CN117808560A

CN117808560A - Article recommendation method and device

Info

Publication number: CN117808560A
Application number: CN202410138026.6A
Authority: CN
Inventors: 龚志刚; 王晶晶
Original assignee: Beijing Jingdong Century Trading Co Ltd; Beijing Wodong Tianjun Information Technology Co Ltd
Current assignee: Beijing Jingdong Century Trading Co Ltd; Beijing Wodong Tianjun Information Technology Co Ltd
Priority date: 2024-01-31
Filing date: 2024-01-31
Publication date: 2024-04-02

Abstract

The invention discloses a method and a device for recommending articles, and relates to the technical field of artificial intelligence. One embodiment of the method for recommending articles comprises the following steps: determining data characteristics of user behavior data in response to receiving the user behavior data of a target user; acquiring regional information of the target user according to the data characteristics, and determining regional related articles of the target user according to the regional information and the user behavior data; recommending the area-associated article to the target user. According to the method and the device, the related articles are recommended to the user based on the user information and the area information of the area where the user is located, so that the accuracy of article recommendation can be improved, better experience is given to the user, and the platform activity of the user is improved.

Description

Article recommendation method and device

Technical Field

The invention relates to the technical field of artificial intelligence, in particular to a method and a device for recommending articles.

Background

In order to assist the user in finding the items of the cardiology among the numerous items, the e-commerce platform often recommends the items to the user. In recommending items to a user, it is common to determine similar users based on historical behavior data of the user and then recommend items preferred by the similar users to the user, or to determine items preferred by the user based on historical behavior data of the user and then recommend other items similar to the items preferred by the user to the user.

In carrying out the present invention, the inventors have found that at least the following problems exist in the prior art:

similar users and preference commodities determined according to the historical behavior data are inaccurate, so that commodities recommended to the users are inaccurate, and user experience is reduced.

Disclosure of Invention

In view of the above, the embodiment of the invention provides a method and a device for recommending articles, which can improve the accuracy of article recommendation and the use experience of users.

To achieve the above object, according to a first aspect of an embodiment of the present invention, there is provided a method for recommending an item, including:

determining data characteristics of user behavior data in response to receiving the user behavior data of a target user;

acquiring regional information of the target user according to the data characteristics, and determining regional related articles of the target user according to the regional information and the user behavior data;

recommending the area-associated article to the target user.

Optionally, the data features include: the data amount of the user behavior data; obtaining the regional information of the target user according to the data characteristics, and determining the regional associated object of the target user according to the regional information and the user behavior data, wherein the method comprises the following steps:

Comparing the data quantity of the user behavior data with a preset data quantity threshold;

under the condition that the data volume of the user behavior data is larger than or equal to the data volume threshold, constructing a similar user set of the target user according to the user behavior data;

acquiring regional user information of the target user according to the geographic position of the target user, and constructing regional user portraits of the target user according to the regional user information;

and according to the regional user portraits, determining target similar users similar to the target users from the similar user sets, and taking the articles associated with the target similar users as regional associated articles of the target users.

Optionally, obtaining the regional user information of the target user according to the geographic location of the target user includes:

determining adjacent users of the target user in a preset distance range according to the geographic position of the target user;

and acquiring user behavior data of the adjacent user, performing grid division on the user behavior data of the adjacent user to obtain a plurality of pieces of grid user behavior data, and taking the plurality of pieces of grid user behavior data as regional user information of the target user.

Optionally, constructing the regional user portrait of the target user according to the regional user information, including:

and respectively carrying out cluster analysis on the plurality of gridding user behavior data to generate a plurality of gridding user portraits, and taking the plurality of gridding user portraits as the regional user portraits of the target user.

Optionally, meshing the user behavior data of the adjacent user to obtain a plurality of meshed user behavior data, including:

determining adjacent geographic positions corresponding to the user behavior data of the adjacent users;

and dividing the user behavior data of the adjacent users by using grids with different grid levels according to the adjacent geographic positions to obtain a plurality of grid user behavior data.

Optionally, in a case where the data amount of the user behavior data is smaller than the data amount threshold, the method further includes:

acquiring user basic data of the target user, and constructing an article feature vector of the target user according to the user behavior data and the user basic data;

acquiring regional store information of the target user according to the geographic position of the target user; constructing a regional article portrait of the geographic position according to the regional store information;

And determining similar articles similar to the article feature vector according to the regional article portrait, and taking the similar articles as regional related articles of the target user.

Optionally, acquiring regional store information of the target user according to the geographic location of the target user includes:

determining adjacent shops of the target user in a preset distance range according to the geographic position of the target user;

and acquiring store data of the adjacent stores, meshing the store data to obtain a plurality of meshed store data, and taking the meshed store data as regional store information of the target user.

Optionally, constructing a regional item portrait of the geographic location according to the regional store information, including:

and performing cluster analysis on the plurality of gridding store data respectively to generate a plurality of gridding article portraits, wherein the plurality of gridding article portraits are taken as regional article portraits of the target user.

According to a second aspect of an embodiment of the present invention, there is provided an apparatus for recommending an item, including:

a first determining module, configured to determine a data feature of user behavior data in response to receiving the user behavior data of a target user;

The second determining module is used for acquiring regional information of the target user according to the data characteristics and determining regional related articles of the target user according to the regional information and the user behavior data;

and the recommending module is used for recommending the area-associated articles to the target user.

Optionally, the apparatus further comprises:

the feature construction module is used for acquiring user basic data of the target user and constructing article feature vectors of the target user according to the user behavior data and the user basic data;

the portrait construction module is used for acquiring regional store information of the target user according to the geographic position of the target user; constructing a regional article portrait of the geographic position according to the regional store information;

and the third determining module is used for determining similar articles similar to the article feature vector according to the regional article portrait, and taking the similar articles as regional associated articles of the target user.

