Merchant Clustering Method and Device, Merchant Recommendation Method and Device, Electronic Device

Abstract

The present application discloses a merchant clustering method, belonging to the field of computer technology, which helps to improve the efficiency of recommending merchants to users based on the clustering results. The merchant clustering method disclosed in the embodiments of the present application includes: obtaining the scenario play methods matched by each merchant included in the target merchant pool, where each merchant included in the target merchant pool matches the target recommendation scenario; clustering the merchants included in the target merchant pool according to the scenario play method matched by each merchant, clustering the merchants that match the same scenario play method into the same merchant category, and each merchant category corresponds to one of the scenario play methods.

Description

Technical Field

[0001] The present application relates to the field of computer technologies, and in particular, to a merchant clustering method and device, a merchant recommendation method and device, an electronic device, and a computer-readable storage medium. Background Art

[0002] With the development of network technologies, people obtain more and more information through network platforms and are more dependent on the information provided by network platforms. Taking travel applications as an example, people hope that the platform recommends merchants suitable for their travel needs. In order to improve the user experience and increase user traffic, network platforms continuously explore technical solutions for recommending merchants that match the user's needs to improve the efficiency of users obtaining information through network platforms. For example, after analyzing the search historical data of users, the network platform recommends the most searched merchants of the users to the target user (i.e., the user currently accessing the network platform); or, according to the search conditions input by the target user, the business scope of merchants is matched, and merchants whose business scope matches the search conditions are recommended to the target user; or, merchants are labeled according to the results of manual identification to determine the recommended merchants, etc.

[0003] The merchant recommendation methods in the prior art cannot well match the needs of the target user, and there are problems of low recommendation efficiency when recommending merchants to the target user. Summary of the Invention

[0004] The present application provides a merchant clustering method and a merchant recommendation method, which can improve the efficiency of recommending merchants to the target user.

[0005] To solve the above problems, in a first aspect, an embodiment of the present application provides a merchant clustering method, including:

[0006] Obtaining the scenario play methods matched by each merchant included in the target merchant pool, where each merchant included in the target merchant pool matches the target recommendation scenario;

[0007] Clustering the merchants included in the target merchant pool according to the scenario play method matched by each merchant, clustering the merchants that match the same scenario play method into the same merchant category, and each merchant category corresponds to one of the scenario play methods.

[0008] In a second aspect, an embodiment of the present application provides a merchant clustering device, including:

[0009] A merchant scenario play method obtaining module, configured to obtain the scenario play methods matched by each merchant included in the target merchant pool, where each merchant included in the target merchant pool matches the target recommendation scenario;

[0010] A scenario-based gameplay clustering module, which is used to cluster the merchants included in the target merchant pool according to the scenario-based gameplay matched by each merchant, cluster the merchants that match the same scenario-based gameplay into the same merchant category, and each merchant category corresponds to one scenario-based gameplay.

[0011] Thirdly, an embodiment of the present application provides a merchant recommendation method, including:

[0012] Determine candidate merchants that match the target recommendation scenario;

[0013] According to the distribution ratio of users with each preset user attribute among the users of each candidate merchant, determine a candidate merchant pool that matches each preset user attribute, where the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one candidate merchant;

[0014] For each candidate merchant pool, respectively perform an operation of sorting the candidate merchants in each candidate merchant pool according to the merchant comprehensive quality score, and determine the sorting result of the candidate merchants in each candidate merchant pool;

[0015] According to the sorting result, determine the target merchant pool that matches each preset user attribute;

[0016] Cluster the merchants included in the target merchant pool through the merchant clustering method in the embodiment of the present application, and determine the merchant categories included in the target merchant pool, where each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario-based gameplay;

[0017] Recommend the merchants included in the target merchant pool according to the merchant categories.

[0018] Fourthly, an embodiment of the present application provides a merchant recommendation device, including:

[0019] A candidate merchant determination module, which is used to determine candidate merchants that match the target recommendation scenario;

[0020] A candidate merchant pool determination module, which is used to determine a candidate merchant pool that matches each preset user attribute according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant, where the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one candidate merchant;

[0021] A sorting module, configured to perform, for each of the candidate merchant pools, an operation of sorting each candidate merchant in the candidate merchant pool according to the comprehensive merchant quality score, and determine the sorting result of the candidate merchants in each candidate merchant pool;

[0022] A target merchant pool determination module, configured to determine a target merchant pool matching each of the preset user attributes according to the sorting result;

[0023] A merchant clustering module, configured to cluster the merchants included in the target merchant pool by the merchant clustering method according to any one of claims 1 to 5, and determine the merchant categories included in the target merchant pool, where each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario play method;

[0024] A merchant recommendation module, configured to classify and recommend the merchants included in the target merchant pool according to the merchant categories.

[0025] In a fifth aspect, an embodiment of the present application further discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the merchant clustering method and / or the merchant recommendation method described in the embodiments of the present application are implemented.

[0026] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the merchant clustering method and / or the merchant recommendation method disclosed in the embodiments of the present application are performed.

[0027] The merchant clustering method disclosed in the embodiments of the present application obtains the scenario play methods matched by each merchant included in the target merchant pool, and each merchant included in the target merchant pool matches the target recommendation scenario. Then, according to the scenario play method matched by each merchant, the merchants included in the target merchant pool are clustered, and the merchants matching the same scenario play method are clustered into the same merchant category. Each merchant category corresponds to one scenario play method, which helps to classify and recommend merchants according to the scenario play method, makes the recommended merchants more matched with the target recommendation scenario where the user is located, and improves the merchant recommendation efficiency. Description of the Drawings

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is the flowchart of the merchant clustering method in the first embodiment of the present application;

[0030] Figure 2 It is the flowchart of the merchant recommendation method in the second embodiment of the present application;

[0031] Figure 3 It is the schematic diagram of the merchant recommendation result in the second embodiment of the present application;

[0032] Figure 4 It is the schematic diagram of the structure of the merchant clustering device in the third embodiment of the present application;

[0033] Figure 5 It is the schematic diagram of the structure of the merchant recommendation device in the fourth embodiment of the present application. Detailed implementation manners

[0034] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0035] Embodiment 1

[0036] A merchant clustering method disclosed in the embodiment of the present application, as Figure 1 shown, the method includes: step 110 and step 120.

[0037] Step 110, obtain the scenario play methods matched by each merchant included in the target merchant pool, and each merchant included in the target merchant pool matches the target recommendation scenario.

[0038] In the embodiment of the present application, the target merchant pool is a set composed of merchants that match a certain target recommendation scenario. Among them, the target recommendation scenario matching includes scenarios for merchant recommendation based on any dimension factors in user, time, and space. For example, the target recommendation scenarios based on the user dimension include: parent-child travel scenario, senior travel scenario, group travel scenario; for another example, the target recommendation scenarios based on the time dimension include: weekend travel scenario, golden week travel scenario, summer vacation travel scenario; and for another example, the target recommendation scenarios based on the space dimension include: surrounding travel scenario.

