CN109242654A - A kind of item recommendation method and system - Google Patents
A kind of item recommendation method and system Download PDFInfo
- Publication number
- CN109242654A CN109242654A CN201811359261.7A CN201811359261A CN109242654A CN 109242654 A CN109242654 A CN 109242654A CN 201811359261 A CN201811359261 A CN 201811359261A CN 109242654 A CN109242654 A CN 109242654A
- Authority
- CN
- China
- Prior art keywords
- user
- purchase
- month
- time
- item
- 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 47
- 238000012549 training Methods 0.000 claims description 61
- 230000006399 behavior Effects 0.000 claims description 27
- 238000002790 cross-validation Methods 0.000 claims description 15
- 230000002452 interceptive effect Effects 0.000 claims description 10
- 238000012163 sequencing technique Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims 3
- 238000005457 optimization Methods 0.000 claims 1
- 230000010365 information processing Effects 0.000 abstract description 2
- 238000012360 testing method Methods 0.000 description 9
- 238000010276 construction Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000012552 review Methods 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 206010000117 Abnormal behaviour Diseases 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000000586 desensitisation Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of item recommendation method, system, equipment and a kind of computer readable storage mediums, are related to technical field of information processing.Historical data of the present invention according to electric business platform for a period of time, and characteristic element is generated by multiple time windows of a period of time building, then the characteristic element is called to be trained collection building according to multiple time windows, obtain LightGBM model, and then regression forecasting is carried out, it obtains including buying behavior and the prediction result progress article recommendation for buying degree, not only increases the accuracy that user buys prediction model, the precision of article recommendation is also improved, user's shopping experience is greatly improved.
Description
Technical Field
The invention relates to the technical field of information processing, in particular to a LightGBM-based item recommendation method, a system and equipment and a computer-readable storage medium.
Background
With the rapid development of e-commerce websites, a large amount of user data including user basic information, user purchasing behavior data, user browsing and collecting data, article information and the like are accumulated in e-commerce websites. How to analyze and mine the accumulated mass user data and construct a user purchase model to recommend articles to users, so that accurate marketing is realized, and the click rate and purchase rate of an e-commerce platform are improved.
In the prediction models used by the E-commerce website at present, single models such as logistic regression and decision trees are applied more, the models have the advantages of stable operation and strong interpretability, but in recent years, with diversification of E-commerce services, the E-commerce services have already finished permeation from a PC (personal computer) end to a mobile end, with diversification of services, user characteristics and service scenes are increasingly complex, and the single models such as logistic regression and decision trees have poor adaptability to complex service scenes and user characteristics and are not adapted to new service scenes and increasingly complex service characteristics; in addition, in the existing model prediction method, users are classified into 'purchase' and 'non-purchase' through a model, and then the users predicted to be 'purchase' are recommended, so that the method is relatively rough and cannot subdivide the degree and possibility of the user purchase. Therefore, an article recommendation system capable of adapting to complex features is needed to improve the purchasing prediction accuracy and the article recommendation accuracy of the user.
Disclosure of Invention
The embodiment of the invention provides an article recommendation method and system, which improve the accuracy of purchasing a prediction model by a user and improve the article recommendation accuracy, so that the shopping experience of the user is improved.
In order to achieve the above object, an embodiment of the present invention provides an item recommendation method applied to an item recommendation device, where the method includes:
acquiring historical data of a period of time, wherein the historical data comprises user information, user order information, user browsing item information, user collection item information or user evaluation information of items;
generating meta-features from the historical data, the meta-features including: user information, article information, interactive behavior characteristics, concern behavior characteristics and user purchase conversion rate;
constructing a time window in the period of time, wherein the time window is obtained by calculating the time width from the label month to the front; the label month refers to a month between a month of a maximum window calculated backwards from a starting month of the period of time and an ending month of the period of time, and the number of the time windows is at least two;
respectively analyzing the meta-features according to the sample set expanded by the time window to construct a training set;
training a LightGBM model according to the training set, wherein the LightGBM model is trained by adopting a cross validation method;
generating a regression prediction result according to the LightGBM model, wherein the regression prediction result comprises user and purchase orders;
recommending the item to the user according to the regression prediction result.
