CN117455579B - Commodity recommendation intervention method, commodity recommendation intervention device, medium and equipment - Google Patents

Abstract

The application relates to a commodity recommendation intervention method, a commodity recommendation intervention device, a medium and a device, wherein the commodity recommendation intervention method comprises the following steps: responding to a commodity searching request, and respectively recalling commodity items matched with the commodity searching request from a plurality of data sources to obtain recalled commodity sets corresponding to the data sources; acquiring a occupying commodity set, deleting commodity items, which are overlapped with the occupying commodity set, in each recall commodity set, wherein the occupying commodity set comprises at least one commodity item and a to-be-occupied position thereof; according to the intervention display scores obtained by the corresponding commodity items in each recall commodity set, carrying out fine-ranking treatment on all commodity items in all recall commodity sets to obtain a fine-ranking commodity set; and adding the commodity items in the occupied commodity set to the ordering order corresponding to the to-be-occupied order of the commodity items in the fine-ranking commodity set to obtain a recommended commodity set for responding to the commodity search request. The method and the device realize optimization of the commodity search results and are favorable for realizing personalized operation of the commodity search results.

Description

Commodity recommendation intervention method, device, medium and equipment

Technical Field

The present application relates to the field of e-commerce information technology, and in particular to a product recommendation intervention method, device, medium and equipment.

Background technique

In e-commerce platforms, collaborative filtering technology is often used to achieve product recommendations. Collaborative filtering is a common recommendation algorithm that analyzes user behavior and preferences, finds other users or products similar to them, and then uses these similarities to make recommendations.

The implementation process of collaborative filtering recommendation system includes two main steps: user similarity calculation and recommendation result generation. Pure collaborative filtering recommendation system has some technical defects and problems.

First, when user behavior data is sparse or cold-started, it is difficult to find sufficiently similar users or products, resulting in inaccurate recommendation results. For e-commerce platforms based on independent sites, since each independent site cannot share user and product information, when new products are listed or new users log in, there is a lack of corresponding historical data, resulting in limited effectiveness of collaborative filtering recommendation algorithms.

Secondly, collaborative filtering recommendation systems are easily affected by data preferences, resulting in the disadvantages of information cocoons, which may cause recommendation results to be too narrow or too consistent.

In addition, collaborative filtering technology is often implemented based on end-to-end deep learning models, and the results it produces are pushed directly to users. It is impossible to intervene in the results from an operational perspective, resulting in the platform being unable to effectively control the product recommendation results and unable to fully meet user needs.

It can be seen that on the basis of the collaborative filtering technology widely used in existing e-commerce platforms to obtain product recommendation results, further technical optimization is still needed to optimize the effectiveness of product recommendations.

Summary of the invention

The purpose of this application is to provide a commodity recommendation intervention method, device, medium and equipment.

According to one aspect of the present application, a commodity recommendation intervention method is provided, comprising:

In response to a product search request, product items matching the product search request are respectively recalled from multiple data sources to obtain a recalled product set corresponding to each data source;

Obtain a placeholder product set, and delete product items in each recalled product set that overlap with the placeholder product set, wherein the placeholder product set includes at least one product item and its order to be occupied;

According to the intervention display scores obtained by the corresponding product items in each recalled product set, all product items in all recalled product sets are refined and sorted to obtain a refined product set;

The commodity items in the placeholder commodity set are added to the sorting order of the refined commodity set corresponding to the order of the commodity items to be occupied, so as to obtain a recommended commodity set for responding to the commodity search request.

According to another aspect of the present application, a commodity recommendation intervention device is provided, comprising:

A data recall module is configured to respond to a product search request, recall product items matching the product search request from multiple data sources, and obtain a recalled product set corresponding to each data source;

A product deduplication module is configured to obtain a placeholder product set and delete product items in each recalled product set that overlap with the placeholder product set, wherein the placeholder product set includes at least one product item and its order to be occupied;

A product fine-ranking module is configured to perform fine-ranking processing on all product items in all recalled product sets according to the intervention display scores obtained by the product items in each recalled product set, so as to obtain a fine-ranked product set;

The placeholder optimization module is configured to add the commodity items in the placeholder commodity set to the sorting order corresponding to the order to be occupied by the commodity items in the refined commodity set, and obtain a recommended commodity set for responding to the commodity search request.

According to another aspect of the present application, a computer device is provided, including a central processing unit and a memory, wherein the central processing unit is used to call and run a computer program stored in the memory to execute the steps of the commodity recommendation intervention method described in the present application.

According to another aspect of the present application, a non-volatile readable storage medium is provided, which stores a computer program implemented according to the commodity recommendation intervention method in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the method are executed.

Compared with the prior art, this application has many technical advantages, including but not limited to:

First, the present application specifies several product items and their corresponding to-be-occupied ranks by pre-setting a placeholder product set, so as to indicate that in a given plurality of ranks, the product items display the corresponding ranks. After the user submits a product search request, a recalled product set matching the product search request is obtained from multiple data sources. The duplicate product items overlapping with the placeholder product set are first deleted from the recalled product set, so that the product items with the specified to-be-occupied ranks do not participate in the process of fine-sorting each recalled product set. After the fine-sorting is completed to obtain the fine-sorted product set, the product items in the placeholder product set are added to the corresponding sorting ranks in the fine-sorted product set according to the to-be-occupied ranks, thereby achieving effective intervention in the multi-source recall results and facilitating personalized operations based on the multi-source recall results.

Secondly, by effectively intervening in the multi-source recall results, for independent sites, this application can optimize the search results of the collaborative filtering recommendation algorithm with the help of a preset placeholder product set in response to situations such as sparse data and cold start of the independent site. At the same time, it can also overcome the disadvantages of data preference and generalize the product recommendation results, so that the product items in the independent site can get more fair recommendation opportunities by being pre-set in the placeholder product set, thereby promoting the increase in the total transaction volume of the platform.

In addition, the placeholder product set of the present application can be customized by the platform or independent site, enriching the technical means for the platform or independent site to perform personalized operations on multi-source recall results, and can further realize the function of product advertising implantation in the product recommendation results. The overall solution is easy to implement, has low deployment costs, high computing efficiency, and can bring higher comprehensive benefits.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

FIG1 is a network architecture of an exemplary e-commerce platform of the present application;

FIG2 is a flow chart of a commodity recommendation intervention method in an embodiment of the present application;

FIG3 is a schematic diagram of a process for finely sorting each recalled product set in an embodiment of the present application;

FIG4 is a schematic diagram of a process of fusing a placeholder product set with a refined product set in an embodiment of the present application;

FIG5 is a schematic diagram of a process of responding to a product search request according to a recommended product set in an embodiment of the present application;

FIG6 is a schematic diagram of the structure of a commodity recommendation intervention device in an embodiment of the present application;

FIG. 7 is a schematic diagram of the structure of a computer device in an embodiment of the present application.

Detailed ways

In the network architecture shown in FIG1 , an e-commerce platform 82 is deployed on the Internet to provide corresponding services to its users, and the merchant user terminal device 80 and the consumer user terminal device 81 of the e-commerce platform 82 are also connected to the Internet to use the services provided by the e-commerce platform. For example, the e-commerce platform can open advertising delivery services for merchant users of various online stores in the e-commerce platform by configuring an advertising system, and when merchant users submit advertising campaign information, the corresponding target audience for advertising delivery is determined for them, and the product activity information in the advertising campaign information is delivered to the corresponding target audience.

