CN113515701A - Information recommendation method and device - Google Patents
Information recommendation method and device Download PDFInfo
- Publication number
- CN113515701A CN113515701A CN202110762301.8A CN202110762301A CN113515701A CN 113515701 A CN113515701 A CN 113515701A CN 202110762301 A CN202110762301 A CN 202110762301A CN 113515701 A CN113515701 A CN 113515701A
- Authority
- CN
- China
- Prior art keywords
- information
- pieces
- value
- click rate
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 239000013598 vector Substances 0.000 claims description 48
- 239000011159 matrix material Substances 0.000 claims description 47
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 6
- 230000008707 rearrangement Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 2
- 230000010365 information processing Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 238000010606 normalization Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9538—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses an information recommendation method and device, and belongs to the technical field of information processing. The information recommendation method comprises the following steps: acquiring M pieces of information associated with a user; wherein, the information arrangement sequence in the M information is a first sequence; determining a similarity value between any two pieces of information in the M pieces of information and a click rate predicted value corresponding to each piece of information; rearranging the M information according to the similarity value and the click rate predicted value to obtain M information arranged according to a second sequence; recommending M pieces of information to the user according to a second sequence; wherein M is a positive integer and M is more than or equal to 2.
Description
Technical Field
The application belongs to the technical field of information processing, and particularly relates to an information recommendation method and device.
Background
With the development of the internet, people have entered an information explosion age, wherein personalized information recommendation has gradually become an important field of the internet. In the process of recommending information to a user, how to select the content which is interested by the user from massive information becomes an urgent problem to be solved.
In the prior art, in order to improve the diversity of information recommendation, an information recommendation method is mainly used for screening and arranging recommendation information based on a certain specific rule, for example, two pieces of information of the same type cannot be continuously recommended by forced limitation, so that all pieces of information recommended in the same information recommendation are prevented from being similar. However, the information recommendation method based on the rule bias cannot guarantee the diversity of information and give consideration to the user interest, so that the success rate of information recommendation is low.
Disclosure of Invention
The embodiment of the application aims to provide an information recommendation method and device, and the problem that the success rate of information recommendation is low due to the fact that the existing information recommendation mode cannot guarantee information diversity and give consideration to user interests is solved.
In a first aspect, an embodiment of the present application provides an information recommendation method, where the method includes:
acquiring M pieces of information associated with a user; wherein, the information arrangement sequence in the M pieces of information is a first sequence;
determining a similarity value between any two pieces of information in the M pieces of information and a click rate predicted value corresponding to each piece of information;
rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information arranged according to a second sequence;
recommending the M pieces of information to the user according to the second sequence;
wherein M is a positive integer and M is more than or equal to 2.
In a second aspect, an embodiment of the present application provides an information recommendation apparatus, including:
an acquisition module for acquiring M pieces of information associated with a user; wherein, the information arrangement sequence in the M pieces of information is a first sequence;
the determining module is used for determining a similarity value between any two pieces of information in the M pieces of information and a click rate predicted value corresponding to each piece of information;
the rearrangement module is used for rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information which are arranged according to a second sequence;
a recommending module, configured to recommend the M pieces of information to the user according to the second order;
wherein M is a positive integer and M is more than or equal to 2.
In a third aspect, an embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program or instructions stored on the memory and executable on the processor, and when executed by the processor, the program or instructions implement the steps of the method according to the first aspect.
In a fourth aspect, embodiments of the present application provide a readable storage medium, on which a program or instructions are stored, which when executed by a processor implement the steps of the method according to the first aspect.
In a fifth aspect, an embodiment of the present application provides a chip, where the chip includes a processor and a communication interface, where the communication interface is coupled to the processor, and the processor is configured to execute a program or instructions to implement the method according to the first aspect.
In the embodiment of the application, the similarity value between any two pieces of information in the M pieces of information to be selected and the click rate predicted value corresponding to each piece of information are determined, the similarity value and the click rate predicted value of a user are comprehensively considered, and the M pieces of information are rearranged.
Drawings
FIG. 1 is a schematic diagram of an information recommendation architecture shown in accordance with an exemplary embodiment;
FIG. 2 is one of the flow diagrams illustrating a method of information recommendation, according to an example embodiment;
FIG. 3 is a second flowchart illustrating a method of information recommendation, according to an example embodiment;
FIG. 4 is a block diagram illustrating the structure of an information recommendation device according to an exemplary embodiment;
FIG. 5 is a block diagram illustrating the structure of an electronic device in accordance with an exemplary embodiment;
fig. 6 is a schematic diagram of a hardware structure of an electronic device implementing an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described clearly below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present disclosure.
The terms first, second and the like in the description and in the claims of the present application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that embodiments of the application may be practiced in sequences other than those illustrated or described herein, and that the terms "first," "second," and the like are generally used herein in a generic sense and do not limit the number of terms, e.g., the first term can be one or more than one. In addition, "and/or" in the specification and claims means at least one of connected objects, a character "/" generally means that a preceding and succeeding related objects are in an "or" relationship.
The information recommendation method, apparatus, electronic device and storage medium provided by the embodiments of the present application are described in detail below with reference to the accompanying drawings through specific embodiments and application scenarios thereof.
The information recommendation method provided by the application can be applied to an application scene for recommending information to a user when the user browses and refreshes information streams, and is specifically described in detail with reference to fig. 1.
FIG. 1 is a schematic diagram illustrating an information recommendation architecture in accordance with an exemplary embodiment.