Correspondingly, the embodiment of the invention also provides an article recommender system, which comprises a data acquisition unit, a feature generation unit, a training set construction unit, a model construction unit, an analysis unit and a recommendation unit;
the data acquisition unit is used for acquiring historical data of a period of time, wherein the historical data comprises user information, article information, user order information, user browsing article information, user collecting article information or user evaluation information of articles;
the feature generation unit is configured to generate a meta-feature according to the history data, where the meta-feature includes: user information, interactive behavior characteristics, attention behavior characteristics and user purchase conversion rate;
the training set constructing unit is used for constructing a time window in the period of time, and the time window is obtained by forward reckoning the time width from the label month; the label month refers to a month between a month of a maximum window calculated backwards from a starting month of the period of time and an ending month of the period of time, and the number of the time windows is at least two; and
respectively analyzing the meta-features according to the sample set expanded by the time window to construct a training set;
the model construction unit is used for training the LightGBM model according to the training set, wherein the LightGBM model is trained by adopting a cross validation method;
the analysis unit is used for generating a regression prediction result according to the LightGBM model, and the regression prediction result comprises the user purchase order number;
and the recommending unit is used for recommending the articles to the user according to the regression prediction result.
According to the method, historical data of the e-commerce platform for a period of time is obtained, a plurality of time windows are built for the period of time, the feature elements are generated according to the historical data, then the feature elements are called according to the time windows to build a training set, cross verification is conducted according to the training set to build a LightGBM model, regression prediction is conducted according to the LightGBM model, prediction results including purchasing behaviors and purchasing degrees are obtained to conduct article recommendation, the accuracy of a user for purchasing the prediction model is improved, the accuracy of article recommendation is improved, and the shopping experience of the user is greatly improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart of a method for recommending items according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating an example of a time window construction according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a five-fold cross-validation method according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an item recommendation system according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
An embodiment of the present invention provides an article recommendation method, which is shown in fig. 1 and is applied to an article recommendation device, and the method specifically includes:
s101, obtaining historical data of a period of time, wherein the historical data comprises user information, article information, user order information, user browsing article information, user collecting article information or user evaluation information of articles.
The period of time refers to a period of time derived from a current predicted behavior time, and may be a year, specifically, depending on the historical data and the prediction purpose, which is not limited in the embodiment of the present invention.
The user information may include, but is not limited to, a user number, a user age, a user gender, and a user rating; the item information may include, but is not limited to, item number, item price, item category number, and item parameters; the user order information may include, but is not limited to, a user number, an item number, an order number, a placing date, a placing area, and a quantity of items purchased; the user browsed item information and user collected item information may include, but is not limited to, a user number, an item number, an action date and an action number; the user rating information of the item may include, but is not limited to, a user number, a review time, an order number, and a review grade. The user age and user gender may be information after being subjected to a desensitization process.
S102, generating meta-characteristics according to the historical data, wherein the meta-characteristics comprise: user information, interactive behavior characteristics, attention behavior characteristics and user purchase conversion rate.
Wherein the user information may include, but is not limited to, user age, user gender, and user rating.
The interactive behavior features may include, but are not limited to, purchasing features, geographic features, parameter information, user spending, user purchase concentration, and user item loyalty. Specifically, the purchase characteristics may include, but are not limited to, an amount of orders, a number of items, a type of items, a number of purchases, a number of days with purchase, and a number of months with purchase; the geographic features may include, but are not limited to, the number of places the user has placed a order and the number of places the user has placed the largest number of orders; the parameter information may include, but is not limited to, a maximum value, a minimum value, an average value, and a median of the price of the user's purchase item; the user cost may include, but is not limited to, a total cost of the user; the user purchase concentration ratio is the ratio of the times of purchasing the articles by the user to the types of the purchased articles; the user item loyalty is the maximum number of times a user purchases the same item.
The behavioral characteristics of interest may include, but are not limited to, the number of items the user browses/collects/adds shopping carts, the types of items the user browses/collects/adds shopping carts, and the number of days the user browses/collects/adds shopping carts.