According to a third aspect of an embodiment of the present invention, there is provided an electronic apparatus including:

one or more processors;

storage means for storing one or more programs,

the one or more processors implement the method of any of the embodiments described above when the one or more programs are executed by the one or more processors.

According to a fourth aspect of embodiments of the present invention, there is provided a computer readable medium having stored thereon a computer program which, when executed by a processor, implements a method according to any of the embodiments described above.

One embodiment of the above invention has the following advantages or benefits: based on the user information and the area information of the area where the user is located, recommending related articles to the user, so that the accuracy of article recommendation can be improved, better platform use experience is given to the user, and the platform activity of the user is improved; determining adjacent users according to geographic positions of the users, performing grid division on user behavior data of the adjacent users to obtain a plurality of pieces of grid user behavior data, adding grid information into the user behavior data to obtain user behavior data corresponding to different grids in the area, and facilitating obtaining more accurate regional user portraits; generating a plurality of corresponding gridding user portraits according to the plurality of gridding user behavior data, adding gridding information into the user portraits, and dividing a plurality of user portraits in a region according to grids so as to be convenient for searching other users more similar to the user; the grids of different grid levels are used for dividing the user behavior data of the adjacent users, so that the flexibility of data division can be improved, a plurality of grid user behavior data can be obtained, and the query efficiency and accuracy of the similar users are improved; determining adjacent shops according to geographic positions of users, meshing shop data of the adjacent shops to obtain a plurality of meshed shop data, adding meshed information into the shop data to obtain shop data corresponding to different grids in a region, and facilitating obtaining more accurate regional shop images; the corresponding multiple gridded article images are generated according to the multiple gridded store data, gridded information can be added into the article images, and the multiple article images in the area are divided according to the grids, so that articles with higher correlation with users can be conveniently searched.

Further effects of the above-described non-conventional alternatives are described below in connection with the embodiments.

Drawings

The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:

FIG. 1 is a schematic diagram of the main flow of a method of item recommendation according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an overall flow of item recommendations according to one referenceable embodiment of the present invention;

FIG. 3 is a schematic diagram of the main flow of data processing according to one referenceable embodiment of the invention;

FIG. 4 is a schematic diagram of the main flow of recommending items based on a regional user representation in accordance with one refereed embodiment of the present invention;

FIG. 5 is a schematic diagram of a main flow of recommending items based on a representation of a regional item according to one refereed embodiment of the present invention;

FIG. 6 is a schematic diagram of the main flow of a method of item recommendation according to one referenceable embodiment of the invention;

FIG. 7 is a schematic diagram of the main modules of an item recommendation device according to an embodiment of the present invention;

FIG. 8 is an exemplary system architecture diagram in which embodiments of the present invention may be applied;

fig. 9 is a schematic diagram of a computer system suitable for use in implementing an embodiment of the invention.

Detailed Description

Exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, in which various details of the embodiments of the present invention are included to facilitate understanding, and are to be considered merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

In the technical scheme of the invention, the related processes of collecting, using, storing, sharing, transferring and the like of the personal information of the user accord with the regulations of related laws and regulations, the user needs to be informed and obtain the consent or the authorization of the user, and when the personal information of the user is applicable, the technical processes of de-identification and/or anonymization and/or encryption are performed on the personal information of the user.

In order to assist the user in finding the items of the cardiology among the numerous items, the e-commerce platform often recommends the items to the user. When recommending commodities to a user, a similar user is usually determined according to historical behavior data of the user, and then commodities preferred by the similar user are recommended to the user; or determining the goods preferred by the user according to the historical behavior data of the user, and then recommending other goods similar to the goods preferred by the user to the user; or clustering the historical behavior data of the user, and recommending corresponding commodities to the user according to a clustering result.

Similar users and preference commodities determined according to the historical behavior data are inaccurate, so that commodities recommended to the users are inaccurate, and user experience is reduced. Based on single types of other commodities recommended by the preference commodity, the clustering method has higher requirements on an initial training sample, a large amount of manpower and material resources are required to be consumed for data processing, and the accuracy of recommending the commodity is lower under the condition that the training sample has larger difference from the requirements.

In view of this, according to a first aspect of embodiments of the present invention, there is provided a method of item recommendation.

FIG. 1 is a schematic diagram of the main flow of a method of item recommendation according to an embodiment of the present invention. As shown in fig. 1, the method for recommending items according to the embodiment of the present invention mainly includes the following steps S101 to S103.

Step S101, in response to receiving user behavior data of a target user, determining data characteristics of the user behavior data.

The user behavior data includes: explicit behavior data and implicit behavior data. Specifically, explicit behavior data refers to behavior in which a user explicitly indicates a preference for an item, and includes: collecting, purchasing, ordering and the like of the articles; implicit behavior data refers to behavior that does not directly represent a preference of a user, and includes: browsing duration of browsing the items, forwarding, sharing, comparing the items, and so on.

The data characteristics of the user behavior data include: data volume, data category, time range, etc. of the user behavior data. For example, acquiring user behavior data of a target user includes: in the past 5 days, 10 articles are collected and ordered for 5 times, and the 3-hour platform page is browsed in an accumulated way, the time range of the user behavior data is 5 days, and the data amount of the user behavior data is 5 when the user behavior data is divided by days, wherein the categories of the user behavior data comprise: explicit behavior data such as collection and ordering, and implicit behavior data such as browsing.

Step S102, obtaining regional information of the target user according to the data characteristics, and determining regional related articles of the target user according to the regional information and the user behavior data.

After determining the data characteristics of the user behavior data, matching the data characteristics with preset information acquisition conditions for determining the regional information to be acquired. Specifically, the localized information relates to a geographic location where the user is located, and the localized information includes: other user information (i.e., regional user information) surrounding the geographic location where the user is located, store information (i.e., regional store information) surrounding the geographic location where the user is located, and so forth.