[0039] Further, the target merchant pool is: a merchant pool composed of several merchants suitable for the parent-child travel scenario, a merchant pool composed of several merchants suitable for the surrounding travel scenario, a merchant pool composed of several merchants suitable for the weekend travel scenario, a merchant pool composed of several merchants suitable for the travel scenario with children aged 6 to 12, etc.

[0040] Merchants in the embodiments of the present application can be merchant users registered on network platforms such as tourist attractions, training institutions, entertainment venues, libraries, science and technology museums, exhibition halls, shopping malls, swimming pools, stadiums, schools, and kindergartens.

[0041] The target merchant pool in the embodiments of the present application is a set of merchants to be recommended to users under a certain target recommendation scenario determined in advance. For example, in the scenario of family travel, a merchant pool composed of several merchants recommended for the scenario of traveling with children aged 6 to 12 (such as children's parks, zoos, botanical gardens, etc.). Another example is that in the scenario of weekend travel, a merchant pool composed of several merchants suitable for weekend travel of users in a certain city (such as tourist attractions, museums, cinemas, etc.).

[0042] When the present application is specifically implemented, each merchant in the target merchant pool matches a certain scenario play method. For example, the scenario play methods matched by the zoo are: going to the zoo, going to the aquarium; the scenario play methods matched by the training institution are: learning art, learning calligraphy, learning dance, learning Go, learning Chinese painting.

[0043] In some embodiments of the present application, several merchants matching each target recommendation scenario can be determined in advance. Usually, the several merchants here are all merchants on the network platform that match the target recommendation scenario. Then, for all merchants matching each target recommendation scenario, according to the associated data of the merchants, the scenario play methods respectively matched by each merchant are further determined.

[0044] Specifically, the scenario play methods matched by each merchant are determined through the following steps S1 to S3.

[0045] Step S1, obtain a merchant pool composed of merchants matching the target recommendation scenario, and the merchants included in the merchant pool include the merchants included in the target merchant pool.

[0046] Taking the example of determining the scenario play methods matched by the merchants matching the target recommendation scenario of family travel, first, it is necessary to determine all merchants suitable for the family travel scenario. When the present application is specifically implemented, the merchants suitable for the family travel scenario can be determined by one or both of the following methods, and a merchant pool is formed by the determined merchants suitable for the family travel scenario.

[0047] First, by mining the user comment data of merchants by users on the network platform, the merchants matching the family travel scenario can be determined. For example, based on the comment words matching the family travel scenario, the user comment data of merchants is mined to determine the merchants matching the family travel scenario.

[0048] When specifically implemented, before mining the user comment data of merchants based on the comment words matching the family travel scenario to determine the merchants matching the family travel scenario, it is first necessary to determine the comment words matching the family travel scenario.

[0049] In some embodiments of the present application, comment words matching the target recommendation scenario can be determined based on expert experience. For example, for the scenario of parent-child outings, the following words are determined as comment words according to expert experience: taking the baby, kids, children, child, bringing the child here, the child likes, kindergarten children, bringing the daughter, bringing the son, children, parent-child, taking the baby for a walk, baby, etc.

[0050] In other embodiments of the present application, keyword frequency analysis can also be performed on the user comment data of users with preset user attributes (such as having parent-child identity information), and keywords whose frequency meets the preset conditions are determined as comment words matching the target recommendation scenario. For example, for users with parent-child identity information, the comments they post are statistically analyzed as a whole, and after sorting the keywords with the highest frequency in the comment content and screening, the keywords ranked in the top 5% with a higher quantile are taken as comment words matching the target recommendation scenario.

[0051] After determining the comment words matching the parent-child outing scenario, the user comment data of merchants on the network platform is mined based on the comment words belonging to the parent-child outing scenario to determine the merchants matching the parent-child outing scenario. For example, for each merchant, it is mined in the user comment data of the merchant whether the above-mentioned determined comment words appear. If the above-mentioned comment words appear in the user comments of a certain merchant, it can be determined that the merchant matches the parent-child outing scenario. In other embodiments of the present application, the number of times the above-mentioned comment words appear in the user comment data of a certain merchant can also be determined, and when the number of times the above-mentioned comment words appear exceeds a preset value (such as 10 times), it is determined that the merchant matches the parent-child outing scenario.

[0052] Second, the historical behavior data of users with parent-child identity information can be further mined according to the identity information of users on the network platform, and all merchants matching the parent-child outing scenario can be determined by mining the user comment data of merchants through the identity information of users on the network platform.

[0053] For example, the merchants followed by users with parent-child identity information are determined as merchants matching the parent-child outing scenario. For another example, the merchants visited by users with parent-child identity information are determined as merchants matching the parent-child outing scenario.

[0054] In the embodiments of the present application, the parent-child identity information of the user is determined according to the information input by the user when registering on the network platform. For example, when the user registers on the network platform, it is input whether there are children and / or the age of the children; or, the user's occupation is father or mother, etc.

[0055] In other embodiments of the present application, other methods can also be used to determine the parent-child identity information of the user, and the present application does not limit this.

[0056] Step S2: Mine the associated data of each merchant in the merchant pool to determine the scenario playstyles covered by the merchants in the merchant pool and the keywords matched by each scenario playstyle.

[0057] In the embodiments of the present application, the scenario playstyle refers to the actions performed for that scenario. The scenario playstyles matched by merchants of different business types may be different. For example, for the merchant "zoo", the scenario playstyles include: going to the zoo, visiting the zoo; while for the merchant "museum", the scenario playstyles include: visiting the museum.

[0058] After determining the merchant pool composed of several merchants that match the parent-child travel scenario, further determine the scenario playstyles supported by each merchant in the merchant pool according to the associated data of each merchant in the merchant pool (such as the business type information of the merchant, the user review data of the merchant).

[0059] In some embodiments of the present application, the mining of the associated data of each merchant in the merchant pool to determine the scenario playstyles covered by the merchants in the merchant pool and the keywords matched by each scenario playstyle includes: determining the scenario playstyles supported by each merchant in the merchant pool and the candidate keywords corresponding to each scenario playstyle according to the business type data of each merchant in the merchant pool; based on the candidate keywords, determining the keywords corresponding to each scenario playstyle by performing word matching on the user review data of each merchant in the merchant pool.

[0060] Still taking the parent-child travel scenario as an example, for each merchant in the merchant pool determined in the previous step, by presetting a list of scenario playstyles that conform to the parent-child travel scenario, and then using web crawler technology to crawl the business type data of each merchant in the merchant pool, determine the scenario playstyles supported by each merchant in the merchant pool, and finally determine all the scenario playstyles covered by all the merchants in the merchant pool.