The meta-feature may be presented in a form of a table, or may be presented in other forms, and the embodiment of the present invention is not limited; the above-mentioned various element features may be presented separately or collectively, and the embodiments of the present invention are not limited thereto.
S103, constructing a time window in the period of time, wherein the time window is obtained by forward calculating the time width from the label month; the label month refers to a month between a month of a maximum window calculated backwards from a starting month of the period of time to an ending month of the period of time, and the number of the time windows is at least two.
Specifically, 5 time windows can be constructed, the maximum time window is 6 months, and the minimum time window is 7 days, including: 7 days, 14 days, 1 month, 3 months and 6 months are respectively estimated forward from the month of the predicted behavior occurrence.
For example, data of 2016 month 5 to 2017 month 4 needs to be used for predicting data of 2017 month 5 of the user, and then the item recommendation is completed. FIG. 2 is a block diagram of an exemplary time window according to an embodiment of the present invention.
As shown in a in fig. 2, the historical data used in the current prediction is the historical data of the user in the year 5 months before 2017, and the label months constructed by determining the time window according to the historical data are respectively: 2016.11, 2016.12, 2017.01, 2017.02, 2017.03 and 2017.04. As shown in B in fig. 2, which is a diagram illustrating the construction of the prediction time window at this time, the construction of the time window can be performed by respectively calculating forward 7 days, 14 days, 1 month, 3 months and 6 months with the 6 labeled months as time bases. By such a construction method, the sample of the period of time can be expanded by 6 times. In addition, through the gradient division of the time window granularity, the characteristic points which can reflect the characteristics of the purchasing and attention behaviors of the user can be fully utilized, so that the subsequent prediction is more accurate. The time width may be determined by an empirical value and an offline test, or may be determined in other manners, and specifically, the embodiment of the present invention is not limited.
And S104, respectively analyzing the meta-features according to the sample set expanded by the time window to construct a training set.
S105, training the LightGBM model according to the training set, wherein the LightGBM model is trained by adopting a cross-validation method.
Specifically, a K-fold cross validation method may be adopted to train the LightGBM model, where the K-fold cross validation method refers to: and randomly dividing the training set into K packets, taking one packet as a test set each time, and taking the rest K-1 packets as sub-training sets for training.
Taking the 5-fold cross validation method as an example, as shown in fig. 3, a schematic diagram of a five-fold cross validation method according to an embodiment of the present invention is shown. The training set is randomly divided into 5 packets, namely, the sub-training sets 1, 2, 3, 4 and 5 are respectively used as test sets, and model training is carried out by using other sub-training sets except the test sets. Taking the sub-training set 1 as the test set as an example, then training by using the sub-training sets 2, 3, 4, 5 and the label 1 corresponding to the test set to obtain the model 1, and training the other four models by the same method.
And S106, generating a regression prediction result according to the LightGBM model, wherein the regression prediction result can comprise the amount of orders purchased by the user.
Specifically, taking training models 1, 2, 3, 4, and 5, which are obtained by training a training set obtained by expanding 5 time windows and illustrated in fig. 3, as an example, prediction results are respectively generated according to each model and each sub-training set, taking the sub-training set 1 illustrated in fig. 3 as an example, a test set is predicted according to the model 1 to obtain a prediction result 1, and prediction results of other four sub-training sets are obtained by the same method. Then, calculating an average value of the prediction results, namely, obtaining a regression prediction result, where the regression prediction result may include not only whether the user purchases the item, but also: if the predicted result is that the user purchases the goods, the user may purchase the amount of orders.
The prediction result may be presented in a table form, or may be presented in other forms, and the embodiment of the present invention is not limited thereto, and table 1 below is described in only one table form, and the table may be in other forms, and the embodiment of the present invention is not limited to the table form.
TABLE 1
| User number | Predicting purchase order numbers |
| 1 | 0.0213 |
| 2 | 2.9213 |
| 3 | 1.3489 |
| 4 | 0.0002 |
| 5 | 3.4156 |
| 6 | 0.8993 |
And S107, recommending the article to the user according to the regression prediction result.
Further, the recommending the item to the user according to the regression prediction result comprises,
and sequencing the regression prediction results from high to low, and recommending the articles to the user according to the sequencing results.