The exemplary e-commerce platform 82 provides supply and demand matching of products and/or services to the general public with the help of Internet infrastructure. In the e-commerce platform 82, products and/or services are provided as commodity information. To simplify the description, the concepts of commodity, product, etc. are used in this application to refer to the products and/or services in the e-commerce platform 82, which may specifically be physical products, digital products, tickets, service subscriptions, other offline services, etc.

In reality, all entities can access the e-commerce platform 82 as users, use various online services provided by the e-commerce platform 82, and achieve the purpose of participating in the business activities achieved by the e-commerce platform 82. These entities can be natural persons, legal persons or social organizations, etc. Corresponding to the two types of entities, merchants and consumers in business activities, the e-commerce platform 82 has two types of users, merchant users and consumer users. All entities in the product distribution chain in business activities, including manufacturers, sellers, retailers, logistics providers, etc., can use online services in the e-commerce platform 82 as merchant users, while consumers in business activities, including real or potential consumers, can use online services in the e-commerce platform 82 as their corresponding consumer users. In actual business activities, the same entity can act as both a merchant user and a consumer user, and this should be understood flexibly.

The infrastructure used to deploy the e-commerce platform 82 mainly includes the backend architecture and frontend equipment. The backend architecture runs various online services through the service cluster, including middleware or frontend services for the platform, services for consumers, services for merchants, etc., to enrich and improve its service functions; the frontend equipment mainly covers the terminal devices used by users to access the e-commerce platform 82 as clients, including but not limited to various mobile terminals, personal computers, point-of-sale devices, etc. For example, a merchant user can enter product information for his online store through the merchant user terminal device 80, or generate his product information using the interface opened by the e-commerce platform; a consumer user can access the webpage of the online store implemented by the e-commerce platform 82 through the consumer user terminal device 81, trigger the shopping process through the shopping button provided on the webpage, and call various online services provided by the e-commerce platform 82 in the shopping process, so as to achieve the purpose of shopping and placing orders.

In some embodiments, the e-commerce platform 82 can be implemented by a processing facility including a processor and a memory, which stores a set of instructions that, when executed, cause the e-commerce platform 82 to perform the e-commerce and supporting functions involved in the present application. The processing facility can be part of a server, client, network infrastructure, mobile computing platform, cloud computing platform, fixed computing platform, or other computing platform, and provides electronic components, merchant equipment, payment gateways, application developers, marketing channels, transportation providers, customer equipment, point-of-sale equipment, etc. of the e-commerce platform 82.

The e-commerce platform 82 can be implemented as an online service such as cloud computing service, software as a service (SaaS), infrastructure as a service (IaaS), platform as a service (PaaS), desktop as a service (DaaS), hosted software as a service, mobile backend as a service (MBaaS), information technology management as a service (ITMaaS), etc. In some embodiments, the various functional components of the e-commerce platform 82 can be implemented to be suitable for operation on various platforms and operating systems. For example, for an online store, its administrator users enjoy the same or similar functions regardless of various embodiments such as iOS, Android, HomonyOS, or web pages.

The e-commerce platform 82 can realize the corresponding independent station for each merchant to run its corresponding online store, and provide the merchant with the corresponding business management engine instance for the merchant to establish, maintain and run one or more online stores in one or more independent stations. The business management engine instance can be used for content management, task automation and data management of one or more online stores, and various specific business processes of the online store can be configured through interfaces or built-in components to support the implementation of business activities. The independent station is the infrastructure of the e-commerce platform 82 with cross-border service functions. Merchants can maintain their online stores more centrally and autonomously based on the independent station. The independent station usually has a domain name and storage space dedicated to the merchant, and different independent stations are relatively independent. The e-commerce platform 82 can provide standardized or personalized technical support for a large number of independent stations, so that merchant users can customize their own business management engine instance and use this business management engine instance to maintain one or more online stores they own.

The online store can implement backend configuration and maintenance by having the merchant user log in to its business management engine instance as an administrator. With the support of various online services provided by the infrastructure of the e-commerce platform 82, the merchant user can configure various functions in its online store as an administrator, view various data, etc. For example, the merchant user can manage various aspects of its online store, such as viewing the online store's recent activities, updating the online store's product catalog, managing orders, recent visit activities, total order activities, etc.; the merchant user can also view more detailed information about the business and visitors to the merchant's online store by obtaining reports or metrics, such as showing a sales summary of the merchant's overall business, specific sales and participation data of active sales marketing channels, etc.

The e-commerce platform 82 may provide communication facilities and associated merchant interfaces for providing electronic communications and marketing, such as utilizing electronic message aggregation facilities to collect and analyze communication interactions between merchants, consumers, merchant devices, customer devices, point-of-sale devices, etc., aggregating and analyzing communications, such as for increasing the potential for providing product sales, etc. For example, a consumer may have a question related to a product, which may generate a dialogue between the consumer and the merchant (or an automated processor-based agent on behalf of the merchant), wherein the communication facility is responsible for interacting and providing the merchant with analysis on how to increase the probability of a sale.

In some embodiments, an application suitable for installation in a terminal device can be provided to serve the access needs of different users, so that various users can access the e-commerce platform 82 by running the application in the terminal device, such as the merchant backend module of the online store in the e-commerce platform 82, etc. In the process of implementing business activities through these functions, the e-commerce platform 82 can implement various functions related to supporting the implementation of business activities as middleware or online services and open corresponding interfaces, and then implant the toolkit corresponding to the interface access function into the application to implement function expansion and task implementation. The business management engine can include a series of basic functions, and expose these functions to online services and/or application calls through APIs. Online services and applications use corresponding functions by remotely calling corresponding APIs.

With the support of various components of the business management engine instance, the e-commerce platform 82 can provide online shopping functions, enabling merchants to establish connections with customers in a flexible and transparent manner. Consumer users can select items online, create product orders, provide the delivery address of the goods in the product order, and complete the payment confirmation of the product order. The merchant can then review and complete or cancel the order. The review component carried by the business management engine instance can implement the compliance use of business processes to ensure that the order is suitable for fulfillment before actual fulfillment. Orders may sometimes be fraudulent and need to be verified (such as ID card checks). There is a payment method that requires merchants to wait to ensure that funds are received, which can play a role in preventing such risks, etc. Order risks may be generated by fraud detection tools submitted by third parties through order risk APIs, etc. Before fulfillment, merchants may need to obtain payment information or wait to receive payment information in order to mark the order as paid, so that the merchant can prepare to deliver the product. Such situations can be reviewed accordingly. The review process can be implemented by the fulfillment component. Merchants can use the fulfillment component to review, adjust work, and trigger related fulfillment services, such as: manual fulfillment services, which are used when merchants pick and pack products in boxes, purchase shipping labels and enter their tracking numbers, or simply mark items as fulfilled; custom fulfillment services, which can define email notifications; API fulfillment services, which can trigger third-party applications to create fulfillment records in third parties; legacy fulfillment services, which can trigger custom API calls from the business management engine to third parties; gift card fulfillment services. You can provide the generation of numbers and activate gift cards. Merchants can use the order printer application to print shipping orders. The fulfillment process can be executed when the items are packed in boxes and ready for shipping, tracking, delivery, consumer receipt verification, etc.