In an exemplary case where the data feature is a data category of the user behavior data, the number of data categories is compared with a preset category number threshold, and in a case where the number of data categories is equal to or greater than the category number threshold, regional user information is acquired, and according to the regional user information, similar users of the target user are determined from other users that are geographically close to the target user, and then commodities associated with the similar users are taken as regional associated items of the target user.

Further, for example, when the number of data categories is smaller than the threshold number of categories, regional store information is acquired, and based on the regional store information, items which are collected (placed and shared) more than a preset threshold number of times are determined from stores which are geographically close to the target user, and the items are collected (placed and shared) more frequently, and the items are associated with the region of the target user as items of interest to other users around the geographic location where the target user is located.

And step S103, recommending the area-associated article to the target user.

After determining the area-related item, the area-related item is recommended to the target user. For example, when a user browses a page, relevant information of area-related items is put into the user in a side bar of the page. For another example, an in-station letter, a short message, and a mail are sent to the user, so that the user can know the area-related articles.

According to one referenceable embodiment of the invention, the data features include: data amount of user behavior data. And when the regional information of the target user is acquired according to the data characteristics and the regional related articles of the target user are determined according to the regional information and the user behavior data, comparing the data quantity of the user behavior data with a preset data quantity threshold value. And under the condition that the data volume of the user behavior data is larger than or equal to the data volume threshold value, the obtained data volume of the user behavior data is larger, and a similar user set of the target user is constructed according to the user behavior data.

Specifically, after receiving user behavior data of a target user, constructing a user behavior matrix of the target user, including: explicit behavior matrices and implicit behavior matrices. For example, vector encoding (i.e., an Embedding operation) is performed on explicit behavior data and implicit behavior data of a target user, each explicit behavior data is converted into a corresponding explicit behavior vector, each implicit behavior data is converted into a corresponding implicit behavior vector, a plurality of explicit behavior vectors are spliced into an explicit behavior matrix, and a plurality of implicit behavior vectors are spliced into an implicit behavior matrix; for another example, the explicit behavior data and the implicit behavior data are both of enumeration type, and then the explicit behavior data and the implicit behavior data of the target user are separately subjected to One-Hot encoding (i.e., one-Hot encoding), the collection behavior is encoded as "0001", the purchasing behavior is encoded as "0010", the next row is encoded as "0100", the forwarding behavior is encoded as "1001", the sharing behavior is encoded as "1010", and the comparison behavior is encoded as "1100", wherein the highest character (i.e., leftmost character) is "0" to indicate that this is the encoding result of One explicit behavior, and the highest character is "1" to indicate that this is the encoding result of One implicit behavior. In the case where the behavior data is of a numerical type, the behavior data is subjected to normalization and discretization processing, for example, the behavior data of a plurality of numerical types is divided by different divisors so that the behavior data of the plurality of numerical types are of the same order of magnitude, the unit-different effects of the behavior data of the plurality of numerical types are eliminated, and for example, a plurality of continuous numerical values are sampled and discretized into a plurality of discrete numerical values.

After obtaining the user behavior matrix of the target user, comparing the user behavior matrix of the target user with the user behavior matrices of other users, and calculating the similarity of the two user behavior matrices, wherein the user compared with the target user is taken as the similar user of the target user under the condition that the similarity is larger than a preset similarity threshold value, and the determined multiple similar users form a similar user set of the target user. Other users that compare to the target user include: other users who are in the same area as the target user (e.g., the same province, the same county, the same street, etc.), other users who have a friend relationship with the target user, other users who have purchased the same merchandise as the target user, and so forth.

According to user behavior data of the target user, a similar user set of the target user is constructed, the similar user similar to the user preference can be determined, a data basis is provided for further identifying the similar user, the user behavior data is subjected to data processing by means of vector coding, independent heat coding, normalization, discretization and the like, the flexibility of the data processing can be improved, personalized data processing is performed aiming at the data type and the data format of the user behavior data, the data processing efficiency and the data processing effect are improved, and the accuracy of the similar user set is improved.

After a similar user set of a target user is constructed according to user behavior data, regional user information of the target user is acquired according to the geographic position of the target user, and a regional user portrait of the target user is constructed according to the regional user information.

Specifically, the execution subject of the embodiment of the present invention obtains the geographic location of the target user, where the geographic location includes: the user fills in the receiving address on the e-commerce platform, the geographical location where the user was in the last time, the geographical location where the user has been staying, and so on. For example, the target user has purchased a store service, and the geographic location of the corresponding store is taken as the geographic location of the target user.

After determining the geographic location, acquiring user behavior data and user basic data of other users nearby the geographic location, and taking the acquired user behavior data and user basic data as regional user information. . By way of example, a circular area is generated by taking the geographic position as a circle center according to a preset distance radius, and user behavior data and user basic data of other users in the circular area are obtained. Further exemplary, the city, county, or country to which the geographic region belongs is determined, and then user behavior data and user base data of other users within the city, county, or country are obtained.

After the regional user information is acquired, generating a regional user portrait according to the regional user information, wherein the regional user portrait represents portraits of other users in a region corresponding to the geographic position of the target user, namely, adding regional features into the user portrait, and the regional user portrait comprises: the user platform registers information such as duration, preferred items, preferred marketing campaigns, consumption habits, consumption ratings, etc. The regional user portrayal may be one or more, for example, the other users in the region corresponding to the geographic location are divided into a plurality of user portrayals based on the regional user information.

Acquiring regional user information according to the geographic position of the target user, constructing regional user portraits according to the regional user information, and determining basic information, consumption habits, consumption levels, preferred articles and activities and the like of other users (namely other users in the vicinity of the target user) in the region where the target user is located; because users in the same area often have the same interest and consumption level and mutually influence each other in consumption concept, the accuracy of the user image can be improved by introducing the area user information when the user image is generated, other users more similar to the target user can be conveniently found, and more suitable articles are recommended to the users.