[0061] In some embodiments of the present application, the list of scenario playstyles supported by the merchants of a certain target recommended scenario can be determined according to expert experience. For example, all the scenario playstyles covered by all the merchants in the finally determined merchant pool include the following scenario playstyles: going to the botanical garden, going to the zoo, watching children's plays, going to the aquarium, going skiing, taking a hot spring bath, going to the film and television base, visiting the park, going to the farmhouse, emotional quotient cultivation, going to the picture book library, learning taekwondo, going to the library, learning programming, learning electronic drums, learning roller skating, learning art, learning Go, learning calligraphy, learning traditional Chinese painting, learning Chinese zither, learning ballet, learning English, baby swimming, learning swimming, going to early education, attending nursery, thinking training, children's theme park, DIY, trampoline, summer camp, social experience, going to the parent-child farm, cool water play, going to the party, playing karting, playing bumper cars, shopping in the mall, getting a haircut, taking pictures, staying in a hotel, staying in a homestay.

[0062] Further, determine the candidate keywords corresponding to each scenario gameplay. For example, the determined keywords corresponding to the scenario gameplay "going to the zoo" include: zoo, wild animals, rare animals; the determined keywords corresponding to the scenario gameplay "going to the botanical garden" include: botanical garden. The candidate keywords corresponding to each scenario gameplay are usually words such as the merchant names included in the scenario gameplay, the representative product names of the merchants, etc.

[0063] After initially determining the candidate keywords matched with each scenario gameplay in the scenario gameplay list, further screen the candidate keywords, delete the candidate keywords that do not meet the requirements, and use the candidate keywords that meet the requirements as the keywords matched with the scenario gameplay.

[0064] In some embodiments of the present application, based on the candidate keywords, by performing word matching on the user comment data of each merchant in the merchant pool, determining the keywords corresponding to each scenario gameplay includes: by performing word matching between the candidate keywords and the user comment data of each merchant in the merchant pool, determining the occurrence frequency of each candidate keyword in the user comment data; using the candidate keywords whose occurrence frequency in the user comment data meets the preset conditions as the keywords corresponding to the corresponding scenario gameplay.

[0065] For example, traverse all the user comment contents of all merchants in the merchant pool. If a candidate keyword matched with a scenario gameplay in the preset gameplay list appears in the user comment content, count the number of times the candidate keyword appears. Finally, determine the number of times each candidate keyword matched with the scenario gameplay appears in the user comment content. If the cumulative occurrence frequency of a certain candidate keyword in the user comment contents of all merchants in the merchant pool is less than the preset frequency threshold (such as 10 times), then discard the candidate keyword; if the cumulative occurrence frequency of a certain candidate keyword in the user comment contents of all merchants in the merchant pool is greater than or equal to the preset frequency threshold, then determine that the candidate keyword is the keyword corresponding to a certain scenario gameplay. Taking the candidate keywords matched with the scenario gameplay "going to the zoo" including: zoo, wild animals, rare animals as an example, if in the user comments of all merchants in the merchant pool, the candidate keyword "wild animals" only appears 8 times, then determine the keywords matched with "going to the zoo" include: zoo, rare animals, and delete the candidate keyword "wild animals".

[0066] In order to match richer, more comprehensive and accurate keywords for each scenario gameplay, in some other embodiments of the present application, after the step of using the candidate keywords whose occurrence frequencies in the user comment data meet the preset conditions as the keywords corresponding to the corresponding scenario gameplay, the following is further included: for each of the keywords corresponding to the scenario gameplay, according to the synonyms or near-synonyms of the keywords that appear in the user comment data, expand the keywords corresponding to the scenario gameplay corresponding to the keyword. For example, for the keywords that have been determined to match the scenario gameplay, that is, the candidate keywords that meet the occurrence frequency conditions, crawl the user comment content and performance of the merchants in the merchant pool again to determine the similar words, synonyms and near-synonyms of the keywords that match the scenario gameplay that appear in the user comment content. For example, crawl the user comment content of the merchant and find that the word "Zoo" (the English expression for zoo) appears. If the word "Zoo" appears in the user comment content of the merchant, then add this word as a keyword for the scenario gameplay of "going to the zoo", and the keywords for the scenario gameplay of "going to the zoo" are expanded to: zoo, rare animals, Zoo.

[0067] Step S3, determine the scenario gameplay matched by each merchant according to the frequency information of the keyword that appears in the user comment data of each merchant in the merchant pool.

[0068] When the word "zoo" appears in a certain user comment of a certain merchant, then there is a possibility that the merchant supports the scenario gameplay of "going to the zoo". In some embodiments of the present application, determine the scenario gameplay matched by each merchant according to the frequency information of the keyword that appears in the user comment data of each merchant in the merchant pool.

[0069] Further, determining the scenario playstyles matched by each merchant according to the frequency information of the keywords appearing in the user review data of each merchant in the merchant pool includes: for each merchant in the merchant pool, perform the following operations respectively: determine the total frequency of all keywords matched by each scenario playstyle appearing in the user review data of the merchant; take the scenario playstyle with the largest total frequency of the matched keywords as the scenario playstyle matched by the merchant. Taking the merchant "zoo" as an example, first traverse all the user review contents of the merchant "zoo", determine the number of user reviews containing each of the previously determined keywords (such as: zoo, rare animals, Zoo, botanical garden, aquarium, etc.), and determine the number of user reviews corresponding to different scenario playstyles according to the number of user reviews containing each keyword. For example, among all the user reviews of the merchant "zoo", there are 1000 user reviews containing the keyword "zoo", 100 user reviews containing the keyword "rare animals", 80 user reviews containing the keyword "Zoo", 2 user reviews containing the keyword "botanical garden", and 100 user reviews containing the keyword "aquarium". Then, the number of user reviews containing all the keywords matched by the scenario playstyle of "going to the zoo" can be accumulated as the number of user reviews matched by the scenario playstyle of "going to the zoo", and the number of user reviews matched by the scenario playstyle of "going to the zoo" is obtained = the number of user reviews containing the keyword "zoo" 1000 + the number of user reviews containing the keyword "rare animals" 100 + the number of user reviews containing the keyword "Zoo" 80 = 1180.

[0070] According to this method, the number of user reviews matched by each scenario playstyle supported by the merchant "zoo" can be determined.

[0071] Still taking the merchant "zoo" as an example, if the specific playstyles corresponding to this merchant include: going to the zoo, going to the aquarium, and among them, the number of user reviews corresponding to "going to the zoo" is 1180, and the number of user reviews corresponding to "going to the aquarium" is 200, then mark the playstyle matched by the merchant "zoo" as "going to the zoo".

[0072] According to this method, the scenario playstyles of each merchant can be marked. By determining the scenario playstyles matched by the merchant according to the frequency of the keywords matched by the scenario playstyles in the user review content, rather than simply determining the scenario playstyles matched by the merchant according to the assumed regulations or merchant-defined attribute information, the scenario playstyles matched by the merchant are more in line with the public's perception, and further, the matching degree of the extracted scenario playstyles with user needs, and the matching degree of merchant classification and user needs when making merchant recommendations based on scenario playstyles can be improved.

[0073] Here, the number of comments for each merchant can also be obtained.

[0074] In specific implementation, a scenario play attribute information can be set for each merchant to mark the scenario play that the merchant matches.