As shown in fig. 1, the architecture diagram may include at least one first electronic device 10 where a client is located and a second electronic device 11 where a server is located. The second electronic device 11 can establish a connection with at least one first electronic device 10 through a wired or wireless network and perform information interaction. The first electronic device 10 may be a device with a communication function, such as a mobile phone, a tablet computer, and an all-in-one machine, and may also be a device simulated by a virtual machine or a simulator. The second electronic device 11 may be a device having storage and computing functions, such as a cloud server or a server cluster.
Based on the above architecture, a user can browse an information stream recommended by a server in the second electronic device 11 on a client in the first electronic device 10, where the information stream is a content stream that can be scrolled, and an information (feed) in the information stream may be an article or a video, which is not limited herein.
Generally, the server can issue a set amount of information for the user to browse and select once, and if the user browses the currently issued information, the user can trigger the server to issue a new set amount of information by continuously sliding the screen in the information flow display interface or clicking a set area in the interface.
Therefore, in order to guarantee the diversity of information and simultaneously give consideration to the interest of the user and improve the click rate of the user, information recommendation can be specifically carried out in the following mode. That is, a user first triggers and generates an information obtaining request through a client in the first electronic device 10, and sends the request to the second electronic device 11, and a server in the second electronic device 11 may first obtain a plurality of pieces of information associated with the user, where an arrangement order of the pieces of information corresponding to the plurality of pieces of information is the first order. Then, determining a similarity value between any two pieces of information in the plurality of pieces of information and a click rate predicted value corresponding to each piece of information, and rearranging the plurality of pieces of information according to the similarity value between any two pieces of information and the click rate predicted value corresponding to each piece of information to obtain a plurality of pieces of information arranged according to a second sequence. And finally, recommending the plurality of information to the user according to the second sequence.
Therefore, the similarity value can reflect the correlation between the two pieces of recommended information, and the click rate predicted value can reflect the user interest tendency, so that the correlation between the information and the user interest degree can be comprehensively considered, the rearranged information can ensure the diversity of the information on the user watching sequence and can cater to the interest of the user, the information diversity is ensured, the user interest is considered, and the information recommendation success rate is improved.
According to the above architecture and application scenarios, the following describes in detail an information recommendation method provided in an embodiment of the present application with reference to fig. 5 to 6, where the information recommendation method can be executed by a server in the second electronic device 12 shown in fig. 1.
Fig. 2 is a flow chart illustrating an information recommendation method according to an example embodiment.
As shown in fig. 2, the information recommendation method may include the steps of:
Wherein M is a positive integer and M is more than or equal to 2.
Here, the information may be information of a text, a picture, a video, or any combination thereof, such as an article including a text and a picture, a video including a text, or the like. The M pieces of information associated with the user may be the same type of information as the information interested by the user, or may be other types of information associated with the information interested by the user, which is not limited herein. Illustratively, the information related to the user can be acquired from an alternative information base arranged in the server side, or the information related to the user can be directly acquired from the historical recommendation information of the user.
Here, the M pieces of information have a certain arrangement order, and specifically, the M pieces of information may be arranged in a first order.
In an optional implementation manner, the M pieces of information may include S pieces of first information and T pieces of second information;
the step 210 may specifically include:
s pieces of first information associated with the user are obtained from the alternative information base according to a preset information recommendation strategy; acquiring T pieces of second information which are not clicked by the user from historical recommendation information corresponding to the user;
wherein S, T is a positive integer.
Here, a candidate information list of size M may be obtained, which is derived from two parts. One part of the information is information recommended from the alternative information base according to the user characteristics and the preset information recommendation strategy, namely first information; the other part is information filtered from the historical recommendation information corresponding to the user, that is, second information, for example, information that is not clicked by the user in the plurality of information recommended to the user last time.
It should be noted that the first information may be information in an information list returned after the fine ranking stage, and the fine ranking stage may be a stage in which a plurality of recalled information are arranged according to a preset rule in the personalized information recommendation process. That is, the information list returned after the fine ranking stage is an ordered information list.
For example, the obtained T second information may be randomly interleaved in the ordered information list including S first information, resulting in a candidate information list arranged based on the first order and having a size of M.
In this way, by acquiring the first information from the candidate information base and acquiring the second information which is not clicked by the user from the historical recommendation information as candidate M pieces of information, since information with similar contents may exist in the first information, the diversity of information recommendation types can be maintained by adding the second information as a type disturbance factor.
Here, the similarity value between two pieces of information can be determined by extracting the feature of each piece of information through comparison between the features. For example, according to the label, classification information, etc. of the information, the information having the same label or the same classification may be determined as similar information, and then a corresponding similarity value may be determined, or a corresponding similarity value may be determined according to the number ratio of the same labels. Of course, the similarity value between two pieces of information can also be determined according to the distance between the feature vectors by acquiring the feature vector corresponding to each piece of information.
In addition, a pCTR (Click-Through Rate) value can be used to reflect the user's interest level. For example, the probability that each of the candidate M pieces of information is clicked by the user, that is, the pCTR value, may be predicted according to the click condition of the user on the history recommendation information.
And step 230, rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information arranged according to a second sequence.
Illustratively, the similarity value between the information and the predicted value of the click rate of the user may be considered in the process of rearranging the M pieces of information, so as to obtain a list after balancing diversity and user interest, where the M pieces of information are arranged according to a second order, that is, the candidate M pieces of information are rearranged from the first order to the second order.
Finally, the M pieces of information arranged according to the second sequence can minimize the similarity value between two adjacent pieces of information while considering the interest of the user, thereby ensuring the diversity of the information.
In an alternative embodiment, the step 230 may specifically include:
constructing a DPP algorithm matrix of the determinant point process according to the similarity value and the click rate predicted value;
and calculating a DPP algorithm matrix based on a greedy algorithm to obtain M pieces of information arranged according to a second sequence.