It can be seen that the services provided by the e-commerce platform are based on products. The corresponding commodity data is the basic data of the e-commerce platform. The commodity information is provided through the commodity data. The mining and utilization of commodity data is the basis for realizing various technical services, including using the user transaction data and commodity data in the commodity data of the e-commerce platform to provide basic services for the operation of the advertising system. Therefore, the advertising system of this application can be run in any one or more servers of the cluster of the e-commerce platform, so as to realize various functions by using various commodity data provided by the e-commerce platform.

Please refer to FIG. 2 . In some embodiments, the commodity recommendation intervention method of the present application can be implemented as a computer program product, which runs on a server in an e-commerce platform and includes:

Step S5100: In response to a product search request, product items matching the product search request are respectively recalled from multiple data sources to obtain a recalled product set corresponding to each data source;

E-commerce platforms often display multiple product items to consumer users through pages to achieve the purpose of promotion. In an exemplary application scenario, a consumer user accesses the product recommendation page of an independent website in an e-commerce platform through a terminal device. There are multiple slots preset in the product recommendation page. Each slot displays the product information of a corresponding product item, showing the product title, product price, and product image of the product item, and links the entire slot to the product details page of the product item. When the user touches one of the slots, it jumps to the product details page of the corresponding product item. The product recommendation page can be a page specifically used to implement product recommendations, or it can be the product details page of the product item that the current consumer user is browsing, or it can be the search result page returned after the consumer user enters keywords related to the product, etc.

The product information of the product items in each slot obtained by the consumer user on the product recommendation page is usually submitted by the terminal device in advance during the consumer user's visit to trigger the server to execute the corresponding product search process, and finally return the corresponding product search results. The terminal device then parses the product information in the product search results and displays it in each slot. For example, the product search request can be automatically triggered and submitted when the user enters the product details page of a certain product item or enters a page specifically used to implement product recommendations; or, the product search request can also be triggered and submitted when the user enters the corresponding keywords or product images on the product search page and confirms the search operation.

After the server of the e-commerce platform receives the product search request triggered and submitted by the user on the terminal device, it executes the product recall business logic in response to the product search request. Usually, the server will preset multiple data recall channels, each data recall channel corresponds to a corresponding data source, and different data recall channels can use their own corresponding recommendation algorithms to provide product items that match the product search request as product search results, and obtain recalled product sets consisting of matching product items. In an exemplary data recall channel, the data recall interface of the e-commerce platform can be called, and the collaborative filtering recommendation system can be called with the help of the data recall interface to perform a product search in the product database of the independent station as the first data source to obtain a recall product set consisting of corresponding matching product items; in another exemplary data recall channel, the local cached backup data can be called as the second data source, and the business logic based on rules or collaborative filtering recommendation algorithms implemented locally can be used to search in the second data source to obtain a recall product set consisting of product items that match the product search request. It is not difficult to understand that multiple such data sources can be set up as needed without being limited by the above examples. Therefore, when the server receives a product search request, it calls the data recall interface corresponding to each data source according to the product search request to return the corresponding recalled product set.

The data stored in each data source may have its own characteristics. For example, the exemplary first data source may be the product data of all the product items listed on the independent site, providing various product information of all the product items; the exemplary second data source may be the ranking data formed by the e-commerce platform after statistics of a certain indicator of all the product items on the independent site, such as a list of best-selling product items in the past seven days. The exemplary second data source is stored in the local cache and becomes the cache backup data, which can overcome the deficiencies caused by data sparsity and cold start, and provide a backup product item for product search requests. Of course, the data stored in various data sources and the media in which the data is stored are not limited to the above examples. For example, each data source may be a data source corresponding to a different product database, etc. The algorithms used for data recall for each data source may also be different. Those skilled in the art may implement it flexibly according to the spirit disclosed herein.

When using a collaborative filtering recommendation algorithm or a collaborative filtering recommendation system implemented thereby for data recall, the collaborative filtering recommendation algorithm can utilize the information carried in the product search request, such as user feature information and/or product feature information, to match similar products based on the historical record data corresponding to the user feature information and the historical access data of the product item corresponding to the product feature information, thereby obtaining the corresponding recalled product set for each product item that matches the product search request. Similarly, as long as the data recall interface corresponding to a data source can search for matching product items from the corresponding data source based on the information carried in the product search request to construct a recalled product set, it can serve the subsequent process of this application without being restricted by the specific algorithm used.

Step S5200: Obtain a placeholder product set, and delete the product items in each recalled product set that overlap with the placeholder product set, wherein the placeholder product set includes at least one product item and its position to be occupied;

A placeholder product set can be preset in the server, and the placeholder product set can be edited and determined in advance by the operating user of the independent website, the management user of the e-commerce platform, or any other legally authorized user. In the placeholder product set, one or more product items are stored, and each product item is assigned a pending order. The pending order is used to indicate that the corresponding product item needs to be displayed in a specific slot among multiple slots on the product recommendation page. That is, the pending order points to a position in the multiple slots on the product recommendation page with the same sorting order as the pending order, indicating that the product item will be displayed in the slot corresponding to the pending order.

The product items in the recalled product set obtained by recalling from various data sources can be further sorted so that the product items that best match the product search request can be displayed closer to the front. However, the product items in the placeholder product set will eventually be displayed in the designated slots in the product recommendation page. In this case, the product items in the placeholder product set do not need to participate in the sorting of the product items in each recalled product set. Therefore, for duplicate product items that appear in both the placeholder product set and any recalled product set, this application deletes the duplicate product items in the recalled product set and only retains the duplicate product items in the placeholder product set to achieve deduplication, so that the product items in the placeholder product set no longer appear in each recalled product set.

Deleting the product items in the placeholder product set from each recalled product set can, on the one hand, avoid repeated recommendation of the same product item in the final recommended product set, and on the other hand, maintain the independence of the product items in the placeholder product set. After completing the fine sorting of the recalled product set to obtain the finely ranked product set, each product item in the placeholder product set is accurately inserted into the corresponding slot in the finely ranked product set according to the order of each product item in the placeholder product set to be occupied.

In one embodiment, there are no duplicate product items in the placeholder product set, but the same product item appears in two or more recalled product sets. In this case, each recalled product set can be deduplicated so that the same product item only appears once in all recalled product sets, maintaining the uniqueness of each product item in all recalled product sets. This ensures that the final refined product set will not contain duplicate product items.

Step S5300: According to the intervention display scores obtained for the product items in each recalled product set, all the product items in all the recalled product sets are sorted to obtain a sorted product set;

In order to complete the precise sorting of multi-way recalled product items, that is, the product items in each recalled product set, each product item in each recalled product set can be scored according to the same preset evaluation rules, and the corresponding intervention display score of each product item can be determined. Then, all product items in all recalled product sets can be screened according to the intervention display score to obtain a precisely ranked product set.