[0075] According to the above method, the corresponding scenario plays that match the merchants for each target recommended scenario can be determined respectively. For example, the scenario plays that match the merchants for weekend outings can be determined. Taking the merchants for weekend outings including hotels, tourist attractions, museums, training institutions, etc. as an example, it can be determined that: the scenario play that matches a certain hotel is accommodation, and the scenario plays that match a certain museum are visiting the museum and visiting exhibitions.

[0076] In some other embodiments of the present application, after determining the candidate merchants that match a certain target recommended scenario, for all candidate merchants of the target recommended scenario, the scenario plays that each candidate merchant matches can be further determined. For example, for the parent-child outing scenario, after determining a merchant pool composed of merchants suitable for parent-child outings in a certain city, a merchant pool can be constructed only based on the merchants suitable for parent-child outings in the network platform, and the scenario plays can be marked for each merchant in the merchant pool, rather than processing the merchants suitable for parent-child outings across the entire network platform, which can improve the data processing efficiency.

[0077] Step 120, cluster the merchants included in the target merchant pool according to the scenario plays that each merchant matches, and cluster the merchants that match the same scenario play into the same merchant category, and each merchant category corresponds to one of the scenario plays.

[0078] After determining the scenario plays that each merchant in the target merchant pool matches, clustering the merchants in the target merchant pool based on the scenario plays that each merchant matches can obtain one or more merchant categories. Each merchant category corresponds to one scenario play, that is, the merchants in each merchant category match the same scenario play.

[0079] The merchant clustering method disclosed in the embodiments of the present application, by obtaining the scenario plays that each merchant included in the target merchant pool matches, and the merchants included in the target merchant pool match the target recommended scenario, and then clustering the merchants included in the target merchant pool according to the scenario plays that each merchant matches, and clustering the merchants that match the same scenario play into the same merchant category, and each merchant category corresponds to one of the scenario plays, helps to classify and recommend or display merchants according to the scenario plays.

[0080] Further, since the merchants in the target merchant pool match the target recommendation scenario, and each merchant matches at least one scenario play method, and the scenario play methods of the merchants are obtained through data mining based on the associated data of the merchants (such as the business type data and user review data of the merchants), the obtained scenario play methods are more accurate. When the target recommendation scenario is a travel scenario, clustering the merchants based on the scenario play methods, the clustering result is more in line with the user's needs and can be clearly presented to the user, improving the merchant recommendation efficiency.

[0081] Embodiment 2

[0082] The following combines Figure 2 to detail a specific application of Embodiment 1 of the present application. As Figure 2 described, a merchant recommendation method disclosed in this embodiment includes: Step 210 to Step 260.

[0083] Step 210, determine candidate merchants that match the target recommendation scenario.

[0084] In some embodiments of the present application, the determining of candidate merchants that match the target recommendation scenario includes: mining the user review data of the merchants based on the review words that match the target recommendation scenario to determine candidate merchants; or, determining candidate merchants according to the merchants followed by the users whose identity information matches the target recommendation scenario and / or the merchants where the users have access behavior.

[0085] The target recommendation scenario described in this embodiment is a target recommendation scenario based on the user dimension. For example, a parent-child travel scenario, a senior travel scenario, a group travel scenario. In this embodiment, the technical solution of merchant recommendation is described in detail by taking the parent-child travel scenario as an example. Then, the user identity information that matches the parent-child travel scenario is parent-child identity information.

[0086] In the embodiments of the present application, the parent-child identity information of the user is determined according to the information input by the user when registering on the network platform. For example, when the user registers on the network platform, the user inputs whether there are children and / or the ages of the children; or the user's occupation is father or mother, etc.

[0087] In other embodiments of the present application, other methods may also be used to determine the parent-child identity information of the user, and the present application does not limit this.

[0088] In some embodiments of the present application, further mine the historical behavior data of the users with parent-child identity information according to the identity information of the users on the network platform, and determine the merchants followed by the users with parent-child identity information as the merchants that match the parent-child travel scenario. For another example, determine the merchants visited by the users with parent-child identity information as the merchants that match the parent-child travel scenario.

[0089] In some other embodiments of the present application, when mining the user review data of merchants based on the review words matching the target recommendation scenario, the steps of determining candidate merchants include: for each merchant, mining the number of the user review data in which review words matching the target recommendation scenario appear in the user review data of the merchant; determining the merchant whose number of the user review data meets a preset quantity threshold as a candidate merchant.

[0090] In specific implementation, when mining the user review data of merchants based on the review words matching the parent-child travel scenario, before determining the merchants matching the parent-child travel scenario, it is first necessary to determine the review words matching the parent-child travel scenario.

[0091] In some embodiments of the present application, the review words matching the target recommendation scenario can be determined according to expert experience. For example, for the parent-child travel scenario, the following words are determined as review words according to expert experience: taking the baby, little friends, children, kids, bringing the child here, the child likes, kindergarten children, bringing the daughter, bringing the son, children, parent-child, walking the baby, baby, etc.

[0092] In some other embodiments of the present application, the keywords whose word frequencies meet the preset conditions can also be determined as the review words matching the target recommendation scenario by performing keyword frequency analysis on the user review data of users with preset user attributes (such as having parent-child identity information). Before mining the user review data of merchants based on the review words matching the target recommendation scenario to determine candidate merchants, it also includes: performing keyword frequency analysis on the user review data of users with the preset user attributes to determine the keywords whose word frequencies meet the preset conditions as the review words matching the target recommendation scenario.

[0093] For example, for users with parent-child identity information, the comments they post are statistically analyzed as a whole, the keywords with the highest frequencies in the comment content are sorted and screened, and the keywords ranked in the top 5% with a higher quantile are taken as the review words matching the target recommendation scenario.

[0094] After determining the review words matching the parent-child travel scenario, the user review data of merchants on the network platform is mined based on the review words belonging to the parent-child travel scenario to determine the merchants matching the parent-child travel scenario. For example, for each merchant, it is mined whether the above-determined review words appear in the user review data of the merchant. If the above review words appear in the user review of a certain merchant, it can be determined that the merchant matches the parent-child travel scenario. In some other embodiments of the present application, the number of times the above review words appear in the user review data of a certain merchant can also be determined, and when the number of times the above review words appear exceeds a preset value (such as 10 times), it is determined that the merchant matches the parent-child travel scenario.

[0095] Step 220: Determine a candidate merchant pool matching each of the preset user attributes according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant.

[0096] The preset user attributes match the target recommendation scenario. For example, in the scenario of family travel with children, the preset user attributes include: age group attributes, such as parents of babies aged 1 to 3, parents of babies aged 3 to 6, and parents of babies aged 6 to 12; in the scenario of senior travel, the preset user attributes include: age group attributes, such as users aged 50 to 60, 60 to 70; in the scenario of group travel, the user attributes include: user occupations, such as students, outdoor sports enthusiasts, teachers.