Here, a DPP (deterministic Point Process) algorithm matrix is a probability model with high performance, and the DPP converts a complex probability calculation into a simple determinant calculation, and calculates the probability of each subset through the determinant of the kernel matrix. The DPP not only can reduce the amount of calculation, but also can improve the operation efficiency. Specifically, the DPP algorithm matrix may be a symmetric matrix.
Specifically, the DPP algorithm matrix may be constructed by setting a value of each element in the DPP algorithm matrix according to a preset construction rule and according to the similarity value and the click rate predicted value. After the matrix is obtained, a certain greedy algorithm can be used for solving the DPP algorithm matrix, an ordered list with the size of M can be output through the greedy algorithm, and the arrangement sequence of M pieces of information in the list is a second sequence.
Therefore, by utilizing the DPP greedy solving algorithm and combining the similarity value and the click rate predicted value, the rearranged result obtained after solving can take the correlation between information and the user interest into consideration, so that the optimal diversity recommendation information is automatically selected, and the user experience is improved.
In an optional embodiment, the step of constructing the DPP algorithm matrix in the determinant point process according to the similarity value and the click rate predicted value may specifically include:
setting the value of the ith row and ith column element in the DPP algorithm matrix as the square of the click rate predicted value corresponding to the ith information in the M information;
setting the value of the ith row and jth column element in the DPP algorithm matrix as the product of the first click rate predicted value, the second click rate predicted value and the target similarity value; the first click rate predicted value is a click rate predicted value corresponding to the ith information in the M information, the second click rate predicted value is a click rate predicted value corresponding to the jth information in the M information, and the target similarity value is a similarity value between the ith information and the jth information;
constructing and obtaining a DPP algorithm matrix based on the value of the ith row and ith column element and the value of the ith row and jth column element;
wherein i and j are positive integers, and i ≠ j.
Illustratively, since the size of the candidate information list is M, the size of the constructed DPP algorithm matrix L is M × M. Assume that the pCTR value of the ith information among the M information is riThen the diagonal elements of matrix L (i.e. the ith row and ith column elements)
Assume that the similarity value between the ith information and the jth information in the M information is sijAnd the pCTR value of the ith information is riThe pCTR value of the jth information is rjThen the non-diagonal elements of the matrix L (i.e., the ith row and jth column elements) Lij＝rirjsij. Further based on the values of the elements in each row and each columnAnd obtaining a DPP algorithm matrix.
Therefore, the diagonal elements of the DPP algorithm matrix are constructed according to the click rate predicted value, the non-diagonal elements of the DPP algorithm matrix are constructed according to the click rate predicted value and the similarity value, the finally constructed DPP algorithm matrix has the click rate predicted value capable of reflecting the user interest degree and the similarity value capable of reflecting the relevance between information, and therefore the matrix is convenient to solve and the user interest degree and the information diversity can be considered when the information is rearranged.
And step 240, recommending M pieces of information to the user according to a second sequence.
For example, the server may sequentially send the M pieces of information to the client used by the user according to a second order, so that the M pieces of information are sequentially displayed on the client for the user to select to browse. Of course, the M pieces of information may also be sent to the client used by the user at one time, so that the M pieces of information are displayed on the client in the second order for the user to select browsing, which is not limited herein.
Therefore, the similarity value between any two pieces of information in the M pieces of information to be selected and the click rate predicted value corresponding to each piece of information are determined, the similarity value and the click rate predicted value of the user are comprehensively considered, the M pieces of information are rearranged, and the click rate predicted value can reflect the interest tendency of the user, so that the rearranged information can ensure the diversity of the information on the watching sequence of the user and cater to the interest of the user, the diversity of the information is ensured, the interest of the user is considered, and the success rate of information recommendation is improved.
On the basis of the foregoing embodiment, in an optional implementation manner, after the foregoing step 230, the information recommendation method provided in the embodiment of the present application may further include:
removing T second information from the M information to obtain P information;
acquiring the first N information from the P information;
wherein P, N is a positive integer;
based on this, step 240 may specifically include:
recommending N pieces of information to the user according to a second sequence.
Here, since the T pieces of second information are already recommended information, the T pieces of second information may be removed when information is recommended this time, and then the first N pieces of information of the filtered ordered list are taken, and further the N pieces of information are information finally recommended to the user.
In this way, by removing the T pieces of second information that have been recommended to the user from the M pieces of information, the repetition rate of information recommendation can be reduced while ensuring the diversity of information.
In addition, in a possible embodiment, as shown in fig. 3, the determining the similarity value between any two pieces of information in the M pieces of information in step 220 may specifically include: step 2201-2204, the specific steps are as follows:
at step 2201, for each of the M pieces of information, an information feature corresponding to each piece of information is extracted from a plurality of dimensions.
Here, the multiple dimensions may specifically include dimensions such as pictures, texts, audio, and the like, and may specifically be determined according to the content of the information.
For example, for each candidate information newly added to the candidate information library, multi-modal information corresponding to the candidate information may be first extracted as features from multiple dimensions. For example, if the information is an article and the article includes a picture and a text, the picture information feature corresponding to the article may be extracted from the picture dimension, and the text information feature corresponding to the article may be extracted from the text dimension, so that the information feature corresponding to the information includes the picture information feature and the text information feature. For another example, if the information is a video and the video includes text and audio, the information features corresponding to the information may include a picture information feature, a text information feature, and an audio information feature, which may be extracted from a picture dimension of a key frame corresponding to the video, a text information feature corresponding to the video, and an audio information feature corresponding to the video.