Different from the traditional way of designing scores for commodity items, the intervention display score can be designed as an indicator that serves two purposes. On the one hand, it is used to measure the possibility of the corresponding commodity item being replaced, and on the other hand, it is used to reflect the potential sales value of the commodity item. Based on this, the evaluation rules of the intervention display score can be set by technicians in this field as needed. For example, the evaluation rules can be set to conduct a comprehensive quantitative evaluation based on one or more data indicators in the commodity information of the commodity item to obtain the corresponding intervention display score. When designing the evaluation rules, from the perspective of whether the commodity item is popular, the various data indicators related to the characterization of the popularity in the commodity information can be examined and determined, so as to obtain the data of the corresponding commodity item under various data indicators for comprehensive quantification. These data indicators can be flexibly selected as needed, for example, they can be the advertising conversion rate, advertising click-through rate, etc. of the commodity item. Each data indicator can also be an indicator related to the sales potential of the commodity item, for example, the ranking of the commodity item in the list on the e-commerce platform, the profit amount, the discount rate, etc. Each data indicator can not only be numerical data, but also enumerated data or Boolean data. For example, it can be Boolean data that represents whether it is a new product or a main product. The data indicators corresponding to enumerated data or Boolean data can be mapped to a certain score to participate in the comprehensive quantification among various data indicators.

For each product item in each recalled product set, after calculating the intervention display score of each product item under various data indicators according to the scoring rules, all product items in each recalled product set can be merged into the same data set, in which each product item is sorted according to the intervention display score, so as to obtain a refined product set.

Intervention display scores are generally quantified according to value. The higher the score, the more potential sales value it has and the more worthy it is to be replaced. Conversely, the lower the potential sales value, the lower the possibility of being replaced. Based on this, in the data set obtained by merging various recalled product sets, each product item can be sorted in reverse order according to the intervention display score, so that each product item is arranged from high to low to form a refined product set.

In some embodiments, the data set can be further screened to delete the commodity items whose intervention display scores do not reach the preset threshold, and only retain the commodity items whose intervention display scores reach the preset threshold to form the refined commodity set, so as to ensure that the commodity items in the refined commodity set have high potential sales value. In some further embodiments, the number of slots required to be displayed on the commodity recommendation page of the terminal device is usually determined. If the total number of commodity items in the refined commodity set is less than the total number of slots in the commodity recommendation page, at this time, commodity items can be selected from any recalled commodity set to supplement the refined commodity set so that the total number of commodity items therein reaches the total number of slots.

Step S5400: Add the commodity items in the placeholder commodity set to the sorting order corresponding to the to-be-occupied order of the commodity items in the refined commodity set, and obtain a recommended commodity set for responding to the commodity search request.

The refined ranking merchandise items already contain multiple merchandise items, and the placeholder merchandise set also contains one or more merchandise items. In order to merge the merchandise items in the placeholder merchandise set into the refined ranking merchandise set, each merchandise item in the placeholder merchandise set can be inserted or replaced into the sorting order corresponding to the to-be-occupied order of the merchandise item in the refined ranking merchandise set according to the to-be-occupied order specified for the merchandise item in the placeholder merchandise set. Since the sorting order of each merchandise item in the refined ranking merchandise set is determined based on its intervention display score, it has the function of reflecting the potential sales value, and the merchandise items in the placeholder merchandise set specify the corresponding insertion or replacement sorting order through the to-be-occupied order, thus ensuring that the merchandise items in the placeholder merchandise set are implanted into the positions with corresponding potential sales value in the refined ranking merchandise set, and realizing the precise intervention of the display of the merchandise items in the refined ranking merchandise set through the to-be-occupied order specified in the placeholder merchandise set, thereby completing the organic integration of the placeholder merchandise set and the refined ranking merchandise set, and obtaining the recommended merchandise set.

The recommended product set can be used directly to answer the product search request submitted by the consumer user, or it can be truncated and screened according to a predetermined rule, such as the total number of slots on the product recommendation page at the terminal device, and then answer the product search request. When answering the product search request, each product item in the recommended product set can be formatted and packaged according to the predetermined format requirements, and then pushed to the terminal device where the consumer user is located as the product search result, parsed and displayed in each corresponding slot on the product recommendation page, thereby completing the response to the product search request.

According to the above embodiments, the present application has many technical advantages, including but not limited to:

First, the present application specifies several product items and their corresponding to-be-occupied ranks by pre-setting a placeholder product set, so as to indicate that in a given plurality of ranks, the product items display the corresponding ranks. After the user submits a product search request, a recalled product set matching the product search request is obtained from multiple data sources. The duplicate product items overlapping with the placeholder product set are first deleted from the recalled product set, so that the product items with the specified to-be-occupied ranks do not participate in the process of fine-sorting each recalled product set. After the fine-sorting is completed to obtain the fine-sorted product set, the product items in the placeholder product set are added to the corresponding sorting ranks in the fine-sorted product set according to the to-be-occupied ranks, thereby achieving effective intervention in the multi-source recall results and facilitating personalized operations based on the multi-source recall results.

Secondly, by effectively intervening in the multi-source recall results, for independent sites, this application can optimize the search results of the collaborative filtering recommendation algorithm with the help of a preset placeholder product set in response to situations such as sparse data and cold start of the independent site. At the same time, it can also overcome the disadvantages of data preference and generalize the product recommendation results, so that the product items in the independent site can get more fair recommendation opportunities by being pre-set in the placeholder product set, thereby promoting the increase in the total transaction volume of the platform.

In addition, the placeholder product set of the present application can be customized by the platform or independent site, enriching the technical means for the platform or independent site to perform personalized operations on multi-source recall results, and can further realize the function of product advertising implantation in the product recommendation results. The overall solution is easy to implement, has low deployment costs, high computing efficiency, and can bring higher comprehensive benefits.

Considering the purpose of not wanting multiple identical product items to appear in the final refined product set, it is necessary to ensure that there are no duplicate product items between the placeholder product set and the recalled product set, and it is also necessary to ensure that there are no duplicate product items between different recalled product sets. This embodiment adopts a unified algorithm to solve this problem to avoid redundant or repeated deduplication operations. Accordingly, based on any embodiment of the method of the present application, the product items in each recalled product set that overlap with the placeholder product set are deleted, including:

Step S5110: Obtain the priorities corresponding to the placeholder product set and each recalled product set, wherein the placeholder product set has the highest priority, and each recalled product set has a different priority configured according to the data source to which it belongs;

The server specifies the corresponding priority for each data source in advance, and also regards the placeholder product set as a data source and sets the corresponding priority, which can be called when needed. That is, the placeholder product set and each recalled product set have their corresponding priorities, among which the placeholder product set is set with the highest priority, and the other recalled product sets are set with different priorities according to the importance of their data sources, but they are all lower than the priority of the placeholder product set. For example, the priority of each recalled product set can be set to the second highest priority for the first data source corresponding to the product database of the e-commerce platform, and the lowest priority for the second data source corresponding to the local cache backup data. In this example, the priority relationship of each data source after being reflected in each data set is: placeholder product set>recalled product set of the first data source>recalled product set of the second data source.

Step S5120: De-duplicate the product items in the placeholder product set and each of the recalled product sets, retain the repeated product items only in the set with the highest relative priority, and delete the repeated product items in the sets with lower relative priorities.