[0097] Each candidate merchant pool includes at least one of the candidate merchants, and the number of designated users of the at least one candidate merchant meets a preset quantity condition, where the designated user refers to a user with a preset user attribute matching the target recommendation scenario. For example, in the scenario of family travel with children, among the users of the determined candidate merchant A, the number of parents of babies aged 1 to 3, parents of babies aged 3 to 6, or parents of babies aged 6 to 12 needs to meet the preset quantity condition (such as more than 10 people).

[0098] Next, group the candidate merchants according to the preset user attributes matching the target recommendation scenario. The merchants in each group have the same preset user attribute, and the candidate merchants in each group form a candidate merchant pool. Taking the scenario of family travel with children as an example, the determined candidate merchant pools may include: a merchant pool matching babies aged 1 to 3, a merchant pool matching babies aged 3 to 6, and a merchant pool matching babies aged 6 to 12.

[0099] In some embodiments of the present application, determining a candidate merchant pool matching each of the preset user attributes according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant includes: respectively determining the preset user attributes matched by each candidate merchant; constructing a candidate merchant pool matching the same preset user attribute according to the candidate merchants matching the same preset user attribute.

[0100] Among them, respectively determining the preset user attributes matched by each candidate merchant includes: for each candidate merchant, respectively performing the following operations: according to the user access history data of the candidate merchant, respectively determining the distribution ratio of users with each preset user attribute among the users accessing the candidate merchant; in response to the comparison result of each distribution ratio with a preset ratio threshold, determining the preset user attribute corresponding to the distribution ratio greater than the preset ratio threshold as the preset user attribute matched by the candidate merchant.

[0101] In some embodiments of the present application, according to the user access history data of the candidate merchant, the distribution ratios of users with each preset user attribute among the users accessing the candidate merchant are determined respectively, including: according to the user access history data of the candidate merchant, determining the number of users with each preset user attribute among the users accessing the candidate merchant; for each of the preset user attributes, taking the ratio of the number of users with the preset user attribute to the number of users accessing the candidate merchant as the distribution ratio of users with the preset user attribute of the candidate merchant.

[0102] For example, for candidate merchant B, first, according to the access history data of all users of candidate merchant B, determine the number of users with each preset user attribute among the users accessing candidate merchant B. For example, determine: the number of users who are parents of babies aged 1 to 3 years old, the number of users who are parents of babies aged 3 to 6 years old, and the number of users who are parents of babies aged 6 to 12 years old. Then, according to the number of users with each preset user attribute, determine the distribution ratio of users with each preset user attribute. For example, when among the users accessing candidate merchant B, the number of parents of babies aged 1 to 3 years old is 10, the number of parents of babies aged 3 to 6 years old is 20, and the number of parents of babies aged 6 to 12 years old is 30, then the distribution ratio of users with the preset user attribute of "being parents of babies aged 1 to 3 years old" of candidate merchant B is 10 / (10 + 20 + 30), the distribution ratio of users with the preset user attribute of "being parents of babies aged 3 to 6 years old" of candidate merchant B is 20 / (10 + 20 + 30), and the distribution ratio of users with the preset user attribute of "being parents of babies aged 6 to 12 years old" of candidate merchant B is 30 / (10 + 20 + 30).

[0103] After that, take the distribution ratios of users with each preset user attribute among the users of candidate merchant B as the distribution ratios corresponding to candidate merchant B and the respective preset user attributes. Specifically, for the above data, the distribution ratio corresponding to candidate merchant B and the preset user attribute of "being parents of babies aged 1 to 3 years old" is 1 / 6, the distribution ratio corresponding to the preset user attribute of "being parents of babies aged 3 to 6 years old" is 2 / 6, and the distribution ratio corresponding to the preset user attribute of "being parents of babies aged 6 to 12 years old" is 3 / 6. The distribution ratios corresponding to candidate merchant B and the respective preset user attributes reflect the proportion of users with the respective preset user attributes among the users of candidate merchant B.

[0104] After determining the distribution ratio of each candidate merchant corresponding to the respective preset user attributes, further compare the distribution ratio corresponding to each preset user attribute with a preset ratio threshold, and determine the preset user attributes matched by the candidate merchant according to the comparison result. For example, when the distribution ratio of candidate merchant B corresponding to the preset user attribute of "being a parent of a baby aged 1 to 3 years old" is greater than the preset ratio threshold, it is determined that candidate merchant B matches the preset user attribute of "being a parent of a baby aged 1 to 3 years old".

[0105] When a certain candidate merchant matches a preset user attribute, it means that the candidate merchant can be used as a recommended merchant for users with the preset user attribute. After determining the preset user attributes matched by each candidate merchant, further form candidate merchant pools for the candidate merchants that match the same preset user attribute, that is, put the merchants that can be recommended to users with the same preset user attribute into one merchant pool.

[0106] So far, one merchant pool for the merchants recommended to users with each preset user attribute can be determined. For example, one merchant pool for the parents of babies aged 1 to 3 years old, one merchant pool for the parents of babies aged 3 to 6 years old, and one merchant pool for the parents of babies aged 6 to 12 years old can be determined. By dividing the merchants according to the user attributes of the merchants to construct merchant pools, so that each merchant pool matches one user attribute, it helps to improve the matching degree between the support capabilities provided by the merchants and the user needs.

[0107] Step 230, for each of the candidate merchant pools, respectively perform the operation of sorting the candidate merchants in each candidate merchant pool according to the comprehensive quality score of the merchants, and determine the sorting results of the candidate merchants in each candidate merchant pool.

[0108] After being processed by the above steps, candidate merchant pools that match different preset user attributes can be obtained. Specifically in the aforementioned parent-child travel scenario, candidate merchant pools that match the parents of babies of different age groups can be obtained. Next, further sort the candidate merchants in each candidate merchant pool to select a certain number of candidate merchants from each candidate merchant pool according to specific recommendation requirements and finally recommend them to the parents of babies of the corresponding age groups.

[0109] In some embodiments of the present application, sorting each candidate merchant in the candidate merchant pool according to the comprehensive quality score of the merchant includes: respectively determining the positive review number index score, high-quality review number index score, daily average user view volume index score, original review number index score matching the target recommendation scenario, and original review proportion index score matching the target recommendation scenario for each candidate merchant in the candidate merchant pool; for each candidate merchant, performing a weighted summation operation on the positive review number index score, high-quality review number index score, daily average user view volume index score, original review number index score matching the target recommendation scenario, and original review proportion index score matching the target recommendation scenario of the candidate merchant with a preset weight to determine the comprehensive quality score of the candidate merchant. The preset weight is determined according to the matching degree verification result of the recommendation result or business requirements.