Specifically, the items are ranked according to the purchase orders in the regression prediction result, and the corresponding users are recommended according to the ranking result, for example, the prediction result ranked 50 before the purchase orders is selected to recommend the items, specifically, how to recommend the items according to the ranking result, which may be determined according to the specific situation, and the embodiment of the present invention does not limit this.
Further, before the analyzing the meta-features respectively according to the time windows and constructing a training set, the method further includes:
the abnormal training samples are discarded. The abnormal training sample may be a meta-feature of the e-commerce promotion time period, or a meta-feature of the user purchasing behavior abnormal time period caused by other reasons, which is not limited in the embodiment of the present invention.
Further, the method further comprises the step of,
and collecting purchase data of the recommended user after receiving the item recommendation, and evaluating the regression prediction result according to the purchase data and the regression prediction result.
Specifically, according to the formula: the user recommended purchase rate is the number of users actually purchased/total number of recommended users, wherein the total number of recommended users can be obtained from the regression prediction result; the actual purchased user number can be obtained from the purchase data;
and evaluating the regression prediction result according to the user recommended purchase rate.
According to the method, historical data of the e-commerce platform for a period of time is obtained, a plurality of time windows are built for the period of time, the feature elements are generated according to the historical data, then the feature elements are called according to the time windows to build a training set, a LightGBM model is built according to the training set, regression prediction is carried out according to the LightGBM model, a prediction result comprising purchasing behaviors and purchasing degrees is obtained to carry out article recommendation, the accuracy of a user purchasing the prediction model is improved, the accuracy of article recommendation is improved, and the shopping experience of the user is greatly improved.
An article recommendation system 10 according to the present invention is shown in fig. 4, where the article recommendation system 10 includes a data obtaining unit 110, a feature generating unit 120, a training set constructing unit 130, a model constructing unit 140, an analyzing unit 150, and a recommending unit 160. Fig. 2 is a schematic diagram, and does not limit other modules of the item recommendation system and the structural relationship of the modules. Wherein,
the data acquiring unit 110 is configured to acquire historical data of a period of time, where the historical data includes user information, item information, user order information, user browsing item information, user favorite item information, or user evaluation information of an item.
The period of time refers to a period of time derived from a current predicted behavior time, and may be a year, specifically, depending on the historical data and the prediction purpose, which is not limited in the embodiment of the present invention.
The user information may include, but is not limited to, a user number, a user age, a user gender, and a user rating; the item information may include, but is not limited to, item number, item price, item category number, and item parameters; the user order information may include, but is not limited to, a user number, an item number, an order number, a placing date, a placing area, and a quantity of items purchased; the user browsed item information and user collected item information may include, but is not limited to, a user number, an item number, an action date and an action number; the user rating information of the item may include, but is not limited to, a user number, a review time, an order number, and a review grade. The user age and user gender may be information after being subjected to a desensitization process.
A feature generating unit 120, configured to generate a meta-feature according to the history data, where the meta-feature includes: user information, interactive behavior characteristics, attention behavior characteristics and user purchase conversion rate.
Wherein the user information may include, but is not limited to, user age, user gender, and user rating.
The interactive behavior features may include, but are not limited to, purchasing features, geographic features, parameter information, user spending, user purchase concentration, and user item loyalty. Specifically, the purchase characteristics may include, but are not limited to, an amount of orders, a number of items, a type of items, a number of purchases, a number of days with purchase, and a number of months with purchase; the geographic features may include, but are not limited to, the number of places the user has placed a order and the number of places the user has placed the largest number of orders; the parameter information may include, but is not limited to, a maximum value, a minimum value, an average value, and a median of the price of the user's purchase item; the user cost may include, but is not limited to, a total cost of the user; the user purchase concentration ratio is the ratio of the times of purchasing the articles by the user to the types of the purchased articles; the user item loyalty is the maximum number of times a user purchases the same item.
The behavioral characteristics of interest may include, but are not limited to, the number of items the user browses/collects/adds shopping carts, the types of items the user browses/collects/adds shopping carts, and the number of days the user browses/collects/adds shopping carts.