According to a referenceable embodiment of the present invention, when obtaining regional user information of a target user according to the geographic location of the target user, determining a neighboring user of the target user within a preset distance range according to the geographic location of the target user. For example, if the preset distance range is 1 km, a circular area is formed by taking the geographic position of the target user as the center and taking 1 km as the radius, and other users in the circular area are taken as the adjacent users of the target user. For another example, if the preset distance range is 3×3 km, a rectangular area is formed with the geographic location of the target user as the center and 3 km as the side length, and other users in the rectangular area are used as the neighboring users of the target user.

And then acquiring user behavior data of the adjacent user, and performing grid division on the user behavior data of the adjacent user to obtain a plurality of grid user behavior data. In an exemplary case where the area corresponding to the geographical location of the target user is a square area, a plurality of squares of unequal side lengths are reformed at the same center within the square area, thereby dividing the square area into one small square area and a plurality of square annular areas. And taking the user behavior data in the small square area and the plurality of square annular areas as grid user behavior data respectively.

Further exemplary, in the case where the area corresponding to the geographical position of the target user is a circular area, a plurality of circles with different radii are formed in the circular area again with the same center of circle, so that the circular area is divided into one small circular area and a plurality of circular areas, for example, in the case where the radius of the circular area is 3 km, one circle is formed every 1 km, then the circular area with the center of circle to within 1 km is taken as a first meshing result, the circular area with the center of circle to within 2 km is taken as a second meshing result, and the circular area with the center of circle to within 3 km is taken as a third meshing result. And taking the user behavior data in the small circular area and the plurality of circular areas as one gridding user behavior data respectively.

And taking the plurality of pieces of gridding user behavior data as regional user information of the target user. And then generating a corresponding user behavior portrait according to each piece of meshed user behavior data, wherein the user behavior portrait is used for searching similar users of the target user.

The method comprises the steps of meshing the area corresponding to the geographic position, dividing the area into a plurality of grid areas, further dividing the area user information to obtain a plurality of meshed user behavior data, generating area user portraits based on a smaller area range, increasing the number of the area user portraits, providing more various data bases for determining similar users, improving the accuracy of the area user portraits, and finding other users more similar to the target user.

According to another embodiment of the present invention, when constructing the regional user representation of the target user based on the regional user information, cluster analysis is performed on the plurality of pieces of meshed user behavior data to generate a plurality of meshed user representations, and the plurality of meshed user representations are used as the regional user representation of the target user. Each of the meshed user behavior data includes: user behavior data for all other users within a grid area near the geographic location of the target user. Illustratively, K-means clustering algorithm analysis is performed on each piece of meshed user behavior data, user portrayal construction is performed on a meshed area near the geographic position of the target user, and an area user portrayal of the meshed area is generated. Each grid region corresponds to a grid-like user representation. All the grid user portraits are taken as regional user portraits. The regional user representation includes: behavioral characteristics of similar populations, such as consumption habits, consumption levels, items and activities of interest, and the like.

The method comprises the steps of gridding the area near the geographic position of the target user to obtain a plurality of grid areas, generating corresponding gridding user portraits, improving the accuracy of the user portraits, subdividing user characteristics of different grid areas in the areas, enhancing the influence of grid information on the user portraits, combining the grid information with the user portraits, facilitating finding other users more similar to the target user, and improving the accuracy of article recommendation.

According to another embodiment of the present invention, when user behavior data of a neighboring user is grid-divided to obtain a plurality of grid-formed user behavior data, a neighboring geographic location corresponding to the user behavior data of the neighboring user is determined. Each user behavior data corresponds to a geographic location where the user behavior data was generated, and the adjacent geographic location represents the geographic location to which the adjacent user corresponds. And dividing the user behavior data of the adjacent users by using grids of different grid levels according to the adjacent geographic positions to obtain a plurality of grid user behavior data.

A plurality of grid levels are preset, wherein the grid levels are used for representing the size, the dimension and the like of the grid, and for example, the grid levels comprise: grid level A1, grid level A2 and grid level A3, wherein grid level A1 corresponds to grid size of 1.5X1.5 km, grid level A2 corresponds to grid size of 2.1X12.1 km, and grid level A3 corresponds to grid size of 3X 3 km. And dividing the areas corresponding to the geographic positions of the target users by using grids of different grid levels respectively to obtain a plurality of pieces of grid user behavior data. For example, firstly, meshing a region within 4×4 km with the geographic position of a target user as the center by using a grid with a grid size of 1×1 km corresponding to a grid level B1 to obtain 16 grid regions, and taking user behavior data in each grid region as meshing user behavior data to obtain 16 meshing user behavior data; and then, using a grid with the grid size of 2 multiplied by 2 kilometers corresponding to the grid level B2 to grid-divide the area within 4 multiplied by 4 kilometers by taking the geographic position of the target user as the center to obtain 4 grid areas, and taking the user behavior data in each grid area as grid user behavior data to obtain 4 grid user behavior data.

According to different grid levels, different grid operation is performed on the region, so that the flexibility of data division can be improved, and as different grid division modes correspond to different grid population densities, grid consumption levels, grid consumption habits and other data, different grid operation can introduce different grid information into user behavior data to obtain a plurality of grid user behavior data, and the query efficiency and accuracy of similar users are improved.

After obtaining the regional user portrayal and the set of similar users, determining a target similar user similar to the target user from the set of similar users according to the regional user portrayal, and recommending the object associated with the target similar user to the target user.

Specifically, similarity calculation is performed on the users included in the regional user portraits and the similar user sets, for example, the similarity is calculated using a cosine similarity, a euclidean distance, or the like. And sorting the users included in the similar user set according to the sequence of the similarity with the regional user figures from large to small, selecting the user ranked at the front as the similar user of the target user according to the number of the preset similar users, or selecting the user with the similarity larger than or equal to the similarity threshold according to the preset similarity threshold as the similar user of the target user.