[0110] Among them, the positive review number index score is the Wilson score determined according to the number of positive reviews and the number of negative reviews in the user review data matching the target recommendation scenario; the high-quality review number index score is the Wilson score determined according to the number of high-quality reviews and the total number of reviews in the user review data matching the target recommendation scenario; the daily average user view volume index score is the normalized logarithmic value determined according to the daily average user view volume of the candidate merchant in the recent preset time period, which is used to indicate the level of the daily average user view volume; the original review number index score matching the target recommendation scenario is the normalized logarithmic value determined according to the amount of user review data matching the target recommendation scenario, which is used to indicate the amount of original reviews; the original review proportion index score matching the target recommendation scenario is the proportion of the amount of user review data matching the target recommendation scenario in the recent preset time period and the total amount of user reviews of the candidate merchant, which is used to indicate the proportion of original reviews matching the target recommendation scenario in all user reviews.

[0111] Taking the comprehensive quality score of the merchant represented as Z as an example, the comprehensive quality score of a certain merchant can be calculated by the formula Z = (a * 0.7 + b * 0.1 + c * 0.2) * 0.5 + d * 0.3 + e * 0.2, where a is the positive review number index score of the candidate merchant, b is the high-quality review number index score, c is the daily average user view volume index score, d is the original review number index score matching the target recommendation scenario, e is the original review proportion index score matching the target recommendation scenario, and the constants are the weights of each index, which are determined according to specific test results or business requirements.

[0112] Further, in some embodiments of the present application, the score a of the positive review number index of the candidate merchant is calculated based on the number of positive reviews of the users with parent-child identity information in the candidate merchant. For example, according to the number of positive reviews, negative reviews and total reviews of the users with parent-child identity information, the Wilson score is calculated through the Wilson algorithm, and the calculated Wilson score is used as the score a of the positive review number index of the candidate merchant.

[0113] In some embodiments of the present application, the score b of the high-quality review number index is calculated based on the number of high-quality reviews and the total number of reviews of the users with parent-child identity information in the candidate merchant. For example, according to the number of high-quality reviews, non-high-quality reviews and total reviews of the users with parent-child identity information, the Wilson score is calculated through the Wilson algorithm, and the calculated Wilson score is used as the score b of the high-quality review number index of the candidate merchant.

[0114] In some embodiments of the present application, for a certain candidate merchant, determine the average daily user view volume Num of the candidate merchant in the past year uv , and normalize the logarithm value obtained after taking the natural logarithm of the average daily user view volume Num uv to obtain a value c, and use the obtained value c as the score of the average daily user view volume index.

[0115] In some embodiments of the present application, for a certain candidate merchant, determine the number Num of user original comments containing keywords matching the target recommendation scenario (such as comment words matching the parent-child travel scenario) in the user original comment data of the candidate merchant ugc , and normalize the logarithm value obtained after taking the natural logarithm of the number Num of user original comments ugc to obtain a value d, and use the obtained value d as the score of the original comment number index.

[0116] In some embodiments of the present application, for a certain candidate merchant, determine the number Num1 of user original comments containing keywords matching the target recommendation scenario (such as comment words matching the parent-child travel scenario) in the user original comment data of the candidate merchant in a recent period (such as within one year) ugc , and all user original comment data NumAll ugc . Then, use the ratio e of the number Num1 of user original comments ugc and all user original comment data NumAll ugc as the score of the original comment proportion index.

[0117] After determining the comprehensive quality scores of each candidate merchant in each candidate merchant pool, for each candidate merchant pool, sort the candidate merchants in the pool in descending order of the comprehensive quality score from front to back.

[0118] By calculating the comprehensive quality score of merchants based on the above multiple indicators, such as: not only including user original data (such as user reviews), but also including daily average visit volume; not only including the intuitive data volume, but also including the quality analysis results of the data (such as the review needs to contain keywords related to parent-child travel), the comprehensive quality score of the merchant is made more objective and reliable.

[0119] In some embodiments of the present application, due to the different development levels of different cities, the industrial formats of different cities are also different, and only the merchants in the same city are comparable. Therefore, for the candidate merchants in a candidate merchant pool, they are first divided into different sub-candidate merchant pools according to different cities, and then, for each sub-candidate merchant pool, they are independently sorted according to the comprehensive quality score.

[0120] In this embodiment, only the example that the candidate merchants in the candidate merchant pool belong to the same city is given to illustrate the merchant recommendation scheme.

[0121] Step 240, according to the sorting result, determine the target merchant pool that matches each of the preset user attributes.

[0122] After determining the sorting results of the candidate merchants in each candidate merchant pool, for any one of the candidate merchant pools, take the top preset number or preset proportion of candidate merchants in the candidate merchant pool and add them to the target merchant pool, and the obtained target merchant pool matches the preset user attribute matched with the candidate merchant pool.

[0123] Step 250, cluster the merchants included in the target merchant pool by the merchant clustering method described in Embodiment 1 to determine the merchant categories included in the target merchant pool.

[0124] Among them, each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario play method. Among them, the scenario play method of the merchant can be marked before executing step 210, or can be determined after determining the candidate merchants that match the target recommendation scenario when executing step 210.

[0125] Taking the example that a number of candidate merchants D determined to match the target recommendation scenario when executing step 210 are represented, if the scenario play method of the merchant is marked after determining the candidate merchants that match the target recommendation scenario, the scenario play method matched by each merchant in the number of candidate merchants D determined when executing step 210 can be determined according to the associated data of the number of candidate merchants D.

[0126] Then, for each target merchant pool, clustering is performed respectively according to the method described in the embodiment. For example, for the target merchant pool 1 that matches the age group attribute "parents of babies aged 3 to 6", it includes multiple merchants such as zoos, playgrounds, etc. Each merchant may match one or more scenario play methods. After clustering all the merchants in the target merchant pool 1 based on the scenario play methods, multiple categories of merchants that match different scenario play methods will be obtained. For example, the scenario play method "going to the zoo" may match merchants such as "Beijing Zoo", "Badaling Safari Park", "Children's Zoo", etc., and the scenario play method "finding a place to play for a day" may match merchants such as "Century Park", "Expo Park", "Fragrant Hills", etc. And the merchants matched by the above different play methods are all suitable for recommendation to the parents of babies aged 3 to 6, that is, the merchants matched by the above different play methods all match the preset user attribute of "parents of babies aged 3 to 6".

[0127] Step 260, classify and recommend the merchants included in the target merchant pool according to the merchant categories.

[0128] Next, according to the merchant categories obtained by clustering, that is, according to different scenario play methods, classify and recommend the merchants included in the target merchant pool. The schematic diagram of the recommendation result is as Figure 3 shown. Figure 3 In [the figure], 310, 320, and 330 are the recommended merchants of three merchant categories corresponding to different scenario play methods respectively.

[0129] The merchant recommendation method disclosed in the embodiment of the present application determines candidate merchants that match the target recommendation scenario; determines candidate merchant pools that match each preset user attribute according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant, where the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one of the candidate merchants; then, for each candidate merchant pool, respectively perform the operation of sorting the candidate merchants in each candidate merchant pool according to the merchant comprehensive quality score to determine the sorting results of the candidate merchants in each candidate merchant pool; and according to the sorting results, determine the target merchant pools that match each preset user attribute; then cluster the merchants included in the target merchant pools based on the scenario play methods to determine the merchant categories included in the target merchant pools, where each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario play method; finally, classify and recommend the merchants included in the target merchant pool according to the merchant categories, so that the merchants recommended to the user match the preset user attributes possessed by the user, are more in line with the user's needs, and further improve the efficiency of recommending merchants.