The meta-feature may be presented in a form of a table, or may be presented in other forms, and the embodiment of the present invention is not limited; the above-mentioned various element features may be presented separately or collectively, and the embodiments of the present invention are not limited thereto.
A training set constructing unit 130, configured to construct a time window within the period of time, where the time window is obtained by calculating a time width from a label month to the front; the label month refers to a month between a month of a maximum window calculated backwards from a starting month of the period of time and an ending month of the period of time, and the number of the time windows is at least two; and
and respectively analyzing the meta-features according to the sample set expanded by the time window to construct a training set.
Specifically, 5 time windows can be constructed, the maximum time window is 6 months, and the minimum time window is 7 days, including: 7 days, 14 days, 1 month, 3 months and 6 months are respectively estimated forward from the month of the predicted behavior occurrence.
The time width may be determined by an empirical value and an offline test, or may be determined in other manners, and specifically, the embodiment of the present invention is not limited.
And a model building unit 140, configured to train the LightGBM model according to the training set, where the LightGBM model is trained by using a cross-validation method.
Specifically, a K-fold cross validation method may be adopted to train the LightGBM model, where the K-fold cross validation method refers to: and randomly dividing the training set into K packets, taking one packet as a test set each time, and taking the rest K-1 packets as sub-training sets for training.
An analyzing unit 150, configured to generate a regression prediction result according to the LightGBM model, where the regression prediction result may include the amount of orders purchased by the user.
Specifically, taking the training models 1, 2, 3, 4, and 5 obtained by training the training set expanded according to the 5 time windows as an example, prediction results may be respectively generated according to each training model and each sub-training set, and then an average value of each prediction result is calculated, which is a regression prediction result, where the regression prediction result may include whether the user purchases an article, and may further include: if the predicted result is that the user purchases the goods, the user may purchase the amount of orders.
The prediction result may be presented in a table form, or may be presented in other forms, and the embodiment of the present invention is not limited.
And the recommending unit 160 is used for recommending the item to the user according to the regression prediction result.
Further, the recommending the item to the user according to the regression prediction result comprises,
and sequencing the regression prediction results from high to low, and recommending the articles to the user according to the sequencing results.
Specifically, the items are ranked according to the purchase orders in the regression prediction result, and the corresponding users are recommended according to the ranking result, for example, the prediction result ranked 50 before the purchase orders is selected to recommend the items, specifically, how to recommend the items according to the ranking result, which may be determined according to the specific situation, and the embodiment of the present invention does not limit this.
Further, the feature generation unit 120 is further configured to,
and discarding abnormal training samples before respectively analyzing the meta-features according to the time window and constructing a training set. The abnormal training sample may be a meta-feature of the e-commerce promotion time period, or a meta-feature of the user purchasing behavior abnormal time period caused by other reasons, which is not limited in the embodiment of the present invention.
Further, the item recommendation system 10 further comprises,
and the optimizing and evaluating unit is used for collecting purchase data of the recommended user after receiving the item recommendation and evaluating the regression prediction result according to the purchase data and the regression prediction result.
Specifically, according to the formula: the user recommended purchase rate is the number of users actually purchased/total number of recommended users, wherein the total number of recommended users can be obtained from the regression prediction result; the actual purchased user number can be obtained from the purchase data;
and evaluating the regression prediction result according to the user recommended purchase rate.
According to the method, historical data of the e-commerce platform for a period of time is obtained, a plurality of time windows are built for the period of time, the feature elements are generated according to the historical data, then the feature elements are called according to the time windows to build a training set, a LightGBM model is built according to the training set, regression prediction is carried out according to the LightGBM model, a prediction result comprising purchasing behaviors and purchasing degrees is obtained to carry out article recommendation, the accuracy of a user purchasing the prediction model is improved, the accuracy of article recommendation is improved, and the shopping experience of the user is greatly improved.
In the several embodiments provided in the present application, it should be understood that the disclosed method and system may be implemented in other ways. For example, the above-described system embodiments are merely illustrative, and for example, the division of the functional blocks is only one logical division, and other divisions may be realized in practice, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing unit, or each unit may be physically included alone, or two or more units may be integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.
The integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium. The software functional unit is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute some steps of the methods according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (14)
1. An item recommendation method, characterized in that the method comprises:
acquiring historical data of a period of time, wherein the historical data comprises user information, item information, user order information, user browsed item information, user collected item information or user evaluation information of items;
generating meta-features from the historical data, the meta-features including: user information, interactive behavior characteristics, attention behavior characteristics and user purchase conversion rate;
constructing a time window in the period of time, wherein the time window is obtained by calculating the time width from the label month to the front; the label month refers to a month between a month of a maximum window calculated backwards from a starting month of the period of time and an ending month of the period of time, and the number of the time windows is at least two;
respectively analyzing the meta-features according to the sample set expanded by the time window to construct a training set;
training a LightGBM model according to the training set, wherein the LightGBM model is trained by adopting a cross validation method;
generating a regression prediction result according to the LightGBM model, wherein the regression prediction result comprises the purchase amount of orders of the user;
recommending the item to the user according to the regression prediction result.
2. The item recommendation method according to claim 1, wherein the interactive behavior features comprise: purchase characteristics, geographic characteristics, parameter information, user spending, user purchase concentration, or user item loyalty;
the attention behavior features include: browse, collect, or add shopping cart features.
3. The item recommendation method according to claim 2, wherein recommending items to the user based on the regression prediction result comprises:
and sequencing the regression prediction results from high to low, and recommending the articles to the corresponding users according to the sequencing results.
4. The item recommendation method according to claim 3, wherein said time window comprises: one month before the label month, 7 days, 14 days, 1 month, 3 months and 6 months are respectively estimated forward.
5. The item recommendation method of claim 4, wherein generating regression prediction results according to said LightGBM model comprises:
and respectively obtaining prediction results according to the LightGBM model obtained after training by adopting a cross-validation method, and calculating the average value of the prediction results to be used as a regression prediction result.
6. The item recommendation method according to claim 5, further comprising:
collecting purchase data after a recommended user receives item recommendation, wherein the purchase data comprises the number of users who actually purchase;
evaluating the regression prediction result according to the user recommended purchase rate;
and the user recommended purchase rate is the number of actually purchased users/total number of recommended users, wherein the total number of recommended users is obtained from the regression prediction result.
7. An item recommendation system, the system comprising:
the data acquisition unit is used for acquiring historical data of a period of time, wherein the historical data comprises user information, article information, user order information, user browsing article information, user collecting article information or user evaluation information of articles;
a feature generation unit configured to generate a meta-feature from the history data, the meta-feature including: user information, interactive behavior characteristics, attention behavior characteristics and user purchase conversion rate;
the training set constructing unit is used for constructing a time window in the period of time, and the time window is obtained by forward reckoning the time width from the label month; the label month refers to a month between a month of a maximum window calculated backwards from a starting month of the period of time and an ending month of the period of time, and the number of the time windows is at least two; and
respectively analyzing the meta-features according to the sample set expanded by the time window to construct a training set;
the model building unit is used for training the LightGBM model according to the training set, wherein the LightGBM model is trained by adopting a cross validation method;
an analysis unit, configured to generate a regression prediction result according to the LightGBM model, where the regression prediction result includes a user purchase order number;
and the recommending unit is used for recommending the articles to the user according to the regression prediction result.
8. The item recommendation system according to claim 7, wherein the interactive behavior features comprise: purchase characteristics, geographic characteristics, parameter information, user spending, user purchase concentration, or user item loyalty;
the attention behavior features include: browse, collect, or add shopping cart features.
9. The item recommendation system of claim 8, wherein said recommending items to the user based on the regression prediction comprises:
and sequencing the regression prediction results from high to low, and recommending the articles to the corresponding users according to the sequencing results.
10. The item recommendation system according to claim 9, wherein said time window comprises: 7 days, 14 days, 1 month, 3 months and 6 months are respectively estimated from the labeled month onwards.
11. The item recommendation system of claim 10, wherein said generating regression predictions from said LightGBM model comprises:
and respectively obtaining prediction results according to the LightGBM model obtained after training by adopting a cross-validation method, and calculating the average value of the prediction results to be used as a regression prediction result.