Illustratively, the regional user representation includes: the user prefers data such as articles, prefers activities, average monthly expenses and the like, data corresponding to users in a similar user set are obtained, the similarity of the data corresponding to the user portrait of the region and the data corresponding to each user included in the similar user set is calculated, and the similarity of a user C1 included in the similar user set is 0.9, the similarity of a user C2 is 0.88, the similarity of a user C3 is 0.85, the similarity of a user C4 is 0.75, and the similarity of a user C5 is 0.6; in the case that the number of the preset similar users is 2, the user C1 and the user C2 are taken as similar users of the target user; in the case where the similarity threshold set in advance is 0.85, the user C1, the user C2, and the user C3 are taken as similar users of the target user.

After determining the similar users of the target users, items associated with the similar users, such as items that the similar users have purchased, items that were recommended to the similar users, items that the similar users have collected or purchased additionally, and so forth, are obtained. Recommending the articles associated with the similar users to the target users.

Because the similar user is similar to the target user in terms of user behavior data and similar to the regional user portraits formed by other users in the vicinity of the target user, the similar user is more similar to the target user in terms of item preferences, consumption habits, consumption levels, and the like. The regional information and the grid information are introduced in the process of searching similar users, users more similar to the target users can be found in the similar users, more interested and more suitable articles are recommended to the target users, the accuracy of article recommendation is improved, the user experience and the platform activity are improved, more purchasing behavior data and lower order data are generated by the users, and the browsing amount, purchasing amount and ordering amount of an electronic commerce platform are improved.

According to a referenceable embodiment of the present invention, in a case where the data size of the user behavior data is smaller than the data size threshold, it is explained that the data size of the acquired user behavior data is smaller, the method further includes: and acquiring user basic data of the target user, and constructing object feature vectors of the target user according to the user behavior data and the user basic data. For example, in the case where the acquired user behavior data is small, it is difficult to accurately determine the preference items, consumption habits, and consumption levels of the user based on the analysis based on the user behavior data alone, and it is difficult to accurately find similar users of the target user, and therefore, it is also necessary to acquire user base data of the target user. Illustratively, from user base data and user behavior data of the target user, determining articles preferred by the target user, collected articles, purchased articles, ordered articles and the like, acquiring article information of the articles, such as article types, prices, styles, sizes, packages and the like, vector coding or independent heat coding the article information, and constructing article feature vectors of the target user.

And acquiring regional store information of the target user according to the geographic position of the target user. Specifically, store information in the vicinity of the geographic position is acquired, and the acquired store information is used as regional store information. Wherein the store information includes: merchant category, store rating, number of store fans, store sales, store acceptance rate, store complaint rate, and the like. By way of example, a circular area is generated by taking the geographic position as a circle center according to a preset distance radius, and store information in the circular area is acquired. Further exemplary, a city, county, or country to which the geographic region belongs is determined, and store information within the city, county, or country is then obtained.

And constructing the regional article portrait of the geographic position according to the acquired regional store information. Specifically, the regional store information further includes: item information such as item category, item price, item sales, item exchange rate, etc. sold by the store. Illustratively, feature vectors of the items are constructed according to the regional store information, the regional store information is subjected to vector coding or independent heat coding to generate item characterization vectors, and regional item representations are generated according to the item characterization vectors. Further exemplary, the article information corresponding to the same article sold by different stores is aggregated to generate a regional article representation for each article or a regional article representation for each category of article. The regional item representation is used to describe item information such as item category, average item price, item efficiency, item rate of change, etc. for an item or a class of items.

After the regional item representation is generated, similar items similar to the item feature vectors are determined according to the regional item representation, the similar items are taken as regional associated items of the target user, and the regional associated items are recommended to the target user. Specifically, similarity calculation is performed on the regional article representation and the article feature vector, and for example, the similarity is calculated by using a cosine similarity method, a euclidean distance method, or the like. And sorting the articles corresponding to the article feature vectors according to the sequence of the similarity of the regional article portrait from large to small, selecting the top-ranked article as the preference article of the target user according to the preset preference article quantity, or selecting the article with the similarity greater than or equal to the similarity threshold according to the preset similarity threshold as the preference article of the target user. And then recommending the preference item and the similar item to the target user by taking the item with the same kind and the same purpose as the preference item as the similar item.

Illustratively, the similarity of the regional item representation to the item feature vector D1 of the target user is 0.89, the similarity to the item feature vector D2 is 0.86, and the similarity to the item feature vector D3 is 0.81; and under the condition that the preset similarity threshold value is 0.85, taking the articles corresponding to the article characteristic vector D2 and the article characteristic vector D3 as the preference articles of the target user, searching for similar articles which are used for the articles with the same type, similar price and the same purpose as the preference articles, and recommending the preference articles and the similar articles to the target user.

Based on a recommendation algorithm based on content, regional and grid information is introduced to generate regional article portraits, articles which are more interesting to a user and are more suitable for the user can be found, the accuracy of article recommendation is improved, the user experience and platform activity are improved, and the browsing amount, purchasing amount and ordering amount of the platform are improved.

According to another exemplary embodiment of the present invention, in acquiring regional store information of a target user according to a geographic location of the target user, the method includes: and determining the adjacent shops of the target user in a preset distance range according to the geographic position of the target user. For example, if the preset distance range is 1 km, a circular area is formed by taking the geographic position of the target user as the center and taking 1 km as the radius, and the shops in the circular area are taken as the neighboring shops of the target user. For another example, if the preset distance range is 3×3 km, a rectangular area is formed with the geographic location of the target user as the center and 3 km as the side length, and the shops in the rectangular area are used as the neighboring shops of the target user.