Here, the information expression model may be first trained before being used.
For example, some information samples can be obtained first, labels and classification information of the information samples are used as labels during training, multi-modal information such as pictures, characters, audio and the like of the information samples is used as input features, and the information expression model is trained through supervised learning until the information expression model converges. And finally outputting the feature vector corresponding to the information by the information expression model.
After the trained information expression model is obtained, for each piece of information in the M pieces of information, multi-modal information of the piece of information can be extracted as information features and input into the trained information expression model, and then a feature vector corresponding to the information can be obtained.
Illustratively, after obtaining the feature vector corresponding to each piece of the M pieces of information, two pieces of information may be selected, and the feature vectors corresponding to the two pieces of information are obtained.
In the embodiment of the application, the similarity value between two pieces of information can be determined by calculating the distance between the feature vectors corresponding to the two pieces of information respectively.
In an optional implementation manner, the step 2204 may specifically include:
normalizing the feature vectors respectively corresponding to any two pieces of information to obtain normalized feature vectors;
and performing inner product calculation on normalized feature vectors corresponding to any two pieces of information to obtain a similarity value between any two pieces of information.
Here, since the feature vectors corresponding to the two pieces of information may have a large difference, but the contents are similar, the two feature vectors may be normalized to unify dimensions, so as to improve the accuracy of the subsequent similarity value calculation result.
For example, after obtaining the normalized feature vector, an inner product may be performed between the normalized feature vectors corresponding to the two arbitrary pieces of information, and a result obtained by the inner product is a similarity value between the two pieces of information.
Therefore, the similarity value is obtained by carrying out normalization processing on the feature vectors and calculating according to the inner product between the normalized feature vectors, the calculation result of the similarity can be more accurate, and the accuracy of the follow-up information recommendation result is improved.
Therefore, the information features corresponding to each piece of information are extracted from multiple dimensions, the information expression model is utilized to output the information features as a feature vector based on the information features, the inner product between the feature vectors corresponding to any two pieces of information is calculated to determine the similarity value between any two pieces of information, and compared with a similarity determination mode which is used for judging whether two pieces of information are similar according to the same label or the same classification, the similarity calculation mode is more accurate and more comprehensive.
Based on this, in an optional implementation manner, before step 210, the information recommendation method provided in the embodiment of the present application may further include:
receiving an information acquisition request sent by a user; the information acquisition request comprises user characteristics and environment characteristics;
determining the click rate prediction value corresponding to each of the M pieces of information in the step 220 may specifically include:
and inputting the information characteristics corresponding to each information in the M information and the user characteristics and the environment characteristics to a click rate prediction model, and outputting to obtain a click rate prediction value corresponding to each information.
Here, the click-through rate prediction model may first be trained before being used.
For example, some historical recommendation information may be obtained as an information sample, and the exposure and click information of the information sample is used to set the positive and negative labels during training, for example, the exposed and clicked information sample is set as a positive label, and the exposed and un-clicked information sample is set as a negative label. Then, the click rate prediction model is trained through supervised learning by taking the user characteristics, the information characteristics, the environment characteristics and the like as input characteristics until the click rate prediction model converges. The environmental characteristics may include information such as time, place, etc. of the user initiated request. In addition, the final output of the click-through rate prediction model may be the pCTR value corresponding to the information.
After the trained click rate prediction model is obtained, for each information acquisition request of the user, the user characteristics and the environmental characteristics carried in the request and the information characteristics corresponding to the candidate information can be input into the click rate prediction model together, and then the pCTR value of the user to the candidate information can be obtained.
In addition, the user can continue to slide the screen or click a set area in the interface in the information flow display interface of the client to trigger and generate an information acquisition request, and the information acquisition request is sent to the electronic equipment where the server is located, so that the server is indicated to recommend new information with a set quantity.
Therefore, the click rate prediction value corresponding to each piece of information is output and obtained by combining the user characteristic, the environment characteristic and the information characteristic corresponding to each piece of information when the information acquisition request is initiated and utilizing the click rate prediction model, so that the click rate prediction result is more accurate, the interestingness of the user can be more accurately reflected, and the accuracy of the information recommendation result is improved.
It should be noted that, in the information recommendation method provided in the embodiment of the present application, the execution subject may be an information recommendation device, or a control module in the information recommendation device for executing the information recommendation method. The information recommendation device provided by the embodiment of the present application is described by taking an example of an information recommendation method executed by an information recommendation device.
Fig. 4 is a block diagram illustrating a structure of an information recommendation apparatus according to an exemplary embodiment.
As shown in fig. 4, the information recommendation apparatus 400 may include:
an obtaining module 401, configured to obtain M pieces of information associated with a user; wherein, the information arrangement sequence in the M pieces of information is a first sequence;
a determining module 402, configured to determine a similarity value between any two pieces of information in the M pieces of information and a click rate prediction value corresponding to each piece of information;
a rearrangement module 403, configured to rearrange the M pieces of information according to the similarity value and the click rate prediction value, to obtain M pieces of information arranged according to a second order;
a recommending module 404, configured to recommend the M pieces of information to the user according to the second order;
wherein M is a positive integer and M is more than or equal to 2.
The information recommendation device 400 is described in detail below, specifically as follows:
in one embodiment, the rearrangement module 403 may specifically include:
the construction submodule is used for constructing a DPP algorithm matrix in the determinant point process according to the similarity value and the click rate predicted value;
and the matrix calculation submodule is used for calculating the DPP algorithm matrix based on a greedy algorithm to obtain M pieces of information arranged according to a second sequence.