[0130] Taking the scenario of parent-child outings as an example, through the merchant recommendation method disclosed in the embodiments of the present application, suitable merchants can be recommended according to the travel needs of parents with babies in different age groups, and the merchants are classified and recommended according to the scenario play methods, providing highly valuable reference information for the parent-child population.

[0131] Embodiment III

[0132] A merchant clustering device disclosed in this embodiment is as Figure 4 shown, and the device includes:

[0133] A merchant scenario play method acquisition module 410, configured to acquire the scenario play methods matched by each merchant included in the target merchant pool, where each merchant included in the target merchant pool matches the target recommendation scenario;

[0134] A clustering module 420 based on scenario play methods is configured to cluster the merchants included in the target merchant pool according to the scenario play methods matched by each merchant, and cluster the merchants that match the same scenario play method into the same merchant category, and each merchant category corresponds to one of the scenario play methods.

[0135] In some embodiments of the present application, the scenario play methods matched by each merchant are determined through the following steps:

[0136] Acquire a merchant pool composed of merchants that match the target recommendation scenario, and the merchants included in the merchant pool include the merchants included in the target merchant pool;

[0137] Mine the associated data of each merchant in the merchant pool to determine the scenario play methods covered by the merchants in the merchant pool and the keywords matched by each scenario play method;

[0138] Determine the scenario play method matched by each merchant according to the frequency information of the keywords appearing in the user comment data of each merchant in the merchant pool.

[0139] In some embodiments of the present application, the step of mining the associated data of each merchant in the merchant pool to determine the scenario play methods covered by the merchants in the merchant pool and the keywords matched by each scenario play method includes:

[0140] According to the business type data of each merchant in the merchant pool, determine the scenario play methods supported by each merchant in the merchant pool and the candidate keywords corresponding to each scenario play method;

[0141] Based on the candidate keywords, determine the keywords corresponding to each scenario play method by performing word matching on the user comment data of each merchant in the merchant pool.

[0142] In some embodiments of the present application, the step of determining the keywords corresponding to each of the scenario playstyles by performing word matching on the user review data of each merchant in the merchant pool based on the candidate keywords includes:

[0143] By performing word matching between the candidate keywords and the user review data of each merchant in the merchant pool, determine the occurrence frequency of each candidate keyword in the user review data;

[0144] Use the candidate keywords whose occurrence frequency in the user review data meets the preset conditions as the keywords corresponding to the respective scenario playstyles.

[0145] In some embodiments of the present application, after the step of using the candidate keywords whose occurrence frequency in the user review data meets the preset conditions as the keywords corresponding to the respective scenario playstyles, it further includes:

[0146] For each keyword corresponding to the scenario playstyle, expand the keywords corresponding to the scenario playstyle corresponding to the keyword according to the synonyms or near-synonyms of the keyword that appear in the user review data.

[0147] The merchant clustering device disclosed in the embodiments of the present application is used to implement the steps of the merchant clustering method described in Embodiment 1 of the present application. For the specific implementation manners of the modules of the device, refer to the corresponding steps of the merchant clustering method, which will not be elaborated here.

[0148] The merchant clustering device disclosed in the embodiments of the present application obtains the scenario playstyles matched by each merchant included in the target merchant pool. Each merchant included in the target merchant pool is matched with the target recommendation scenario. Then, according to the scenario playstyles matched by each merchant, cluster the merchants included in the target merchant pool, and cluster the merchants that match the same scenario playstyle into the same merchant category. Each merchant category corresponds to one of the scenario playstyles, which helps to classify and recommend or display merchants according to the scenario playstyles.

[0149] Furthermore, since the merchants in the target merchant pool are matched with the target recommendation scenario, and each merchant matches at least one scenario playstyle, and the scenario playstyles of the merchants are obtained through data mining based on the associated data of the merchants (such as the business type data, user review data, etc.) of the merchants, the obtained scenario playstyles are more accurate. When the target recommendation scenario is a travel scenario, clustering the merchants based on the scenario playstyles, the clustering result is more in line with the user's needs and can be clearly displayed to the user, improving the merchant recommendation efficiency.

[0150] Embodiment 4

[0151] A merchant recommendation device disclosed in this embodiment, such as Figure 5As shown, the device includes:

[0152] A candidate merchant determination module 510, configured to determine candidate merchants that match the target recommendation scenario;

[0153] A candidate merchant pool determination module 520, configured to determine a candidate merchant pool that matches each preset user attribute according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant, where the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one of the candidate merchants;

[0154] A sorting module 530, configured to perform an operation of sorting each candidate merchant in each candidate merchant pool according to the comprehensive quality score of the merchant for each candidate merchant pool, and determine the sorting result of the candidate merchants in each candidate merchant pool;

[0155] A target merchant pool determination module 540, configured to determine a target merchant pool that matches each preset user attribute according to the sorting result;

[0156] A merchant clustering module 550, configured to cluster the merchants included in the target merchant pool by the merchant clustering method described in Embodiment 1, and determine the merchant categories included in the target merchant pool, where each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario play method;

[0157] A merchant recommendation module 560, configured to classify and recommend the merchants included in the target merchant pool according to the merchant categories.

[0158] In some embodiments of the present application, the candidate merchant pool determination module 520 is further configured to:

[0159] Determine the preset user attributes matched by each candidate merchant respectively;

[0160] Construct a candidate merchant pool that matches the same preset user attribute according to the candidate merchants that match the same preset user attribute;

[0161] Wherein, determining the preset user attributes matched by each candidate merchant respectively includes:

[0162] For each candidate merchant, perform the following operations respectively:

[0163] According to the user access historical data of the candidate merchant, determine the distribution ratio of users with each preset user attribute among the users who access the candidate merchant respectively;

[0164] Determine, in response to the comparison result between each of the distribution ratios and a preset ratio threshold, the preset user attributes corresponding to the distribution ratios greater than the preset ratio threshold as the preset user attributes matched by the candidate merchants.

[0165] In some embodiments of the present application, the step of sorting each of the candidate merchants in the candidate merchant pool according to the merchant comprehensive quality score includes:

[0166] Respectively determine the good review number index score, high-quality review number index score, average daily user view volume index score, original review number index score matching the target recommendation scenario, and original review proportion index score matching the target recommendation scenario of each candidate merchant in the candidate merchant pool;

[0167] For each candidate merchant, perform a weighted summation operation on the good review number index score, high-quality review number index score, average daily user view volume index score, original review number index score matching the target recommendation scenario, and original review proportion index score matching the target recommendation scenario of the candidate merchant with preset weights to determine the merchant comprehensive quality score of the candidate merchant.