Then, store data of adjacent stores are acquired, the store data of the adjacent stores are subjected to grid division, a plurality of pieces of grid store data are obtained, and the plurality of pieces of grid store data are used as regional store information of target users. Illustratively, a plurality of grid levels are preset, the grid levels being used to represent the size, dimension, etc. of the grid, for example, the grid levels include: grid level A1, grid level E2 and grid level E3, wherein grid level E1 corresponds to grid size of 1×1 km, grid level E2 corresponds to grid size of 2×2 km, and grid level E3 corresponds to grid size of 3×3 km. And dividing the areas corresponding to the geographic positions of the target users by using grids with different grid levels respectively to obtain a plurality of gridded store data. Firstly, meshing a region in 3×3 kilometers with the geographic position of a target user as the center by using a grid with a grid size of 1×1 kilometer corresponding to a grid level E1 to obtain 9 grid regions, and taking store data in each grid region as meshed store data to obtain 9 meshed store data; and then, using a grid with the grid size of 2×2 kilometers corresponding to the grid level E2 to grid-divide the area within 3×3 kilometers with the geographic position of the target user as the center to obtain 4 grid areas, and using store data in each grid area as one grid store data to obtain 4 grid store data.

Further exemplary, in the case where the area corresponding to the geographical position of the target user is a circular area, a plurality of circles with different radii are formed in the circular area again with the same center of circle, so that the circular area is divided into one small circular area and a plurality of circular areas, for example, in the case where the radius of the circular area is 3 km, one circle is formed every 1 km, then the circular area with the center of circle to within 1 km is taken as a first meshing result, the circular area with the center of circle to within 2 km is taken as a second meshing result, and the circular area with the center of circle to within 3 km is taken as a third meshing result. Store data in the small circular area and the plurality of circular areas are used as one piece of gridding store data.

The region corresponding to the geographic position is gridded, the region is divided into a plurality of grid regions, regional store information can be further divided, and a plurality of gridded store data are obtained, so that regional object portraits can be conveniently generated based on a smaller regional range, the number of regional object portraits is increased, a more various data basis is provided for determining the preference objects of the target user, the accuracy of the regional object portraits is improved, and the preference objects of the target user can be conveniently and efficiently found.

According to different grid levels, different gridding operations are performed on the areas, so that the flexibility of data division can be improved, and as different grid division modes correspond to different grid population densities, grid consumption levels, grid consumption habits and other data, store data can be correspondingly different, different gridding operations can introduce different gridding information into the store data, a plurality of gridding store data are obtained, and preference objects and similar objects can be searched more accurately.

According to still another embodiment of the present invention, when creating a regional item representation of a geographic location based on regional store information, a plurality of meshed store data are each subjected to a cluster analysis to create a plurality of meshed item representations, and the plurality of meshed item representations are used as regional item representations of target users. Each gridded store data includes: within a grid area near the geographic location of the target user, business data, item information, etc. for all stores. For example, K-means clustering algorithm analysis is performed for each piece of gridded store data, and a product portrait included in a store is constructed for a grid region near the geographic location of a target user, so as to generate a regional product portrait for the grid region. Each grid region corresponds to a representation of a grid-like article. All the grid-like article representations are taken as regional article representations. The regional article representation includes: other users within the area prefer features, such as information about the category, price, use, packaging, style, size, etc. of the item.

The method comprises the steps of meshing an area near the geographic position of a target user to obtain a plurality of grid areas, generating corresponding meshed article portraits, improving the accuracy of the article portraits, finely dividing the article characteristics of different grid areas in the areas, enhancing the influence of grid information on the article portraits, combining the grid information with the article portraits, facilitating finding out preferred articles of the target user and other articles similar to the preferred articles, improving the accuracy of article recommendation, improving user experience, and improving platform browsing amount, purchasing amount and ordering amount.

FIG. 2 is a schematic diagram of an overall flow of item recommendations according to one referenceable embodiment of the present invention. As shown in fig. 2, an execution subject of the embodiment of the present invention obtains user behavior data of a target user, constructs feature vectors of the user behavior data under the condition that the obtained user behavior data is more, and determines a similar user set of the target user according to the feature vectors of the user behavior data; meanwhile, obtaining regional user data and constructing regional user portraits; a similar user for the target user is then determined based on the set of similar users and the regional user profile. Under the condition that the acquired user behavior data are fewer, acquiring user basic data of a target user, and constructing article feature vectors of the target user according to the user behavior data and the user basic data; meanwhile, obtaining regional store data and constructing regional object portraits; a preferred item of the target user is then determined based on the item feature vector and the regional item representation. In the data recall link, according to the similar users of the target users, the preference articles of the similar users and the corresponding similar articles are obtained, and the preference articles of the target users and the corresponding similar articles are obtained. And then sorting the preferred items and similar items, for example, sorting the preferred items according to the collection time, the purchase frequency, the ordering quantity, the ordering frequency and other data of the user, sorting the similar items according to the similarity with the preferred items, determining a target item according to the sorting result, and recommending the target item to the target user.

Fig. 3 is a schematic diagram of the main flow of data processing according to one referenceable embodiment of the invention. Illustratively, as shown in fig. 3, the executing body of the embodiment of the present invention acquires user behavior data, user basic data, store data, article data, and gridding data, performs data preprocessing on the data, and includes: data cleaning, data standardization, data formatting, data rejection, data filling and other operations; the text class feature data obtained from the user behavior data and the user basic data comprises the names and the like of operation activities in which the user participates, and the numerical class feature data comprises: frequency of user participation in operation activities, etc.; the obtaining text class feature data from store data includes: store classification, store rating, etc., the numerical class feature data includes: vermicelli count, sales, etc.; the obtaining text type feature data from the article data comprises: commodity classification, commodity use, etc., the numerical class feature data includes: price, collection number, purchase number, etc.; the obtaining text class feature data from the gridding data comprises: grid level, grid size, etc., the numeric class feature data includes: grid population density, grid consumption level, and the like. And performing single-heat coding or vector coding on the text type characteristic data to generate a characteristic vector, and performing normalization processing or discretization processing on the numerical value type characteristic data. Feature screening is carried out on the data after feature processing, and features with low correlation, irrelevant and redundancy with article recommendation are removed; and then determining the object items of interest to the object user by using the filtered characteristics, and recommending the object items to the object user.