After determining the corresponding priorities of the placeholder product set and each recalled product set, a unified algorithm is used to detect repeated product items across multiple product sets. Then, the repeated product items are retained only in the product set with the highest priority in each product set, and the corresponding repeated product items that appear in other relatively lower priority product sets are deleted. After deduplication according to this unified algorithm, each product item will only appear in the product set with the highest priority. Among them, since the product items in the placeholder product set themselves correspond to the highest priority, as long as the product items appear in the placeholder product set, they will not be removed during deduplication, ensuring that the product items set by manual intervention can still be integrated with the refined product set in the end.

In the above embodiment, priorities are set based on the data sources of the placeholder product set and each recalled product set, and a unified algorithm is applied to deduplicate product items according to the priorities. This can not only ensure that product items in the placeholder product set are not mistakenly deleted by the deduplication operation, but also effectively avoid duplicate product items in the final refined product set. The operation is efficient and the data is ordered.

Based on any embodiment of the method of the present application, please refer to FIG. 3 , according to the intervention display scores obtained by the product items in each recalled product set, all the product items in all the recalled product sets are refined and sorted to obtain a refined product set, including:

Step S5210: filtering the commodity items in each recalled commodity set according to the preset filtering condition;

The quality of the recalled product sets obtained from various data sources may vary. Therefore, it is necessary to filter the product items in each recalled product set to eliminate the false and retain the fine, so that the remaining product items after filtering can more effectively match the product search request.

When filtering each recalled product set, a preset filtering condition is applied, and the filtering condition can be flexibly designed, for example, filtering according to whether any product information such as price and sales volume reaches a corresponding set value, or filtering according to a whitelist or blacklist of product items, etc. It is not difficult to understand that after filtering, the product items in each recalled product set have a higher matching degree with the search condition in the product search request.

Step S5220: Calculate the intervention display score of each product item in the recalled product set according to multiple data indicators of the product item;

As disclosed in the above embodiments, multiple data indicators can be specified by the same scoring rule to implement scoring of each commodity item, so as to determine the intervention display score corresponding to each commodity item in each recalled commodity set after filtering. Since the intervention display score of each commodity item is comprehensively and quantitatively determined in association with the same data indicator, although each recalled commodity set comes from a different data source, the commodity items therein are unified in evaluation dimension through the intervention display score, and the commodity items in each recalled commodity set are comparable through the corresponding intervention display score.

Step S5230: Sort the commodity items in the intermediate commodity set according to the intervention display scores of the commodity items in the intermediate commodity set formed by each recalled commodity set;

In order to achieve precise sorting of all the items in each recalled product set, all the items in each recalled product set can be merged into the same data set, and the data set is used as the intermediate product set. Considering that the intervention display score has been designed so that the higher the score, the higher the potential sales ability of the product item, therefore, on the basis of the intermediate product set, each product item is sorted in reverse order according to the intervention display score, so that the higher the ranking, the higher the corresponding potential sales ability, which is exactly the same as the role of the slot order in the product recommendation page of the terminal device.

Step S5240: select some product items from the intermediate product set according to the intervention display scores of the product items to construct a refined product set.

Since the intermediate product set is obtained by collecting and sorting multiple recalled product sets, and the total number of slots in the product recommendation page of the terminal device is limited, and if the intervention display score of the product items in the intermediate product set, it also means that the promotion value of the corresponding product items may not be high, therefore, on the basis of the intermediate product set, further screening can be done according to the intervention display score to screen out better quality product items to form a refined product set.

When screening, first use the preset threshold corresponding to the intervention display score to screen the intermediate product set, screen out the product items whose intervention display score is higher than the threshold, and delete the product items whose intervention display score is lower than the threshold. If the number of product items in the intermediate product set exceeds the total number of slots in the product recommendation page of the terminal device, the intermediate data set can also be truncated according to the total number of slots to obtain multiple product items with the same number of slots to form the final refined product set. If the total number of product items in the refined product set does not reach the total number of slots, product items can be obtained from each recalled product set to supplement the refined product set.

It can be seen from the above embodiments that in the process of fine-ranking after filtering the recalled product sets corresponding to multiple data sources, the intervention display score of the product items plays a key role. After screening the product items of each recalled product set according to the intervention display score, a fine-ranked product set is constructed, in which the product items not only highly match the search conditions in the product search request, thereby improving the accuracy of the product search, but also under the influence of the function of the intervention display score, the sorting order of the product items can effectively reflect the potential sales value of the product items, thereby ensuring that the operation of pre-inserting or directly replacing the product items with the corresponding sorting order in the fine-ranked product set according to the to-be-occupied order in the placeholder product set has corresponding value.

Based on any embodiment of the method of the present application, the intervention display score of each product item in the recalled product set is calculated according to multiple data indicators of the product item, including:

Step S5221, calculating the normalized scores of each product item in the recalled product set under multiple preset data indicators;

When comprehensively determining the intervention display score of each item in each recalled product set based on each data indicator, the normalized score corresponding to each data indicator of each item can be calculated separately. For ease of calculation, the normalized score can be controlled in the numerical range of [0,1]. For the case where the data indicator is numerical, the maximum and minimum value normalization method can be applied to determine the normalized score, refer to the following formula:

in, It is a value that falls within the range of 0-1 after normalization; /> The current value of the current data indicator; /> The minimum value of the current type of data indicator set; /> It is the maximum value among the current type of data indicators.

For data indicators of enumeration type or Boolean type, a mapping relationship between the data and the value can be established. Similarly, it can be mapped to a value interval of [0,1]. For this, technicians in this field can flexibly set it.

Step S5222: Using the preset intervention display probability corresponding to each data indicator as the weight of the data indicator, the weighted total score of the normalized scores of each data indicator of each commodity item is calculated to serve as the intervention display score of each commodity item.

When designing the scoring rules for intervention display scoring, the intervention display probability corresponding to each data indicator can also be set in advance to represent the weight of each indicator in determining the intervention display score. Based on this, after the normalized scores of each data indicator of each commodity item are determined, the normalized scores of each data indicator of each commodity item can be weighted and summarized according to the following formula to obtain a weighted total score as the intervention display score of the commodity item:

in, For product items/> The intervention presentation score; /> The data index of this product item/> The normalized score obtained is, It is a data indicator in the scoring rule/> Corresponding to the preset intervention display probability.

In the above embodiment, for each commodity item, the normalized score of each commodity item under the same data indicator is first scored according to the maximum and minimum values between the numerical values of each commodity item under the same data indicator, and the data of each commodity item under each data indicator is normalized to the same numerical range, and then the corresponding preset weight is adapted for each data indicator to achieve the fusion of the normalized scores of multiple data indicators of each commodity item into the same intervention display score. The intervention display score can effectively quantify the value attributes of each commodity item under the same data indicators, so that each commodity item can be compared through its own intervention display score, thereby ensuring that the result of the refined ranking of the commodity items is more accurate and reliable.

Based on any embodiment of the method of the present application, some commodity items are selected from the intermediate commodity set according to the intervention display scores of the commodity items to form a refined commodity set, including:

Step S5241, select a number of commodity items whose intervention display scores exceed a preset threshold from the intermediate commodity set to form a refined commodity set;

As disclosed above, a number of product items with intervention display scores exceeding the preset threshold can be directly screened out from the intermediate product set according to the preset threshold to form a refined product set. Moreover, if necessary, the total number of product items in the refined product set can also be controlled, for example, so that it does not exceed the total number of slots in the product recommendation page of the terminal device.