[0168] In some embodiments of the present application, the candidate merchant determination module 510 is further configured to:

[0169] Mine the user review data of the merchant based on the review words matching the target recommendation scenario to determine candidate merchants; or,

[0170] Determine candidate merchants according to the merchants followed by the users whose identity information matches the target recommendation scenario and / or the merchants where access behaviors occur.

[0171] In some embodiments of the present application, before mining the user review data of the merchant based on the review words matching the target recommendation scenario to determine candidate merchants, the candidate merchant determination module 510 is further configured to:

[0172] Perform keyword frequency analysis on the user review data of the users with the preset user attributes to determine the keywords whose word frequencies meet the preset conditions as the review words matching the target recommendation scenario.

[0173] The merchant recommendation device disclosed in the embodiments of the present application is used to implement the steps of the merchant recommendation method described in the second embodiment of the present application. For the specific implementation manners of the modules of the device, refer to the corresponding steps of the merchant recommendation method, which will not be elaborated here.

[0174] The merchant recommendation device disclosed in the embodiments of the present application, through the merchant recommendation method disclosed in the embodiments of the present application, determines candidate merchants that match the target recommendation scenario; determines candidate merchant pools that match each preset user attribute according to the distribution ratios of users with each preset user attribute among the users of each candidate merchant, where the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one of the candidate merchants; then, for each candidate merchant pool, respectively perform an operation of sorting the candidate merchants in each candidate merchant pool according to the comprehensive merchant quality score to determine the sorting results of the candidate merchants in each candidate merchant pool; and according to the sorting results, determine the target merchant pools that match each preset user attribute; then cluster the merchants included in the target merchant pools based on the scenario gameplay to determine the merchant categories included in the target merchant pools, where each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario gameplay; finally, classify and recommend the merchants included in the target merchant pools according to the merchant categories, so that the merchants recommended to users match the preset user attributes that the users possess, are more in line with the needs of users, and further improve the efficiency of recommending merchants.

[0175] Taking the scenario of parent-child outings as an example, through the merchant recommendation device disclosed in the embodiments of the present application, it is possible to recommend suitable merchants for the travel needs of parents with babies in different age ranges, and classify and recommend the merchants according to the scenario gameplay, providing highly valuable reference information for the parent-child population.

[0176] Correspondingly, the present application also discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the merchant clustering method described in Embodiment 1 of the present application and / or the merchant recommendation method described in Embodiment 2 of the present application. The electronic device can be a PC, a mobile terminal, a personal digital assistant, a tablet computer, etc.

[0177] The present application also discloses a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps of the merchant clustering method described in Embodiment 1 of the present application and / or the steps of the merchant recommendation method described in Embodiment 2 of the present application.

[0178] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments.

[0179] The above has introduced in detail a merchant clustering method and device, and a merchant recommendation method and device provided by the present application. Specific examples are used in this article to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present application.

[0180] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

Claims

1. A merchant recommendation method, characterized in that, it includes: Determine candidate merchants that match the target recommendation scenario; According to the distribution ratio of users with each preset user attribute among the users of each candidate merchant, determine a candidate merchant pool that matches each preset user attribute, where the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one of the candidate merchants; For each candidate merchant pool, respectively perform the operation of sorting the candidate merchants in each candidate merchant pool according to the merchant comprehensive quality score, and determine the sorting result of the candidate merchants in each candidate merchant pool; According to the sorting result, determine the target merchant pool that matches each preset user attribute; Cluster the merchants included in the target merchant pool through a merchant clustering method, and determine the merchant categories included in the target merchant pool, where each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario play method; Recommend the merchants included in the target merchant pool according to the merchant categories.

2. The method according to claim 1, characterized in that, The step of determining a candidate merchant pool that matches each preset user attribute according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant includes: Respectively determine the preset user attributes matched by each candidate merchant; According to the candidate merchants that match the same preset user attribute, construct a candidate merchant pool that matches the same preset user attribute; Among them, respectively determining the preset user attributes matched by each candidate merchant includes: For each candidate merchant, respectively perform the following operations: According to the user access history data of the candidate merchant, respectively determine the distribution ratio of users with each preset user attribute among the users who visit the candidate merchant; In response to the comparison result of each distribution ratio with a preset ratio threshold, determine the preset user attribute corresponding to the distribution ratio greater than the preset ratio threshold as the preset user attribute matched by the candidate merchant.

3. The method according to claim 2, characterized in that, The step of sorting the candidate merchants in each candidate merchant pool according to the merchant comprehensive quality score includes: Respectively determine the positive comment number index score, high-quality comment number index score, average daily user view volume index score, original comment number index score that matches the target recommendation scenario, and original comment ratio index score that matches the target recommendation scenario of each candidate merchant in the candidate merchant pool; For each candidate merchant, through weighted summation of the positive comment number index score, high-quality comment number index score, average daily user view volume index score, original comment number index score that matches the target recommendation scenario, and original comment ratio index score that matches the target recommendation scenario of the candidate merchant with a preset weight, determine the merchant comprehensive quality score of the candidate merchant.

4. The method according to claim 3, characterized in that, The step of determining candidate merchants that match the target recommendation scenario includes: Mining the user review data of merchants based on review words that match the target recommendation scenario to determine candidate merchants; Alternatively, determining candidate merchants according to the merchants followed by users whose identity information matches the target recommendation scenario and / or the merchants where access behaviors occur.

5. According to the method described in claim 4, wherein, Before the step of mining the user review data of merchants based on review words that match the target recommendation scenario to determine candidate merchants, it further includes: By performing keyword frequency analysis on the user review data of users with the preset user attributes, determining keywords whose word frequencies meet the preset conditions as review words that match the target recommendation scenario.

6. A merchant recommendation device, wherein, It includes: A candidate merchant determination module for determining candidate merchants that match the target recommendation scenario; A candidate merchant pool determination module for determining a candidate merchant pool that matches each preset user attribute according to the distribution ratio of users with each preset user attribute among the users of each candidate merchant, wherein the preset user attribute matches the target recommendation scenario, and each candidate merchant pool includes at least one of the candidate merchants; A sorting module for, for each of the candidate merchant pools, respectively performing an operation of sorting the candidate merchants in each candidate merchant pool according to the comprehensive quality score of the merchants to determine the sorting result of the candidate merchants in each candidate merchant pool; A target merchant pool determination module for determining a target merchant pool that matches each preset user attribute according to the sorting result; A merchant clustering module for clustering the merchants included in the target merchant pool by a merchant clustering method to determine the merchant categories included in the target merchant pool, wherein each merchant category includes at least one merchant in the target merchant pool, and the merchants in each merchant category correspond to the same scenario play method; A merchant recommendation module for classifying and recommending the merchants included in the target merchant pool according to the merchant categories.

7. An electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, When the processor executes the computer program, it implements the merchant recommendation method described in any one of claims 1 to 5.

8. A computer-readable storage medium, on which a computer program is stored, wherein, When the program is executed by the processor, it implements the steps of the merchant recommendation method described in any one of claims 1 to 5.

Images