FIG. 4 is a schematic diagram of the main flow of recommending items based on a regional user representation in accordance with one refereed embodiment of the present invention. Illustratively, as shown in fig. 4, the execution subject of the embodiment of the present invention acquires explicit user behavior data and implicit user behavior data, constructs feature vectors of the user behavior data, and constructs a similar user set; then, based on a preset multi-stage grid, building regional user information, specifically, the multi-stage grid comprises: square primary grid with side length of 1.5 km, square secondary grid with side length of 2.1 km, square tertiary grid with side length of 3 km; constructing an area user portrait according to the area user information; and calculating the similarity between the similar user set and the regional user portrait, determining similar users of the target user, and recommending the articles associated with the similar users to the target user.

FIG. 5 is a schematic diagram of the main flow of recommending items based on regional item representations according to one referenceable embodiment of the present invention. Illustratively, as shown in fig. 5, the execution subject of the embodiment of the present invention acquires user behavior data and user base data, constructs feature vectors of the user behavior data, and constructs article feature vectors of the target user; then, based on a preset multi-level grid, regional store information is constructed, specifically, the multi-level grid includes: square primary grid with side length of 1.6 km, square secondary grid with side length of 2.3 km, square tertiary grid with side length of 3.2 km; constructing a regional article representation according to regional store information; and calculating the similarity between the item feature vector and the regional item portrait, determining the preference item of the target user, and recommending the preference item of the target user and similar items similar to the preference item to the target user.

FIG. 6 is a schematic diagram of the main flow of a method of item recommendation according to one referenceable embodiment of the invention. As shown in fig. 6, the method for recommending the item may include:

step S601, receiving user behavior data of a target user;

step S602, judging whether the data amount of the user behavior data is larger than a quantity threshold, if yes, jumping to step S603, otherwise jumping to step S606;

step S603, constructing a similar user set of the target user according to the user behavior data;

step S604, obtaining regional user information of the target user according to the geographic position of the target user, and constructing regional user portraits of the target user according to the regional user information;

step S605, determining target similar users similar to the target users from the similar user set according to the regional user portraits, and recommending articles associated with the target similar users to the target users;

step S606, obtaining user basic data of a target user, and constructing article feature vectors of the target user according to the user behavior data and the user basic data;

step S607, obtaining regional store information of the target user according to the geographic position of the target user; constructing a regional article portrait of the geographic position according to regional store information;

Step S608, determining similar articles similar to the article feature vector according to the regional article portrait, and recommending the similar articles to the target user.

The specific implementation of the method for recommending an article according to the present invention is described in detail in the above method for recommending an article, and thus the description thereof will not be repeated here.

According to a second aspect of embodiments of the present invention, there is provided an apparatus for item recommendation.

FIG. 7 is a schematic diagram of main modules of an apparatus for recommending items according to an embodiment of the present invention, and as shown in FIG. 7, the apparatus 700 for recommending items mainly includes:

a first determining module 701, configured to determine, in response to receiving user behavior data of a target user, a data feature of the user behavior data;

a second determining module 702, configured to obtain, according to the data feature, localized information of the target user, and determine, according to the localized information and the user behavior data, a local association object of the target user;

and the recommending module 703 is configured to recommend the area-related item to the target user.

According to a referenceable embodiment of the invention, the data amount of the user behavior data; obtaining the regional information of the target user according to the data characteristics, and determining the regional associated object of the target user according to the regional information and the user behavior data, wherein the method comprises the following steps:

According to a referenceable embodiment of the present invention, obtaining regional user information of the target user according to the geographic location of the target user includes:

According to another embodiment of the present invention, constructing a regional user representation of a target user based on the regional user information includes:

According to another embodiment of the present invention, the meshing of the user behavior data of the neighboring users to obtain a plurality of meshed user behavior data includes:

According to yet another exemplary embodiment of the present invention, the apparatus 700 for recommending items further includes:

According to still another embodiment of the present invention, the obtaining regional store information of the target user according to the geographic location of the target user includes:

According to a referenceable embodiment of the present invention, constructing a representation of the geographically located regional item from the regional store information comprises:

In the embodiment of the present invention, the specific implementation of the device for recommending an item is described in detail in the method for recommending an item, so that the description is not repeated here.

According to the technical scheme provided by the embodiment of the invention, the related articles are recommended to the user based on the user information and the area information of the area where the user is located, so that the accuracy of article recommendation can be improved, better platform use experience is given to the user, and the platform activity of the user is improved; determining adjacent users according to geographic positions of the users, performing grid division on user behavior data of the adjacent users to obtain a plurality of pieces of grid user behavior data, adding grid information into the user behavior data to obtain user behavior data corresponding to different grids in the area, and facilitating obtaining more accurate regional user portraits; generating a plurality of corresponding gridding user portraits according to the plurality of gridding user behavior data, adding gridding information into the user portraits, and dividing a plurality of user portraits in a region according to grids so as to be convenient for searching other users more similar to the user; the grids of different grid levels are used for dividing the user behavior data of the adjacent users, so that the flexibility of data division can be improved, a plurality of grid user behavior data can be obtained, and the query efficiency and accuracy of the similar users are improved; determining adjacent shops according to geographic positions of users, meshing shop data of the adjacent shops to obtain a plurality of meshed shop data, adding meshed information into the shop data to obtain shop data corresponding to different grids in a region, and facilitating obtaining more accurate regional shop images; the corresponding multiple gridded article images are generated according to the multiple gridded store data, gridded information can be added into the article images, and the multiple article images in the area are divided according to the grids, so that articles with higher correlation with users can be conveniently searched.