In one embodiment, the above-mentioned construction sub-module may specifically include:
the first setting unit is used for setting the value of the ith row and ith column element in the DPP algorithm matrix as the square of the click rate predicted value corresponding to the ith information in the M information;
the second setting unit is used for setting the value of the ith row and the jth column element in the DPP algorithm matrix as the product of the first click rate predicted value, the second click rate predicted value and the target similarity value; the first click rate predicted value is a click rate predicted value corresponding to the ith information in the M information, the second click rate predicted value is a click rate predicted value corresponding to the jth information in the M information, and the target similarity value is a similarity value between the ith information and the jth information;
the matrix construction unit is used for constructing and obtaining the DPP algorithm matrix based on the value of the ith row and ith column element and the value of the ith row and jth column element;
wherein i and j are positive integers, and i ≠ j.
In one embodiment, the above mentioned M pieces of information may include S pieces of first information and T pieces of second information;
the obtaining module 401 may specifically include:
the first obtaining submodule is used for obtaining S pieces of first information related to the user from an alternative information base according to a preset information recommendation strategy; and the number of the first and second groups,
the first obtaining sub-module is used for obtaining T pieces of second information which are not clicked by the user from historical recommendation information corresponding to the user;
wherein S, T is a positive integer.
In one embodiment, the information recommendation apparatus 400 may further include:
the information removing module is used for rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information which are arranged according to a second sequence, and then removing the T pieces of second information from the M pieces of information to obtain P pieces of information;
the information acquisition module is used for acquiring the first N pieces of information from the P pieces of information;
wherein P, N is a positive integer;
the recommending module 404 may specifically include:
and the information recommending submodule is used for recommending the N pieces of information to the user according to the second sequence.
In one embodiment, the determining module 402 may specifically include:
a feature extraction sub-module, configured to, for each of the M pieces of information, extract an information feature corresponding to the each piece of information from a plurality of dimensions;
the vector output submodule is used for respectively inputting the information characteristics corresponding to each piece of information into the information expression model and outputting to obtain the characteristic vector corresponding to each piece of information;
a vector obtaining submodule, configured to obtain, based on the feature vector corresponding to each piece of information, feature vectors corresponding to any two pieces of information in the M pieces of information, respectively;
and the similarity operator module is used for calculating a similarity value between any two pieces of information based on the feature vectors respectively corresponding to the any two pieces of information.
In one embodiment, the similarity operator module may specifically include:
the normalization unit is used for performing normalization processing on the feature vectors respectively corresponding to the arbitrary two pieces of information to obtain normalized feature vectors;
and the inner product calculating unit is used for carrying out inner product calculation on the normalized feature vectors corresponding to the arbitrary two pieces of information to obtain a similarity value between the arbitrary two pieces of information.
In one embodiment, the information recommendation apparatus 400 may further include:
the request receiving module is used for receiving an information acquisition request sent by a user before M pieces of information associated with the user are acquired; the information acquisition request comprises user characteristics and environment characteristics;
the determining module 402 may specifically include:
and the click rate prediction sub-module is used for inputting the information characteristics corresponding to each piece of information in the M pieces of information and the user characteristics and the environment characteristics to a click rate prediction model respectively and outputting to obtain a click rate prediction value corresponding to each piece of information.
Therefore, the similarity value between any two pieces of information in the M pieces of information to be selected and the click rate predicted value corresponding to each piece of information are determined, the similarity value and the click rate predicted value of the user are comprehensively considered, the M pieces of information are rearranged, and the click rate predicted value can reflect the interest tendency of the user, so that the rearranged information can ensure the diversity of the information on the watching sequence of the user and cater to the interest of the user, the diversity of the information is ensured, the interest of the user is considered, and the success rate of information recommendation is improved.
The information recommendation device in the embodiment of the present application may be a device, or may also be a component, an integrated circuit, or a chip in a terminal. The device may be a server, and the embodiment of the present application is not particularly limited.
The information recommendation device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system (Android), an iOS operating system, or other possible operating systems, which is not specifically limited in the embodiments of the present application.
The information recommendation device provided in the embodiment of the present application can implement each process implemented by the method embodiments of fig. 2 to fig. 3, and is not described here again to avoid repetition.
Optionally, as shown in fig. 5, an electronic device 500 is further provided in this embodiment of the present application, and includes a processor 501, a memory 502, and a program or an instruction stored in the memory 502 and executable on the processor 501, where the program or the instruction is executed by the processor 501 to implement each process of the above-mentioned information recommendation method embodiment, and can achieve the same technical effect, and in order to avoid repetition, details are not repeated here.
It should be noted that the electronic device in the embodiment of the present application includes the mobile electronic device and the non-mobile electronic device described above.
Fig. 6 is a schematic diagram of a hardware structure of an electronic device implementing an embodiment of the present application.
The electronic device 600 includes, but is not limited to: a radio frequency unit 601, a network module 602, an audio output unit 603, an input unit 604, a sensor 605, a display unit 606, a user input unit 607, an interface unit 608, a memory 609, a processor 610, and the like.
Those skilled in the art will appreciate that the electronic device 600 may also include various componentsA power supply (e.g., a battery) that supplies power may be logically connected to the processor 610 via a power management system, such that functions of managing charging, discharging, and power consumption are implemented via the power management system. Drawing (A)6The electronic device structures shown in the figures do not constitute limitations of the electronic device, and the electronic device may include more or less components than those shown, or combine some components, or arrange different components, and thus, the description is not repeated here.