12. The item recommendation system according to claim 11, further comprising:
the model optimization unit is used for collecting purchase data after the recommended user receives item recommendation, and the purchase data comprises the number of users who actually purchase;
evaluating the regression prediction result according to the user recommended purchase rate;
and the user recommended purchase rate is the number of actually purchased users/total number of recommended users, wherein the total number of recommended users is obtained from the regression prediction result.
13. An item recommendation device, characterized in that the device comprises:
a memory for storing a computer program;
a processor for executing the computer program to implement the item recommendation method of any one of claims 1-6.
14. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the item recommendation method according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811359261.7A CN109242654A (en) | 2018-11-15 | 2018-11-15 | A kind of item recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811359261.7A CN109242654A (en) | 2018-11-15 | 2018-11-15 | A kind of item recommendation method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109242654A true CN109242654A (en) | 2019-01-18 |
Family
ID=65075355
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811359261.7A Pending CN109242654A (en) | 2018-11-15 | 2018-11-15 | A kind of item recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109242654A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163525A (en) * | 2019-05-29 | 2019-08-23 | 中国联合网络通信集团有限公司 | Terminal recommended method and terminal recommender system |
| CN110415048A (en) * | 2019-08-06 | 2019-11-05 | 即悟(上海)智能科技有限公司 | A kind of advertisement recommended method, equipment, system, computer equipment and storage medium |
| CN110490685A (en) * | 2019-03-27 | 2019-11-22 | 南京国科双创信息技术研究院有限公司 | A kind of Products Show method based on big data analysis |
| CN110795634A (en) * | 2019-10-31 | 2020-02-14 | 秒针信息技术有限公司 | Commodity recommendation method and device, computer equipment and readable storage medium |
| CN112766995A (en) * | 2019-10-21 | 2021-05-07 | 招商证券股份有限公司 | Article recommendation method and device, terminal device and storage medium |
| CN113781134A (en) * | 2020-07-28 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Item recommendation method and device and computer-readable storage medium |
| CN113781139A (en) * | 2020-10-19 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Item recommendation method, item recommendation device, equipment and medium |
| CN113807957A (en) * | 2020-06-11 | 2021-12-17 | Sap欧洲公司 | Determining categories of data objects based on machine learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127546A (en) * | 2016-06-20 | 2016-11-16 | 重庆房慧科技有限公司 | A kind of Method of Commodity Recommendation based on the big data in intelligence community |
| CN107705183A (en) * | 2017-09-30 | 2018-02-16 | 深圳乐信软件技术有限公司 | Recommendation method, apparatus, storage medium and the server of a kind of commodity |
| US20180075516A1 (en) * | 2016-09-14 | 2018-03-15 | Microsoft Technology Licensing, Llc | System for producing recommendations and predicting purchases of products based on usage patterns |
| CN107909433A (en) * | 2017-11-14 | 2018-04-13 | 重庆邮电大学 | A kind of Method of Commodity Recommendation based on big data mobile e-business |
| CN107944913A (en) * | 2017-11-21 | 2018-04-20 | 重庆邮电大学 | High potential user's purchase intention Forecasting Methodology based on big data user behavior analysis |
| US20180260840A1 (en) * | 2017-03-10 | 2018-09-13 | Facebook, Inc. | Selecting content for presentation to an online system user based on categories associated with content items |
| CN108629665A (en) * | 2018-05-08 | 2018-10-09 | 北京邮电大学 | A kind of individual commodity recommendation method and system |
-
2018
- 2018-11-15 CN CN201811359261.7A patent/CN109242654A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127546A (en) * | 2016-06-20 | 2016-11-16 | 重庆房慧科技有限公司 | A kind of Method of Commodity Recommendation based on the big data in intelligence community |
| US20180075516A1 (en) * | 2016-09-14 | 2018-03-15 | Microsoft Technology Licensing, Llc | System for producing recommendations and predicting purchases of products based on usage patterns |
| US20180260840A1 (en) * | 2017-03-10 | 2018-09-13 | Facebook, Inc. | Selecting content for presentation to an online system user based on categories associated with content items |
| CN107705183A (en) * | 2017-09-30 | 2018-02-16 | 深圳乐信软件技术有限公司 | Recommendation method, apparatus, storage medium and the server of a kind of commodity |