Step S5242, determine whether the total number of product items in the refined ranking product set reaches the preset total number of slots. If not, extract the top-ranked product items from the recalled product set belonging to the ranked product set to supplement the refined ranking product set. The ranked product set contains multiple product items with a preset ranking relationship.

In order to avoid the total number of commodity items in the refined ranking commodity set being insufficient to fill the preset slots in the commodity recommendation page, the total number of commodity items in the refined ranking commodity set can be judged to determine whether it reaches the preset total number of slots. When the preset total number of slots is reached, no additional processing is required. When the preset total number of slots is not reached, commodity items can be called from the preset recalled commodity set belonging to the ranking commodity set to supplement the refined ranking commodity set. The recalled commodity set belonging to the ranking commodity set can be the cached bottom-up data referred to above, that is, the second data source. The various commodity items contained in the ranking commodity set are generated by ranking and statistics of the commodity items according to certain rules, in which the ranking relationship of each commodity item is determined, and can be called according to the ranking order. The ranking basis between each commodity item can be set on demand according to the popularity of the commodity, the click-through rate of the commodity, the conversion rate of the commodity, etc., or the e-commerce ranking list of the e-commerce platform can be directly obtained to directly constitute the ranking commodity set. When it is necessary to call commodity items from the ranking commodity set to supplement the refined ranking commodity set, it can be called in sequence according to the ranking order.

In the above embodiments, in order to avoid that the product items in the refined product set are insufficient to correspond to each slot in the product recommendation page, the product items in the recalled product set belonging to the ranked product set are called to supplement the refined product set, which plays a role in providing a backup for the refined product set and ensures that the product search request can obtain multiple product items corresponding to the total number of preset slots, avoiding the product search results being too narrow, and overcoming problems such as data sparsity and cold start.

Based on any embodiment of the method of the present application, referring to FIG. 4 , adding a commodity item in the placeholder commodity set to a ranking order corresponding to the place to be occupied by the commodity item in the refined commodity set includes:

Step S5410: determine whether the total number of commodity items in the refined commodity set reaches the preset total number of slots;

Depending on the two different situations of whether the total number of product items in the refined product set reaches the total number of preset slots on the product recommendation page of the terminal device, different strategies can be applied to the fusion of the placeholder product set and the refined product set. Based on this, first judge whether the total number of product items in the refined product set exceeds the total number of slots to obtain the corresponding judgment result.

Step S5420: When the preset total number of slots is reached, according to the correspondence between the to-be-occupied order of the commodity items in the placeholder commodity set and the sorting order in the refined commodity set, the commodity items in the placeholder commodity set are added to the refined commodity set by insertion and/or replacement to obtain a recommended commodity set;

When the judgment result shows that the total number of commodity items in the refined ranking commodity set reaches the total number of slots, it indicates that the number of commodity items in the refined ranking commodity set can fill the slots in the commodity recommendation page. In this case, in one embodiment, for each commodity item in the placeholder commodity set, it is replaced by the commodity item stored in the refined ranking commodity set at the ranking order corresponding to the position to be occupied by the commodity item, ensuring that the number of commodity items in the refined ranking commodity set does not increase, thereby obtaining a recommended commodity set. In another embodiment, each commodity item in the placeholder commodity set can also be inserted into the position in the refined ranking commodity set corresponding to its position to be occupied, starting with the commodity items at the front of the position to be occupied, so that the commodity items in the refined ranking commodity set are pushed back one slot due to being inserted in front, and then the next commodity item to be occupied is inserted accordingly, and so on, and a recommended commodity set can also be obtained. At the same time, the product items of the placeholder product set can also be added to the refined product set by both insertion and replacement. In one embodiment, the products with higher rankings in the placeholder product set are added to the refined product set by insertion, while the products with lower rankings are added to the refined product set by replacement, replacing the products with corresponding rankings after re-ordering, thereby finally forming a recommended product set.

Step S5430: When the preset total number of slots is not reached, the commodity items in the placeholder commodity set are inserted into the corresponding sorting order of the refined commodity set according to the order of the commodity items in the placeholder commodity set to be occupied, so as to obtain a recommended commodity set.

When the judgment result shows that the total number of product items in the refined product set does not reach the total number of preset slots, it means that the number of product items in the refined product set is insufficient to fill the slots in the product recommendation page. In this case, for each product item in the placeholder product set, it is inserted into the sorting order corresponding to the position to be occupied by the product item in the refined product set, so that the sorting order and the original product items behind it are sorted backwards, so that the number of product items in the refined product set is increased, thereby obtaining the corresponding recommended product set.

Of course, in one embodiment, after inserting a single product item of the placeholder product set into the refined product set when the preset total number of slots has not been reached, the process can return to step S5410 to continue iterating the judgment, and so on, until the total number of product items in the refined product set reaches the preset total number of slots. If there are still product items to be processed in the placeholder product set, step S5420 will be executed according to the judgment result, and the remaining product items to be processed will be replaced with the refined product set, and the number of product items in the recommended product set finally obtained will be exactly equal to the preset total number of slots.

It can be seen from the above embodiments that by analyzing the relationship between the total number of product items in the refined product set and the total number of slots on the product recommendation page, different strategies can be flexibly applied to insert or replace product items in the placeholder product set into the refined product set, so as to reasonably control the total number of product items in the refined product set so that it matches the total number of slots in the product recommendation page as much as possible, ensuring that the slots in the product recommendation page can fully display the product information of each product item.

Based on any embodiment of the method of the present application, referring to FIG. 5 , after adding the commodity items in the placeholder commodity set to the sorting order corresponding to the position to be occupied by the commodity items in the refined commodity set, and obtaining the recommended commodity set for responding to the commodity search request, the method includes:

Step S5500: Obtain product information of a product item in the recommended product set, wherein the product information includes a product image, a product title, and a product page link;

After determining the search product set that matches the product search request, the product items in the recommended product set can be data packaged to meet the needs of displaying the product recommendation page in the terminal device. To this end, the product information corresponding to each product item is first obtained from the product database of the independent station based on the product items in the recommended product set.

In this embodiment, the product information obtained for each product item includes but is not limited to a product image, a product title, and a product page link. The product image preferably uses the main product image set by default for the product item in the product database.

Step S5600: encapsulate the product information of the product items in the recommended product set into a product display message according to a preset format and associate it with the sorting order in the recommended product set to construct a product display list corresponding to each product item in the recommended product set;

In order to adapt to the format requirements of displaying the product information of each product item in the recommended product set on the product recommendation page, the product information of each product item in the recommended product set can be encapsulated according to the corresponding format, that is, according to a certain preset format, the product image, product title and product page link of each product item are encapsulated as the corresponding product display message. At the same time, the product display message of each product item is associated with the sorting order of the product item in the recommended product set. In this way, the encapsulation configuration of the product information of each product item in the recommended product set is realized, and the corresponding product display list is obtained. The product display list can be used to display the search result data corresponding to the product search request.

Step S5700: Push the product display list to the terminal device that submits the product search request, and the terminal device parses and displays each product display message in the product display list in a slot corresponding to the sorting order in the display page of the terminal device.