Wherein, the processor 610 is configured to obtain M pieces of information associated with a user; wherein, the information arrangement sequence in the M pieces of information is a first sequence; determining a similarity value between any two pieces of information in the M pieces of information and a click rate predicted value corresponding to each piece of information; rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information arranged according to a second sequence; recommending the M pieces of information to the user according to the second sequence; wherein M is a positive integer and M is more than or equal to 2.
Therefore, the similarity value between any two pieces of information in the M pieces of information to be selected and the click rate predicted value corresponding to each piece of information are determined, the similarity value and the click rate predicted value of the user are comprehensively considered, the M pieces of information are rearranged, and the click rate predicted value can reflect the interest tendency of the user, so that the rearranged information can ensure the diversity of the information on the watching sequence of the user and cater to the interest of the user, the diversity of the information is ensured, the interest of the user is considered, and the success rate of information recommendation is improved.
Optionally, the processor 610 is further configured to construct a DPP algorithm matrix in a determinant point process according to the similarity value and the click rate predicted value; and calculating the DPP algorithm matrix based on a greedy algorithm to obtain M pieces of information arranged according to a second sequence.
Optionally, the processor 610 is further configured to set a value of an ith row and an ith column element in the DPP algorithm matrix to be a square of a click rate predicted value corresponding to an ith information in the M pieces of information; setting the value of the ith row and the jth column element in the DPP algorithm matrix as the product of a first click rate predicted value, a second click rate predicted value and a target similarity value; the first click rate predicted value is a click rate predicted value corresponding to the ith information in the M information, the second click rate predicted value is a click rate predicted value corresponding to the jth information in the M information, and the target similarity value is a similarity value between the ith information and the jth information; constructing and obtaining the DPP algorithm matrix based on the value of the ith row and ith column element and the value of the ith row and jth column element; wherein i and j are positive integers, and i ≠ j.
Optionally, the processor 610 is further configured to obtain S pieces of first information associated with the user from an alternative information base according to a preset information recommendation policy; acquiring T pieces of second information which are not clicked by the user from historical recommendation information corresponding to the user; wherein S, T is a positive integer.
Optionally, the processor 610 is further configured to remove the T second information from the M information, to obtain P information; acquiring the first N information from the P information; wherein P, N is a positive integer; and recommending the N pieces of information to the user according to the second sequence.
Optionally, the processor 610 is further configured to, for each of the M pieces of information, extract an information feature corresponding to each of the M pieces of information from multiple dimensions; inputting the information characteristics corresponding to each piece of information into an information expression model respectively, and outputting to obtain a characteristic vector corresponding to each piece of information; acquiring feature vectors corresponding to any two pieces of information in the M pieces of information respectively based on the feature vectors corresponding to each piece of information; and calculating a similarity value between any two pieces of information based on the feature vectors respectively corresponding to the any two pieces of information.
Optionally, the processor 610 is further configured to perform normalization processing on the feature vectors respectively corresponding to the two arbitrary pieces of information to obtain normalized feature vectors; and carrying out inner product calculation on the normalized feature vectors corresponding to the arbitrary two pieces of information to obtain a similarity value between the arbitrary two pieces of information.
Optionally, the network module 602 is configured to receive an information obtaining request sent by the user; the information acquisition request comprises user characteristics and environment characteristics;
the processor 610 is further configured to determine a click-through rate predicted value corresponding to each of the M pieces of information, and includes: inputting the information characteristics corresponding to each piece of information in the M pieces of information and the user characteristics and the environment characteristics to a click rate prediction model, and outputting to obtain a click rate prediction value corresponding to each piece of information.
Therefore, by utilizing a DPP greedy solving algorithm, combining an information expression model and a click rate prediction model, comprehensively considering the correlation and the user interest degree between information, automatically selecting optimal diversity recommendation information, improving the user experience and effectively solving the problems of the traditional diversity strategy.
It is to be understood that, in the embodiment of the present application, the input Unit 604 may include a Graphics Processing Unit (GPU) 6041 and a microphone 6042, and the Graphics Processing Unit 6041 processes image data of a still picture or a video obtained by an image capturing apparatus (such as a camera) in a video capturing mode or an image capturing mode. The display unit 606 may include a display panel 6061, and the display panel 6061 may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like. The user input unit 607 includes a touch panel 6071 and other input devices 6072. A touch panel 6071, also referred to as a touch screen. The touch panel 6071 may include two parts of a touch detection device and a touch controller. Other input devices 6072 may include, but are not limited to, a physical keyboard, function keys (e.g., volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which are not described in detail herein. The memory 609 may be used to store software programs as well as various data including, but not limited to, application programs and an operating system. The processor 610 may integrate an application processor, which primarily handles operating systems, user interfaces, applications, etc., and a modem processor, which primarily handles wireless communications. It will be appreciated that the modem processor described above may not be integrated into the processor 610.
The embodiment of the present application further provides a readable storage medium, where a program or an instruction is stored on the readable storage medium, and when the program or the instruction is executed by a processor, the program or the instruction implements each process of the above information recommendation method embodiment, and can achieve the same technical effect, and in order to avoid repetition, details are not repeated here.
The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and so on.
The embodiment of the present application further provides a chip, where the chip includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is configured to run a program or an instruction to implement each process of the above information recommendation method embodiment, and can achieve the same technical effect, and in order to avoid repetition, the chip is not described here again.
It should be understood that the chips mentioned in the embodiments of the present application may also be referred to as system-on-chip, system-on-chip or system-on-chip, etc.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element. Further, it should be noted that the scope of the methods and apparatus of the embodiments of the present application is not limited to performing the functions in the order illustrated or discussed, but may include performing the functions in a substantially simultaneous manner or in a reverse order based on the functions involved, e.g., the methods described may be performed in an order different than that described, and various steps may be added, omitted, or combined. In addition, features described with reference to certain examples may be combined in other examples.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present application may be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present application.