According to a third aspect of an embodiment of the present invention, there is provided an electronic apparatus including: one or more processors; and the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors are enabled to realize the method provided by the first aspect of the embodiment of the invention.

According to a fourth aspect of embodiments of the present invention, there is provided a computer readable medium having stored thereon a computer program which when executed by a processor implements the method provided by the first aspect of embodiments of the present invention.

FIG. 8 illustrates an exemplary system architecture 800 of an item recommendation method or an item recommendation device to which embodiments of the present invention may be applied.

As shown in fig. 8, a system architecture 800 may include terminal devices 801, 802, 803, a network 804, and a server 805. The network 804 serves as a medium for providing communication links between the terminal devices 801, 802, 803 and the server 805. The network 804 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

A user may interact with the server 805 through the network 804 using the terminal devices 801, 802, 803 to receive or send messages or the like. Various communication client applications may be installed on the terminal devices 801, 802, 803, such as an item recommendation class application, an item query class application, a search class application, an instant messaging tool, a mailbox client, social platform software, and the like (by way of example only).

The terminal devices 801, 802, 803 may be a variety of electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop and desktop computers, and the like.

The server 805 may be a server providing various services, such as a background management server (by way of example only) that provides support for requests for item recommendations sent upstream using the terminal devices 801, 802, 803. The background management server can respond to receiving the user behavior data of the target user and determine the data characteristics of the user behavior data; acquiring regional information of the target user according to the data characteristics, and determining regional related articles of the target user according to the regional information and the user behavior data; recommending the area-associated item to the target user; and feeds back the recommended condition of the item (only by way of example) to the terminal device.

It should be noted that, the method for recommending an item according to the embodiment of the present invention is generally performed by the server 805, and accordingly, the device for recommending an item is generally provided in the server 805. The method for recommending the articles provided by the embodiment of the invention can also be executed by the terminal equipment 801, 802 and 803, and correspondingly, the device for recommending the articles can be arranged in the terminal equipment 801, 802 and 803.

It should be understood that the number of terminal devices, networks and servers in fig. 8 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

Referring now to FIG. 9, there is illustrated a schematic diagram of a computer system 900 suitable for use in implementing an embodiment of the present invention. The terminal device shown in fig. 9 is only an example, and should not impose any limitation on the functions and the scope of use of the embodiment of the present invention.

As shown in fig. 9, the computer system 900 includes a Central Processing Unit (CPU) 901, which can execute various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 902 or a program loaded from a storage section 908 into a Random Access Memory (RAM) 903. In the RAM 903, various programs and data necessary for the operation of the system 900 are also stored. The CPU 901, ROM 902, and RAM 903 are connected to each other through a bus 904. An input/output (I/O) interface 905 is also connected to the bus 904.

The following components are connected to the I/O interface 905: an input section 906 including a keyboard, a mouse, and the like; an output portion 907 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and a speaker; a storage portion 908 including a hard disk or the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to the I/O interface 905 as needed. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is installed as needed on the drive 910 so that a computer program read out therefrom is installed into the storage section 908 as needed.

In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 909 and/or installed from the removable medium 911. When the computer program is executed by a Central Processing Unit (CPU) 901, the above-described functions defined in the system of the embodiment of the present invention are performed.

It should be noted that, the computer readable medium shown in the embodiments of the present invention may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In embodiments of the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In contrast, in embodiments of the present invention, the computer-readable signal medium may comprise a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer programs according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules involved in the embodiments of the present invention may be implemented in software or in hardware. The described modules may also be provided in a processor, for example, as: a processor comprises a first determination module, a second determination module, a recommendation module, wherein the names of these modules do not in some cases constitute a limitation of the module itself, e.g. the first building module may also be described as a "building similar user set module".

As another aspect, the embodiment of the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiment; or may be present alone without being fitted into the device. The computer readable medium carries one or more programs which, when executed by a device, implement the method of: determining data characteristics of user behavior data in response to receiving the user behavior data of a target user; acquiring regional information of the target user according to the data characteristics, and determining regional related articles of the target user according to the regional information and the user behavior data; recommending the area-associated article to the target user.

The above detailed description should not be construed as limiting the scope of the embodiments of the present invention. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives can occur depending upon design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the embodiments of the present invention should be included in the scope of the embodiments of the present invention.

Claims

1. A method of item recommendation, comprising:

recommending the area-associated article to the target user.

2. The method of claim 1, wherein the data features comprise: the data amount of the user behavior data; obtaining the regional information of the target user according to the data characteristics, and determining the regional associated object of the target user according to the regional information and the user behavior data, wherein the method comprises the following steps:

3. The method of claim 2, wherein obtaining regional user information for the target user based on the geographic location of the target user comprises:

4. A method according to claim 3, wherein constructing an area user representation of the target user based on the area user information comprises:

5. A method according to claim 3, wherein meshing the user behavior data of the neighboring users to obtain a plurality of meshed user behavior data comprises:

6. The method according to claim 2, wherein in case the data amount of the user behavior data is smaller than the data amount threshold, the method further comprises:

7. The method of claim 6, wherein obtaining regional store information for the target user based on the geographic location of the target user comprises:

8. The method of claim 7, wherein constructing a representation of the geographically located regional item based on the regional store information comprises:

9. An apparatus for recommending items, comprising:

10. An electronic device, comprising:

one or more processors;

storage means for storing one or more programs,

the one or more processors implement the method of any of claims 1-8 when the one or more programs are executed by the one or more processors.

11. A computer readable medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any of claims 1-8.