| CN107909433A (en) * | 2017-11-14 | 2018-04-13 | 重庆邮电大学 | A kind of Method of Commodity Recommendation based on big data mobile e-business |
| CN107944913A (en) * | 2017-11-21 | 2018-04-20 | 重庆邮电大学 | High potential user's purchase intention Forecasting Methodology based on big data user behavior analysis |
| CN108629665A (en) * | 2018-05-08 | 2018-10-09 | 北京邮电大学 | A kind of individual commodity recommendation method and system |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110490685A (en) * | 2019-03-27 | 2019-11-22 | 南京国科双创信息技术研究院有限公司 | A kind of Products Show method based on big data analysis |
| CN110163525A (en) * | 2019-05-29 | 2019-08-23 | 中国联合网络通信集团有限公司 | Terminal recommended method and terminal recommender system |
| CN110415048A (en) * | 2019-08-06 | 2019-11-05 | 即悟(上海)智能科技有限公司 | A kind of advertisement recommended method, equipment, system, computer equipment and storage medium |
| CN112766995A (en) * | 2019-10-21 | 2021-05-07 | 招商证券股份有限公司 | Article recommendation method and device, terminal device and storage medium |
| CN112766995B (en) * | 2019-10-21 | 2024-09-24 | 招商证券股份有限公司 | Article recommendation method, device, terminal equipment and storage medium |
| CN110795634A (en) * | 2019-10-31 | 2020-02-14 | 秒针信息技术有限公司 | Commodity recommendation method and device, computer equipment and readable storage medium |
| CN113807957A (en) * | 2020-06-11 | 2021-12-17 | Sap欧洲公司 | Determining categories of data objects based on machine learning |
| CN113781134A (en) * | 2020-07-28 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Item recommendation method and device and computer-readable storage medium |
| CN113781139A (en) * | 2020-10-19 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Item recommendation method, item recommendation device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109242654A (en) | A kind of item recommendation method and system | |
| CN106485562B (en) | Commodity information recommendation method and system based on user historical behaviors | |
| CN108665329B (en) | Commodity recommendation method based on user browsing behavior | |
| CN108665311B (en) | Electric commercial user time-varying feature similarity calculation recommendation method based on deep neural network | |
| CN104239338A (en) | Information recommendation method and information recommendation device | |
| CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
| CN103164804A (en) | Personalized method and personalized device of information push | |
| Yeo et al. | Conversion prediction from clickstream: Modeling market prediction and customer predictability | |
| CN112001754A (en) | User portrait generation method, device, equipment and computer readable medium | |
| KR20160071990A (en) | Customer data analysis and verification system | |
| CN107193831A (en) | Information recommendation method and device | |
| CN105335518A (en) | Method and device for generating user preference information | |
| JP5824602B1 (en) | Information processing apparatus, information processing method, and information processing program | |
| CN117745349A (en) | Personalized coupon pushing method and system based on user characteristics | |
| CN107180044A (en) | Recognize Internet user's sex method and system | |
| CN116127184A (en) | Product recommendation method and device, nonvolatile storage medium and electronic equipment | |
| CN105630185A (en) | Temporal dynamics in display advertising prediction | |
| US10467654B2 (en) | Forecasting customer channel choice using cross-channel loyalty | |
| CN111680213A (en) | Information recommendation method, data processing method and device | |
| JP5603678B2 (en) | Demand forecasting apparatus and demand forecasting method | |
| CN117909569A (en) | Method for accurately recommending commodities in electronic commerce based on internet data analysis | |
| JP5525501B2 (en) | Customer survival rate estimation system, customer survival rate estimation method, customer survival rate estimation processing server, and program for customer survival rate estimation processing server | |
| CN111241382A (en) | Data processing method and device, storage medium and electronic equipment | |
| CN115660203A (en) | Life cycle value prediction method and device, storage medium and electronic equipment | |
| CN112015970A (en) | Product recommendation method, related equipment 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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190118 |
|
| RJ01 | Rejection of invention patent application after publication |