While the present embodiments have been described with reference to the accompanying drawings, it is to be understood that the invention is not limited to the precise embodiments described above, which are meant to be illustrative and not restrictive, and that various changes may be made therein by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (12)
1. An information recommendation method, comprising:
acquiring M pieces of information associated with a user; wherein, the information arrangement sequence in the M pieces of information is a first sequence;
determining a similarity value between any two pieces of information in the M pieces of information and a click rate predicted value corresponding to each piece of information;
rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information arranged according to a second sequence;
recommending the M pieces of information to the user according to the second sequence;
wherein M is a positive integer and M is more than or equal to 2.
2. The method according to claim 1, wherein the rearranging the M information according to the similarity value and the click-through rate prediction value to obtain M information arranged according to a second order comprises:
constructing a DPP algorithm matrix of the determinant point process according to the similarity value and the click rate predicted value;
and calculating the DPP algorithm matrix based on a greedy algorithm to obtain M pieces of information arranged according to a second sequence.
3. The method according to claim 2, wherein constructing a determinant point process DPP algorithm matrix according to the similarity value and the click-through rate prediction value comprises:
setting the value of the ith row and ith column element in the DPP algorithm matrix as the square of the click rate predicted value corresponding to the ith information in the M information;
setting the value of the ith row and the jth column element in the DPP algorithm matrix as the product of a first click rate predicted value, a second click rate predicted value and a target similarity value; the first click rate predicted value is a click rate predicted value corresponding to the ith information in the M information, the second click rate predicted value is a click rate predicted value corresponding to the jth information in the M information, and the target similarity value is a similarity value between the ith information and the jth information;
constructing and obtaining the DPP algorithm matrix based on the value of the ith row and ith column element and the value of the ith row and jth column element;
wherein i and j are positive integers, and i ≠ j.
4. The method according to claim 1, wherein the M pieces of information include S pieces of first information and T pieces of second information;
the acquiring of the M pieces of information associated with the user includes:
s pieces of first information related to the user are obtained from an alternative information base according to a preset information recommendation strategy; and the number of the first and second groups,
acquiring T pieces of second information which are not clicked by the user from historical recommendation information corresponding to the user;
wherein S, T is a positive integer.
5. The method of claim 4, wherein after rearranging the M information according to the similarity value and the click-through rate prediction value to obtain M information arranged in a second order, the method further comprises:
removing the T second information from the M information to obtain P information;
acquiring the first N information from the P information;
wherein P, N is a positive integer;
the recommending the M pieces of information to the user according to the second sequence comprises:
and recommending the N pieces of information to the user according to the second sequence.
6. The method of claim 1, wherein determining a similarity value between any two of the M information comprises:
for each of the M pieces of information, extracting information features corresponding to the each information from a plurality of dimensions;
inputting the information characteristics corresponding to each piece of information into an information expression model respectively, and outputting to obtain a characteristic vector corresponding to each piece of information;
acquiring feature vectors corresponding to any two pieces of information in the M pieces of information respectively based on the feature vectors corresponding to each piece of information;
and calculating a similarity value between any two pieces of information based on the feature vectors respectively corresponding to the any two pieces of information.
7. An information recommendation apparatus, comprising:
an acquisition module for acquiring M pieces of information associated with a user; wherein, the information arrangement sequence in the M pieces of information is a first sequence;
the determining module is used for determining a similarity value between any two pieces of information in the M pieces of information and a click rate predicted value corresponding to each piece of information;
the rearrangement module is used for rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information which are arranged according to a second sequence;
a recommending module, configured to recommend the M pieces of information to the user according to the second order;
wherein M is a positive integer and M is more than or equal to 2.
8. The apparatus of claim 7, wherein the reordering module comprises:
the construction submodule is used for constructing a DPP algorithm matrix in the determinant point process according to the similarity value and the click rate predicted value;
and the matrix calculation submodule is used for calculating the DPP algorithm matrix based on a greedy algorithm to obtain M pieces of information arranged according to a second sequence.
9. The apparatus of claim 8, wherein the build submodule comprises:
the first setting unit is used for setting the value of the ith row and ith column element in the DPP algorithm matrix as the square of the click rate predicted value corresponding to the ith information in the M information;
the second setting unit is used for setting the value of the ith row and the jth column element in the DPP algorithm matrix as the product of the first click rate predicted value, the second click rate predicted value and the target similarity value; the first click rate predicted value is a click rate predicted value corresponding to the ith information in the M information, the second click rate predicted value is a click rate predicted value corresponding to the jth information in the M information, and the target similarity value is a similarity value between the ith information and the jth information;
the matrix construction unit is used for constructing and obtaining the DPP algorithm matrix based on the value of the ith row and ith column element and the value of the ith row and jth column element;
wherein i and j are positive integers, and i ≠ j.
10. The apparatus according to claim 7, wherein the M pieces of information include S pieces of first information and T pieces of second information;
the acquisition module includes:
the first obtaining submodule is used for obtaining S pieces of first information related to the user from an alternative information base according to a preset information recommendation strategy; and the number of the first and second groups,
the first obtaining sub-module is used for obtaining T pieces of second information which are not clicked by the user from historical recommendation information corresponding to the user;
wherein S, T is a positive integer.