After the server obtains the product display list corresponding to the product search request, it pushes it to the terminal device where the consumer user who submitted the product search request is located. The terminal device then parses the product display list to obtain each product display message therein, and then parses each product display message accordingly to obtain product information such as product images, product titles, and product page links of each product item and the sorting order. Then, the product image and product title of each product item are displayed in the slot of the product recommendation page corresponding to the sorting order of the product item, thereby realizing the parsing and display of the product display list, and displaying the product search results corresponding to the product search request in each slot of the product recommendation page. The product images and product titles displayed in each slot are associated with the product page link of the product item corresponding to the slot. When the user clicks on the product image or product title of a slot, he can jump to the corresponding product page link to load the product details page of the corresponding product item.

According to the above embodiments, it can be seen that according to the recommended product set corresponding to the product search request, the product information of each product item in the recommended product set is packaged in a preset format and associated with the sorting order of each product item, and synchronized to the terminal device for parsing and display. This allows the product items implanted by the present application according to the placeholder product set to be displayed in the corresponding slots of the product recommendation page of the terminal device, effectively realizing manual intervention in the search results recalled from multiple data sources and achieving search business expansion.

Please refer to FIG. 6 . Another embodiment of the present application further provides a product recommendation intervention device, which includes a data recall module 5100, a product deduplication module 5200, a product fine ranking module 5300, and a placeholder optimization module 5400. The data recall module 5100 is configured to respond to a product search request, respectively recall product items that match the product search request from multiple data sources, and obtain a recalled product set corresponding to each data source; the product deduplication module 5200 is configured to obtain a placeholder product set, delete product items in each recalled product set that overlap with the placeholder product set, and the placeholder product set includes at least one product item and its to-be-occupied order; the product fine ranking module 5300 is configured to perform fine ranking processing on all product items in all recalled product sets according to the intervention display scores corresponding to the product items in each recalled product set, and obtain a fine ranking product set; the placeholder optimization module 5400 is configured to add the product items in the placeholder product set to the sorting order corresponding to the to-be-occupied order of the product item in the fine ranking product set, and obtain a recommended product set for answering the product search request.

Based on any embodiment of the device of the present application, the product deduplication module 5200 includes: a priority determination unit, configured to obtain the priorities corresponding to the placeholder product set and each recalled product set, wherein the placeholder product set has the highest priority, and each recalled product set is configured with different priorities according to the data source to which it belongs; a deduplication processing unit, configured to deduplicate product items in the placeholder product set and each recalled product set, retaining repeated product items only in the set with the highest relative priority, and deleting repeated product items in the set with a relatively lower priority.

Based on any embodiment of the device of the present application, the product precise ranking module 5300 includes: a product filtering unit configured to filter product items in each recalled product set according to a preset filtering condition;

The scoring determination unit is configured to calculate the intervention display score of the commodity items in each recalled commodity set according to multiple data indicators of the commodity items; the sorting processing unit is configured to sort the commodity items in the intermediate commodity set according to the intervention display score of the commodity items in the intermediate commodity set constituted by each recalled commodity set; the commodity screening unit is configured to screen out some commodity items from the intermediate commodity set according to the intervention display score of the commodity items to construct a refined commodity set.

Based on any embodiment of the device of the present application, the scoring determination unit includes: an indicator quantification subunit, which is configured to calculate the normalized scores of each product item in the recalled product set under multiple preset data indicators; a weighted scoring subunit, which is configured to use the preset intervention display probability corresponding to each data indicator as the weight of the data indicator, and obtain the weighted total score of the normalized scores of each data indicator of each product item as the intervention display score of each product item.

Based on any embodiment of the device of the present application, the product screening unit includes: a score selection sub-unit, which is configured to screen out a number of product items whose intervention display scores exceed a preset threshold from the intermediate product set to form a refined ranking product set; a product completion sub-unit, which is configured to determine whether the total number of product items in the refined ranking product set reaches a preset total number of slots. If not, top-ranked product items are extracted from a recalled product set belonging to the ranked product set to supplement the refined ranking product set, and the ranked product set contains multiple product items with a pre-set ranking relationship.

Based on any embodiment of the device of the present application, the placeholder optimization module 5400 includes: a matching judgment unit, configured to judge whether the total number of commodity items in the refined ranking commodity set reaches a preset total number of slots; a replacement processing unit, configured to add the commodity items in the placeholder commodity set to the refined ranking commodity set by insertion and/or replacement when the preset total number of slots is reached, according to the correspondence between the order of the commodity items in the placeholder commodity set and the sorting order in the refined ranking commodity set, so as to obtain a recommended commodity set; an insertion processing unit, configured to insert the commodity items in the placeholder commodity set into the corresponding sorting order of the refined ranking commodity set when the preset total number of slots is not reached, according to the order of the commodity items in the placeholder commodity set to obtain a recommended commodity set.

On the basis of any embodiment of the device of the present application, after the operation of the placeholder optimization module 5400, the product recommendation intervention device of the present application includes: an information acquisition module, configured to obtain product information of product items in the recommended product set, the product information including product pictures, product titles, and product page links; a message encapsulation module, configured to encapsulate the product information of the product items in the recommended product set into product display messages according to a preset format and associate them with the sorting order in the recommended product set, thereby constructing a product display list corresponding to each product item in the recommended product set; a product display module, configured to push the product display list to a terminal device that submits the product search request, and the terminal device parses and displays each product display message in the product display list in a slot corresponding to the sorting order in a display page of the terminal device.

On the basis of any embodiment of the present application, please refer to FIG. 7. Another embodiment of the present application further provides a computer device, as shown in FIG. 7, a schematic diagram of the internal structure of the computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and a computer program encapsulating computer-readable instructions. The database may store a control information sequence. When the computer-readable instruction is executed by the processor, the processor can implement a commodity recommendation intervention method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instruction is executed by the processor, the processor can execute the commodity recommendation intervention method of the present application. The network interface of the computer device is used to connect and communicate with the terminal. It can be understood by those skilled in the art that the structure shown in FIG. 7 is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

In this embodiment, the processor is used to execute the specific functions of each module and its submodule in FIG6 , and the memory stores the program code and various data required to execute the above modules or submodules. The network interface is used to transmit data between user terminals or servers. The memory in this embodiment stores the program code and data required to execute all modules/submodules in the commodity recommendation intervention device of the present application, and the server can call the program code and data of the server to execute the functions of all submodules.

The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the product recommendation intervention method described in any embodiment of the present application.

The present application also provides a computer program product, including a computer program/instruction, which, when executed by one or more processors, implements the steps of the product recommendation intervention method described in any embodiment of the present application.

A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments of the present application can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

The above is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

To summarize, the present application utilizes the to-be-occupied order specified for product items in the placeholder product set to implant corresponding product items into the corresponding sorting order of the refined product set obtained after multi-source recall, thereby implementing intervention in the multi-source recall results, overcoming the defects caused by unfavorable factors such as data sparsity, cold start, and data preference, optimizing product search results, and helping to realize personalized operation of product search results.

Claims (8)

1. A method of merchandise recommendation intervention, comprising:

responding to a commodity searching request, and respectively recalling commodity items matched with the commodity searching request from a plurality of data sources to obtain recalled commodity sets corresponding to the data sources;

acquiring a occupying commodity set, deleting commodity items, which are overlapped with the occupying commodity set, in each recall commodity set, wherein the occupying commodity set comprises at least one commodity item and a to-be-occupied position thereof;

According to the intervention display scores obtained by the corresponding commodity items in each recall commodity set, carrying out fine-ranking treatment on all commodity items in all recall commodity sets to obtain a fine-ranking commodity set;

adding the commodity items in the occupied commodity set to the ordering order corresponding to the to-be-occupied order of the commodity items in the fine-ranking commodity set to obtain a recommended commodity set for responding to the commodity search request;

the deleting commodity items in each recall commodity set, which are overlapped with the occupied commodity set, comprises the following steps:

acquiring priorities corresponding to the occupied commodity set and each recall commodity set, wherein the occupied commodity set has the highest priority, and each recall commodity set is configured with different priorities according to the data source to which the occupied commodity set belongs;

carrying out de-duplication treatment on the commodity items in the occupying commodity set and the recall commodity sets, only keeping the repeatedly-appearing commodity items in the set with the highest relative priority, and deleting the repeated commodity items in the set with the lower relative priority;

the adding the commodity item in the occupying commodity set to the ordering order corresponding to the to-be-occupied order of the commodity item in the precisely arranged commodity set comprises the following steps:

Judging whether the total number of commodity items in the fine-arranged commodity set reaches the preset pit total number or not;

when the total number of the preset pits is reached, adding the commodity items in the occupied commodity set into the fine-ranking commodity set in an inserting or replacing mode according to the corresponding relation between the to-be-occupied positions of the commodity items in the occupied commodity set and the ordering positions in the fine-ranking commodity set so as to obtain a recommended commodity set;

and when the total number of the preset pits is not reached, inserting the commodity items in the occupied commodity set into the sequencing cis corresponding to the fine-ranking commodity set according to the to-be-occupied cis of the commodity items in the occupied commodity set so as to obtain the recommended commodity set.

2. The commodity recommendation intervention method according to claim 1, wherein the performing fine-ranking processing on all commodity items in all recalled commodity sets according to the intervention display scores obtained by the commodity items in each recalled commodity set, to obtain a fine-ranking commodity set, comprises:

filtering commodity items in each recall commodity set according to preset filtering conditions;

calculating intervention display scores of all recalled commodity items in the commodity set according to a plurality of data indexes of the commodity items;

sorting the commodity items in the intermediate commodity set according to the intervention display scores of the commodity items in the intermediate commodity set formed by the recall commodity sets;

And screening partial commodity items from the intermediate commodity sets according to the intervention display scores of the commodity items to construct a fine-ranking commodity set.

3. The merchandise recommendation intervention method of claim 2, wherein calculating an intervention presentation score for each recalled merchandise item in the merchandise set based on the plurality of data metrics for the merchandise item comprises:

calculating normalized scores of commodity items in all recall commodity sets under a plurality of preset data indexes;

and taking the preset intervention display probability corresponding to each data index as the weight of the data index, and solving the weighted total score of the normalized scores of each data index of each commodity item to serve as the intervention display score of each commodity item.

4. The commodity recommendation intervention method of claim 2, wherein selecting a portion of the commodity items from the intermediate commodity collection according to the intervention display score of the commodity items is configured as a fine-ranked commodity collection, comprising:

screening a plurality of commodity items with intervention display scores exceeding a preset threshold value from the middle commodity set to form a fine-ranking commodity set;

judging whether the total number of commodity items in the fine-ranking commodity set reaches the preset pit total number, and when the total number of commodity items does not reach the preset pit total number, extracting commodity items with the front ranking from the recall commodity set belonging to the ranking commodity set to complement the fine-ranking commodity set, wherein the ranking commodity set comprises a plurality of commodity items with preset ranking relations.

5. The commodity recommendation intervention method according to any one of claims 1 to 4, wherein adding the commodity item in the occupied commodity set to a ranking order corresponding to a to-be-occupied order of the commodity item in the fine-ranked commodity set, to obtain a recommended commodity set for responding to the commodity search request, comprises:

acquiring commodity information of commodity items in the recommended commodity set, wherein the commodity information comprises commodity pictures, commodity titles and commodity page links;

packaging commodity information of the commodity items in the recommended commodity set into commodity display information according to a preset format, and associating the commodity display information with ordering positions in the recommended commodity set to construct a commodity display list corresponding to each commodity item in the recommended commodity set;

pushing the commodity display list to terminal equipment submitting the commodity search request, and analyzing and displaying each commodity display message in the commodity display list to pits corresponding to the ordering order in a display page of the terminal equipment by the terminal equipment.

6. A merchandise recommendation intervention device, comprising:

the data recall module is used for responding to the commodity search request and recalling commodity items matched with the commodity search request from a plurality of data sources respectively to obtain recalled commodity sets corresponding to the data sources;

The commodity duplicate removal module is used for acquiring a occupying commodity set, deleting commodity items in each recall commodity set, which are overlapped with the occupying commodity set, wherein the occupying commodity set comprises at least one commodity item and a to-be-occupied position thereof;

the commodity precision arranging module is used for carrying out precision arranging treatment on all commodity items in all recall commodity sets according to the intervention display scores correspondingly obtained by the commodity items in all recall commodity sets to obtain a precision arranging commodity set;

the occupation optimization module is used for adding the commodity items in the occupation commodity set to the ordering order corresponding to the to-be-occupied order of the commodity items in the fine-arranged commodity set to obtain a recommended commodity set for responding to the commodity search request;

the commodity duplicate removal module comprises:

the priority determining unit is used for obtaining the priority corresponding to the occupied commodity set and each recall commodity set, wherein the occupied commodity set has the highest priority, and each recall commodity set is configured with different priorities according to the data source to which the occupied commodity set belongs;

the duplicate removal processing unit is used for carrying out duplicate removal processing on the commodity items in the occupied commodity set and the recall commodity sets, keeping the repeatedly-appearing commodity items only in the set with the highest relative priority, and deleting the repeated commodity items in the set with the lower relative priority;

The occupation optimization module comprises:

the matching judging unit is used for judging whether the total number of commodity items in the fine-arranged commodity set reaches the preset pit total number or not;

the replacement processing unit is used for adding the commodity items in the occupied commodity set into the fine-ranking commodity set in an inserting and/or replacing mode according to the corresponding relation between the to-be-occupied positions of the commodity items in the occupied commodity set and the ordering positions in the fine-ranking commodity set when the total number of the preset pits is reached, so that a recommended commodity set is obtained;

and the inserting processing unit is used for inserting the commodity items in the occupied commodity set into the sequencing order corresponding to the fine-ranking commodity set according to the to-be-occupied order of the commodity items in the occupied commodity set when the total number of the preset pits is not reached, so as to obtain the recommended commodity set.

7. A computer device comprising a central processor and a memory, characterized in that the central processor is arranged to invoke a computer program stored in the memory for performing the steps of the method according to any of claims 1 to 5.

8. A non-transitory readable storage medium, characterized in that it stores in form of computer readable instructions a computer program implemented according to the method of any one of claims 1 to 5, which when invoked by a computer, performs the steps comprised by the corresponding method.