11. The apparatus of claim 10, further comprising:
the information removing module is used for rearranging the M pieces of information according to the similarity value and the click rate predicted value to obtain M pieces of information which are arranged according to a second sequence, and then removing the T pieces of second information from the M pieces of information to obtain P pieces of information;
the information acquisition module is used for acquiring the first N pieces of information from the P pieces of information;
wherein P, N is a positive integer;
the recommendation module comprises:
and the information recommending submodule is used for recommending the N pieces of information to the user according to the second sequence.
12. The apparatus of claim 7, wherein the determining module comprises:
a feature extraction sub-module, configured to, for each of the M pieces of information, extract an information feature corresponding to the each piece of information from a plurality of dimensions;
the vector output submodule is used for respectively inputting the information characteristics corresponding to each piece of information into the information expression model and outputting to obtain the characteristic vector corresponding to each piece of information;
a vector obtaining submodule, configured to obtain, based on the feature vector corresponding to each piece of information, feature vectors corresponding to any two pieces of information in the M pieces of information, respectively;
and the similarity operator module is used for calculating a similarity value between any two pieces of information based on the feature vectors respectively corresponding to the any two pieces of information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110762301.8A CN113515701A (en) | 2021-07-06 | 2021-07-06 | Information recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110762301.8A CN113515701A (en) | 2021-07-06 | 2021-07-06 | Information recommendation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113515701A true CN113515701A (en) | 2021-10-19 |
Family
ID=78066865
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110762301.8A Pending CN113515701A (en) | 2021-07-06 | 2021-07-06 | Information recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113515701A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023087914A1 (en) * | 2021-11-19 | 2023-05-25 | 腾讯科技（深圳）有限公司 | Method and apparatus for selecting recommended content, and device, storage medium and program product |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086439A (en) * | 2018-08-15 | 2018-12-25 | 腾讯科技（深圳）有限公司 | Information recommendation method and device |
| CN110781377A (en) * | 2019-09-03 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Article recommendation method and device |
| CN111400615A (en) * | 2020-03-19 | 2020-07-10 | 腾讯科技（深圳）有限公司 | Resource recommendation method, device, equipment and storage medium |
| CN111428133A (en) * | 2020-03-19 | 2020-07-17 | 腾讯科技（北京）有限公司 | Artificial intelligence based recommendation method and device, electronic equipment and storage medium |
| CN111475730A (en) * | 2020-04-09 | 2020-07-31 | 腾讯科技（北京）有限公司 | Information recommendation method and device based on artificial intelligence and electronic equipment |
-
2021
- 2021-07-06 CN CN202110762301.8A patent/CN113515701A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086439A (en) * | 2018-08-15 | 2018-12-25 | 腾讯科技（深圳）有限公司 | Information recommendation method and device |
| CN110781377A (en) * | 2019-09-03 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Article recommendation method and device |
| CN111400615A (en) * | 2020-03-19 | 2020-07-10 | 腾讯科技（深圳）有限公司 | Resource recommendation method, device, equipment and storage medium |
| CN111428133A (en) * | 2020-03-19 | 2020-07-17 | 腾讯科技（北京）有限公司 | Artificial intelligence based recommendation method and device, electronic equipment and storage medium |
| CN111475730A (en) * | 2020-04-09 | 2020-07-31 | 腾讯科技（北京）有限公司 | Information recommendation method and device based on artificial intelligence and electronic equipment |
Non-Patent Citations (1)
| Title |
|---|
| 夕小瑶: "聊聊如何提升推荐系统的结果多样性", Retrieved from the Internet <URL:https://blog.csdn.net/xixiaoyaoww/article/details/112551295> * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023087914A1 (en) * | 2021-11-19 | 2023-05-25 | 腾讯科技（深圳）有限公司 | Method and apparatus for selecting recommended content, and device, storage medium and program product |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111079022B (en) | Personalized recommendation method, device, equipment and medium based on federal learning | |
| CN109819284B (en) | Short video recommendation method and device, computer equipment and storage medium | |
| US20150169710A1 (en) | Method and apparatus for providing search results | |
| CN109190049B (en) | Keyword recommendation method, system, electronic device and computer readable medium | |
| US20220284327A1 (en) | Resource pushing method and apparatus, device, and storage medium | |
| US9576305B2 (en) | Detecting competitive product reviews | |
| EP4180991A1 (en) | Neural network distillation method and apparatus | |
| US20180217972A1 (en) | Populating values in a spreadsheet using semantic cues | |
| CN109993627B (en) | Recommendation method, recommendation model training device and storage medium | |
| US20220261591A1 (en) | Data processing method and apparatus | |
| CN113688310B (en) | Content recommendation method, device, equipment and storage medium | |
| CN110008396B (en) | Object information pushing method, device, equipment and computer readable storage medium | |
| CN113407814B (en) | Text searching method and device, readable medium and electronic equipment | |
| CN113190741A (en) | Searching method, searching device, electronic equipment and storage medium | |
| CN112907334A (en) | Object recommendation method and device | |
| CN112989146A (en) | Method, apparatus, device, medium, and program product for recommending resources to a target user | |
| US10229212B2 (en) | Identifying Abandonment Using Gesture Movement | |
| CN113869063A (en) | Data recommendation method and device, electronic equipment and storage medium | |
| CN113869377A (en) | Training method and device and electronic equipment | |
| CN113515701A (en) | Information recommendation method and device | |
| CN117573973A (en) | Resource recommendation method, device, electronic equipment and storage medium | |
| CN112766995A (en) | Article recommendation method and device, terminal device and storage medium | |
| CN112231546B (en) | Heterogeneous document ordering method, heterogeneous document ordering model training method and device | |
| CN115309487A (en) | Display method, display device, electronic equipment and readable storage medium | |
| CN110275986B (en) | Video recommendation method based on collaborative filtering, server and computer storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |