CN118350904B

CN118350904B - User clothing order behavior analysis method and system based on big data

Info

Publication number: CN118350904B
Application number: CN202410750730.7A
Authority: CN
Inventors: 王丽莉; 王润邦; 王润基; 周洪涛
Original assignee: Qingzhou City Tanboer Clothing Corp ltd
Current assignee: Qingzhou City Tanboer Clothing Corp ltd
Priority date: 2024-06-12
Filing date: 2024-06-12
Publication date: 2024-09-17
Anticipated expiration: 2044-06-12
Also published as: CN118350904A

Abstract

The application provides a user clothing order behavior analysis method and a system based on big data, which can deeply analyze the collaborative interest state and the user interest correlation degree between users by acquiring the user clothing order behavior big data of a target user, thereby accurately identifying user groups with similar purchase interests and behaviors, namely user collaborative groups. Further, a target user collaborative knowledge graph is generated according to the analysis results, the target user collaborative knowledge graph intuitively displays interest relations and collaborative relations among users, and a powerful tool is provided for deep understanding of user behaviors. Based on the target user collaborative knowledge map, a collaborative interest path of each target user can be accurately determined, which means that propagation and influence paths of user interests can be clearly depicted. By utilizing the collaborative interest paths, highly personalized information pushing service can be provided for users, and the accuracy and the effectiveness of information pushing are greatly improved.

Description

User clothing order behavior analysis method and system based on big data

Technical Field

The application relates to the technical field of big data, in particular to a user clothing order behavior analysis method and system based on big data.

Background

With the rapid development of internet technology, the data volume in the electronic commerce field is explosively increased, and particularly in the clothing electronics business field, the order behavior data generated by users is particularly rich. The data not only records the purchasing behavior of the user, but also contains multidimensional information such as browsing history, evaluation content and the like of the user, and provides precious data resources for analyzing the interests and behavior patterns of the user.

However, how to effectively extract valuable information in the face of massive user clothing order behavior data and analyze the collaborative interest state and the user interest correlation degree between users, so as to provide more personalized services for users, which is a problem to be solved. Conventional related art often has difficulty coping with large-scale, complex data sets, and capturing deep-level connections between users. In addition, however, in the related art, only the purchasing behavior of the individual users is often focused, and the collaborative interests and influences among the users are ignored.

Disclosure of Invention

In view of the above-mentioned problems, in combination with the first aspect of the present application, an embodiment of the present application provides a method for analyzing a user clothing order behavior based on big data, the method including:

acquiring user clothing order behavior big data of a target user;

Determining a user collaborative interest state and a user interest correlation degree between the target users according to the user clothing order behavior big data, wherein the user collaborative interest state reflects whether a plurality of target users form a user collaborative group, and the user interest correlation degree reflects the interest correlation degree between a plurality of target users forming the user collaborative group;

Generating a target user collaborative knowledge graph according to the user collaborative interest state and the user interest correlation degree, wherein the target user collaborative knowledge graph is composed of knowledge members and knowledge links among the knowledge members, different knowledge members correspond to different target users, and the knowledge links reflect the user interest correlation degree among a plurality of target users forming the user collaborative group;

And determining a collaborative interest path of each target user based on the target user collaborative knowledge map, and pushing information to each target user based on the collaborative interest path of each target user.

In a possible implementation manner of the first aspect, the generating a target user collaborative knowledge graph according to the user collaborative interest state and the user interest correlation degree includes:

determining the number of collaborative groups of each target user according to the user collaborative interest state, wherein the number of collaborative groups represents the number of the user collaborative groups formed by the target users;

Determining knowledge member characteristics of the knowledge members corresponding to the target users according to the number of the collaborative groups, wherein the knowledge member characteristics comprise at least one of knowledge member liveness and knowledge member influence, and the knowledge member liveness and the number of the collaborative groups are in forward association;

Generating the knowledge members according to the knowledge member characteristics;

Determining the deviation degree between the knowledge members corresponding to the target users according to the user interest correlation degree between the target users, wherein the deviation degree and the user interest correlation degree are in a negative association relationship;

when the position optimization of knowledge members is completed according to the deviation degree, generating the knowledge link between the knowledge members corresponding to the target users forming the user cooperation group;

And generating the target user collaborative knowledge graph composed of the knowledge members and the knowledge links.

In a possible implementation manner of the first aspect, after the generating the target user collaborative knowledge graph according to the user collaborative interest state and the user interest correlation degree, the method includes:

Based on an enhanced expression instruction of a first target user in the target user collaborative knowledge graph, performing enhanced expression on a first knowledge member and a second knowledge member in the target user collaborative knowledge graph, wherein the first knowledge member is a knowledge member corresponding to the first target user, the second knowledge member is a knowledge member corresponding to the second target user, k-level interest links are arranged between the second target user and the first target user, and k is a positive integer;

and carrying out reinforced expression on the knowledge links between the first knowledge member and the second knowledge member and the knowledge links between the second knowledge member.

In a possible implementation manner of the first aspect, the method further includes:

performing shrinkage optimization on knowledge members other than the first knowledge member and the second knowledge member based on a selection instruction of the first target user;

Under the shrinkage condition, a target knowledge member in the second knowledge members is connected with the shrinkage knowledge members, k-level interest correlation exists between the second target user corresponding to the target knowledge members and the first target user, and the quantity of the knowledge members which are connected with the target knowledge members and are shrunk is expressed in the shrinkage knowledge members;

The method further comprises the steps of:

and based on the selected instruction of the contracted knowledge member, presenting a third knowledge member with a first-level interest correlation degree with the target knowledge member.

Acquiring a filtering requirement based on a filtering instruction of the target user collaborative knowledge graph, wherein the filtering requirement comprises at least one of a user characteristic label, a threshold user interest correlation degree and an interest correlation level;

And weakening expression of knowledge members which do not meet the filtering requirement in the target user collaborative knowledge graph.

Based on a redundant optimization instruction for the target user cooperative knowledge graph, determining the optimization parameter number of the target user cooperative knowledge graph according to the redundant optimization weight characterized by the redundant optimization instruction;

And optimizing the feature vector generated by the knowledge member and/or the knowledge link according to the optimization parameter, wherein the information which can be displayed by the knowledge member comprises a user mark, a user name and the number of collaborative groups, and the information which can be displayed by the knowledge link comprises a user interest correlation degree.

In a possible implementation manner of the first aspect, the determining, according to the user clothing order performance big data, a user collaborative interest state and a user interest correlation between the target users includes:

Determining a reference user set, wherein the reference user set consists of a plurality of target users;

Determining collaborative feature data of the reference user set according to the user clothing order behavior big data of each target user in the reference user set;

When the collaborative feature data meets a collaborative matching requirement, determining that a plurality of target users in the reference user set form the user collaborative group;

determining the user interest correlation degree of the user cooperative group according to the cooperative characteristic data;

In a possible implementation manner of the first aspect, the collaborative feature data includes a collaborative purchase rate promotion vector, a collaborative browsing rate promotion vector, and a user preference matching degree, where the collaborative purchase rate promotion vector is used to characterize an influence condition of the reference user set on a purchase rate, the collaborative browsing rate promotion vector is used to characterize an influence of the reference user set on a commodity browsing rate, and the user preference matching degree is used to characterize whether there is a similar clothing preference between at least two target users;

the method comprises the following steps:

When the collaborative purchase rate lifting vector or the collaborative browsing rate lifting vector is a negative lifting vector, determining that the collaborative matching requirement is not met; or alternatively, the first and second heat exchangers may be,

Determining to meet the collaborative matching requirement when the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors, the user preference matching degree represents similar preference, and the significant lifting vector values of the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are larger than a threshold value; or alternatively, the first and second heat exchangers may be,

When the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors and the user preference matching degree represents similar preference, but the significant lifting vector values of the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are smaller than a threshold value, determining that the collaborative matching requirement is not met; or alternatively, the first and second heat exchangers may be,

Determining to meet the collaborative matching requirement when the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors and the user preference matching degree indicates that similar preference does not exist, but a significant lifting vector value of the collaborative purchase rate lifting vector is larger than a threshold value; or alternatively, the first and second heat exchangers may be,

And when the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors, the user preference matching degree indicates that similar preference does not exist, and the significant lifting vector value of the collaborative purchase rate lifting vector is smaller than a threshold value, determining that the collaborative matching requirement is not met.

In a possible implementation manner of the first aspect, the determining the user interest relevance of the user cooperation group according to the cooperation feature data includes:

And determining the user interest correlation degree of the user collaborative group according to the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector, wherein the user interest correlation degree and the collaborative purchase rate lifting vector are in a forward association relationship, and the user interest correlation degree and the collaborative browsing rate lifting vector are in a forward association relationship.

In a possible implementation manner of the first aspect, the step of determining a collaborative interest path of each target user based on the target user collaborative knowledge graph and pushing information to each target user based on the collaborative interest path of each target user includes:

Carrying out liveness analysis on each knowledge member in the target user collaborative knowledge graph to generate liveness of each knowledge member, wherein the liveness is measured according to the interaction frequency, the order number and the evaluation frequency of the target user corresponding to each knowledge member on the clothing electronic commerce platform;

Selecting knowledge members with liveness meeting preset conditions as initial knowledge members of a collaborative interest path, starting from the initial knowledge members, expanding along a knowledge link, and primarily identifying the collaborative interest path associated with a target user, wherein the collaborative interest path consists of a plurality of knowledge members and knowledge links which are associated with each other, and reflecting interest transfer and influence relation among the target users;

Performing weight assignment on the preliminarily identified collaborative interest paths based on the interest correlation among target users, the similarity of purchasing behavior and the consistency of user evaluation, and optimizing and pruning the collaborative interest paths according to the assigned weights;

In the optimized and pruned collaborative interest path, identifying key knowledge members, and acquiring the position characteristics of each key knowledge member in the collaborative interest path, wherein the position characteristics comprise the specific position of each key knowledge member in the collaborative interest path and the path cost relation between each key knowledge member and other knowledge members in the collaborative interest path;

identifying character features of the key knowledge members in the collaborative interest path according to user behaviors and collaborative influence of the key knowledge members in the collaborative interest path, wherein the character features comprise one of leading characters, transferring characters and following characters;

And mining target interest preference data from the collaborative behavior data between each key knowledge member and other knowledge members according to the position characteristics and the role characteristics of each key knowledge member in the collaborative interest path, and generating personalized push content in collaborative association with other target users according to the target interest preference data.

In yet another aspect, an embodiment of the present application further provides a system for analyzing a user's clothing order behavior based on big data, including a processor, and a machine-readable storage medium, where the machine-readable storage medium is connected to the processor, and the machine-readable storage medium is used to store a program, an instruction, or a code, and the processor is used to execute the program, the instruction, or the code in the machine-readable storage medium, so as to implement the method described above.

Based on the above aspects, the embodiment of the application can deeply analyze the collaborative interest state and the user interest correlation degree between users by acquiring the user clothing order behavior big data of the target user, thereby accurately identifying the user group with similar purchase interests and behaviors, namely the user collaborative group. Further, a target user collaborative knowledge graph is generated according to the analysis results, the target user collaborative knowledge graph intuitively displays interest relations and collaborative relations among users, and a powerful tool is provided for deep understanding of user behaviors. Based on the target user collaborative knowledge map, a collaborative interest path of each target user can be accurately determined, which means that propagation and influence paths of user interests can be clearly depicted. By utilizing these collaborative interest paths, highly personalized information push services can be provided to users, such as recommending clothing styles or trending trends to users that highly match their interests. The pushing mode based on the collaborative interest path not only greatly improves the accuracy and the effectiveness of information pushing, but also remarkably improves the user experience and the satisfaction. Therefore, by generating the target user collaborative knowledge map and determining the collaborative interest path, powerful data support is provided for the clothing electronic commerce platform, and the platform is helped to realize more accurate user portraits and personalized service strategies, so that the competitiveness and user viscosity of the platform are improved.

Drawings

Fig. 1 is a schematic diagram of an execution flow of a user clothing order behavior analysis method based on big data according to an embodiment of the present application.

Fig. 2 is a schematic diagram of a hardware architecture of a customer clothing order behavior analysis system based on big data according to an embodiment of the present application.

Detailed Description

The present application is specifically described below with reference to the accompanying drawings, and fig. 1 is a schematic flow chart of a method for analyzing customer clothing order behavior based on big data according to an embodiment of the present application, and the method for analyzing customer clothing order behavior based on big data is described in detail below.

Step S110, user clothing order behavior big data of a target user are obtained.

In this embodiment, the server is first connected to the database of the clothing e-commerce platform, and extracts all clothing order data of the target user on the clothing e-commerce platform, where the clothing order data includes, but is not limited to, clothing items browsed by the user, commodities added to shopping carts, commodities purchased in order, order amount, purchase time, receiving address, and the like. In addition, the method also comprises the evaluation of the commodity by the user, the return record and the interaction record with customer service, wherein the clothing order data form the basis of the large data of the clothing order behavior of the user, and a rich information source is provided for the subsequent analysis of the interests and the cooperative behavior of the user.

For example, the server obtains all clothing order records for user A, B, C, etc. over the past year from the database of the e-commerce platform, which shows that user A often purchases casual style clothing, user B prefers business forward, and user C is more interested in fashion clothing.

Step S120, determining a user collaborative interest state and a user interest correlation degree between the target users according to the user clothing order behavior big data, wherein the user collaborative interest state reflects whether a plurality of target users form a user collaborative group, and the user interest correlation degree reflects the interest correlation degree between a plurality of target users forming the user collaborative group.

In detail, the user collaborative interest state refers to determining whether a user collaborative group is formed by analyzing clothing order behavior data of users and judging whether similar interests and behavior patterns exist among a plurality of target users. The user interest correlation is used to quantify the similarity of interests between users in the collaborative group. For example, after analyzing the clothing order performance data of user A, B, C, the server finds that user A and user B have similar purchase records and ratings on certain specific categories (e.g., outdoor sports equipment), thus judging that there is a collaborative interest between them and grouping them into a user collaborative group. Meanwhile, the server calculates the interest correlation degree of the users according to the purchase frequency, the purchase amount, the evaluation consistency and other factors, for example, the interest correlation degree can be a numerical value between 0 and 1, and the higher the numerical value is, the higher the interest correlation degree is.

In this example, the interest relevance of user A and user B may be high, as they often purchase similar outdoor sports equipment, and the ratings are similar. While user C may also be interested in fashion clothing, the interest correlation between them may be low due to the large variance in purchase records and ratings from user A, B.

That is, in this embodiment, the server uses big data analysis techniques to mine the user's clothing order behavior big data deep. By comparing purchase histories, browsing records and evaluation contents of different users, the server can identify user groups with similar interests, wherein the user groups are user cooperation groups, and target users have higher interest correlation degree.

For example, server analysis identified that although user A and user B were different in daily purchasing style, they all purchased the same type of clothing during certain specific activities (e.g., outdoor sports). Thus, the server determines that user A and user B have a collaborative interest in outdoor athletic apparel and groups them into a collaborative group of users. Meanwhile, the server calculates the user interest correlation between them according to their purchase frequency and amount.

And step S130, generating a target user collaborative knowledge graph according to the user collaborative interest state and the user interest correlation degree, wherein the target user collaborative knowledge graph is composed of knowledge members and knowledge links among the knowledge members, wherein different knowledge members correspond to different target users, and the knowledge links reflect the user interest correlation degree among a plurality of target users composing the user collaborative group.

In this embodiment, based on the user collaborative interest state and the user interest correlation determined in the previous step, the server starts to construct the target user collaborative knowledge graph. In the target user collaborative knowledge graph, each knowledge member represents a target user, and the knowledge links among the knowledge members reflect the interest correlation among the users. The thickness or shade of the knowledge link may be used to represent the strength of interest correlation.

For example, in the target user collaborative knowledge graph, users A, B and C are represented as three knowledge members. Because user a and user B have a synergistic interest in outdoor sports wear, a thicker knowledge link is formed between them. And the interest correlation degree of the user C and other two users is low, so that the knowledge link between the user C and the user B is relatively thin.

And step S140, determining a collaborative interest path of each target user based on the target user collaborative knowledge map, and pushing information to each target user based on the collaborative interest path of each target user.

In detail, the collaborative interest path is determined based on a collaborative knowledge graph of target users, reflects a specific path of interest propagation and influence between users, and reveals how users are affected by other users and how such influence is transferred between users. For example, suppose user A often knows new outdoor sports equipment information through user B and purchases similar equipment under the influence of user B. Then a collaborative interest path is formed from user B to user a. The server may determine such paths by analyzing knowledge members and knowledge links in the target user collaborative knowledge graph and customize personalized information push policies for each user based on these paths. For example, when user B browses or purchases new outdoor sports equipment, the server may automatically push related information to user a to improve accuracy and user satisfaction of information push.

That is, in this embodiment, the server determines a collaborative interest path for each target user by analyzing knowledge members and knowledge links in the target user collaborative knowledge graph, and these collaborative interest paths reveal propagation and influence relationships of interests between users. Based on these collaborative interest paths, the server may customize a personalized information push policy for each user. For example, the server recognizes that user a often knows about new style of outdoor athletic clothing through user B. Therefore, when the user B browses or purchases the new outdoor sportswear, the server can automatically push related information to the user A, and the pushing mode based on the collaborative interest path greatly improves the accuracy of information pushing and the satisfaction of the user.

Based on the steps, the user clothing order behavior big data of the target user are obtained, and the collaborative interest state and the user interest correlation degree among the users can be deeply analyzed, so that the user group with similar purchase interests and behaviors, namely the user collaborative group, can be accurately identified. Further, a target user collaborative knowledge graph is generated according to the analysis results, the target user collaborative knowledge graph intuitively displays interest relations and collaborative relations among users, and a powerful tool is provided for deep understanding of user behaviors. Based on the target user collaborative knowledge map, a collaborative interest path of each target user can be accurately determined, which means that propagation and influence paths of user interests can be clearly depicted. By utilizing these collaborative interest paths, highly personalized information push services can be provided to users, such as recommending clothing styles or trending trends to users that highly match their interests. The pushing mode based on the collaborative interest path not only greatly improves the accuracy and the effectiveness of information pushing, but also remarkably improves the user experience and the satisfaction. Therefore, by generating the target user collaborative knowledge map and determining the collaborative interest path, powerful data support is provided for the clothing electronic commerce platform, and the platform is helped to realize more accurate user portraits and personalized service strategies, so that the competitiveness and user viscosity of the platform are improved.

In one possible implementation, step S130 includes:

Step S131, determining the number of collaborative groups of each target user according to the user collaborative interest state, where the number of collaborative groups represents the number of collaborative groups of the target user.

Step S132, determining knowledge member characteristics of the knowledge members corresponding to the target users according to the collaborative group quantity, wherein the knowledge member characteristics comprise at least one of knowledge member liveness and knowledge member influence, and the knowledge member liveness and the collaborative group quantity are in a forward association relationship.

And step S133, generating the knowledge members according to the knowledge member characteristics.

Step S134, determining the deviation degree between the knowledge members corresponding to the target users according to the user interest correlation degree between the target users, wherein the deviation degree and the user interest correlation degree are in a negative association relationship.

And step S135, generating the knowledge link between the knowledge members corresponding to the target users forming the user cooperation group when the knowledge member position optimization is completed according to the deviation degree.

And step S136, generating the target user collaborative knowledge graph composed of the knowledge members and the knowledge links.

In this embodiment, the server first analyzes the user collaborative interest state, which is based on the user clothing order behavior big data extracted from the clothing e-commerce platform database in the previous step, and these data reveal which users have similar shopping interests and behaviors. Based on these collaborative interest states, the server determines the number of collaborative groups each target user participates in.

For example, suppose that the server recognizes that user A participates in 3 different collaborative groups formed around "outdoor sports equipment", "business forward", and "fashion apparel", respectively. User B participates in 2 collaborative groups: "outdoor sports equipment" and "fashion accessories". While user C is only engaged in 1 collaborative group: "fashion trend apparel".

Next, the server determines features of the corresponding knowledge members based on the number of collaborative groups each target user participates in, the features mainly including knowledge member liveness and knowledge member influence. If the user participates in a plurality of cooperative groups, the user has a hunting in different interesting fields, so that the knowledge member liveness is relatively high.

For example, since user a is engaged in 3 collaborative groups, its knowledge member liveness is marked as "high". User B participates in 2 groups, whose liveness is marked as "medium". While user C is engaged in only 1 group, its liveness is marked as "low".

Based on the knowledge member characteristics determined in the previous step, the server starts to generate knowledge members. Each knowledge member represents a target user and carries characteristic information of the user, such as liveness, influence, and the like.

For example, the server generates three knowledge members for the user A, B, C, respectively, each of which contains a corresponding liveness flag and other relevant information.

The server continues to analyze the user interest correlations between the target users by comparing their similarities in clothing order behavior. Based on this correlation, the server calculates the degree of deviation between knowledge members. The deviation degree is an index reflecting the interest difference among knowledge members and is negatively correlated with the interest correlation degree of the user.

For example, user A and user B have a high degree of interest relevance in the "outdoor sports equipment" group, and therefore the degree of deviation between their corresponding knowledge members is low. While user a and user C have intersections in the "fashion trend apparel" group, their overall interest relevance is not as high as a and B, and therefore the degree of deviation between their corresponding knowledge members is relatively high.

After determining the degree of deviation, the server begins to optimize the location of knowledge members, taking into account the degree of deviation and the degree of closeness within the collaborative group. After the optimization is completed, the server generates knowledge links between knowledge members corresponding to the target users forming the collaborative group, and the links intuitively display interest links between the users.

For example, during the optimization process, the server may adjust the locations of knowledge members such that knowledge members with higher interest-related degrees are closer together in the target user collaborative knowledge graph. For example, the knowledge members corresponding to user A and user B may be placed in relatively close proximity and connected by a thicker knowledge link to reflect the high degree of interest correlation between them.

Finally, the server generates a target user cooperative knowledge graph composed of knowledge members and knowledge links, and the graph intuitively displays the cooperative relationship and interest relation among the target users.

For example, in the finally generated collaborative knowledge graph of the target users, the users A, B, C are respectively represented by three knowledge members, and are connected through knowledge links, and the thickness and the color of the knowledge links may be different due to different interest relatedness between the users, so that an intuitive view is provided to understand the collaborative interest relationship between the users.

In a possible implementation manner, after the step S130, the method includes:

Step A110, based on a reinforced expression instruction of a first target user in the target user collaborative knowledge graph, performing reinforced expression on a first knowledge member and a second knowledge member in the target user collaborative knowledge graph, wherein the first knowledge member is a knowledge member corresponding to the first target user, the second knowledge member is a knowledge member corresponding to the second target user, k-level interest connection is arranged between the second target user and the first target user, and k is a positive integer.

And step A120, carrying out reinforced expression on the knowledge links between the first knowledge member and the second knowledge member and the knowledge links between the second knowledge member.

In this embodiment, the server receives an enhanced expression instruction, where the enhanced expression instruction specifies that the first target user (assumed to be user a) in the target user collaborative knowledge graph is to be enhanced. According to the reinforced expression instruction, the server first identifies the knowledge member corresponding to the user A, namely the knowledge member A. At the same time, the server also identifies a knowledge member, namely knowledge member B, corresponding to the second target user (assumed to be user B) with k-level interest contact for user a.

In this scenario, a k-level interest connection may be understood as a connection between user A and user B through k intermediary users or k interest links. The server may determine such a relationship by analyzing knowledge links in the target user collaborative knowledge graph.

Once a knowledge member is identified that requires an enhanced expression, the server may begin performing operations to enhance the expression, which may include increasing the visual prominence of the knowledge member (e.g., changing color, size, or shape), or adding additional labels or information next to the knowledge member to emphasize its importance or specificity.

After the enhanced expression of knowledge members a and B is completed, the server may continue to perform the enhanced expression of knowledge links between the two knowledge members, which means that the server may particularly highlight knowledge links connecting knowledge members a and B to emphasize their connections.

In addition, if knowledge links exist between knowledge member B and other knowledge members (assuming knowledge member C corresponding to user C), the server will also express these links in an enhanced manner in order to demonstrate the broader association of user B in the user collaborative group and how these associations are associated with user a.

Methods of enhancing the expression of knowledge links may include using thicker or brighter lines to represent the links, or adding additional instructions or labels alongside the links so that the user can more easily notice the enhanced links when looking at the target user collaborative knowledge graph, thereby better understanding the collaborative interests and connections between users.

Therefore, key users and relations in the target user collaborative knowledge graph are highlighted by strengthening expression of specific knowledge members and knowledge links, so that users are helped to more intuitively understand the collaborative relations and interest relations among the users.

In one possible embodiment, the method further comprises: and performing shrinkage optimization on knowledge members except the first knowledge member and the second knowledge member based on the selected instruction of the first target user.

Under the shrinkage condition, a target knowledge member in the second knowledge members is connected with the shrinkage knowledge members, k-level interest correlation degrees exist between the second target users corresponding to the target knowledge members and the first target users, and the quantity of the knowledge members which are connected with the target knowledge members and are shrunk is expressed in the shrinkage knowledge members.

The method further comprises the steps of:

In this embodiment, the server receives a selection instruction for a first target user (e.g., user a), where the selection instruction may be sent by the user through an interface operation, and indicates that the user wants to view the collaboration relationship related to user a more clearly.

After receiving the selected instruction, the server performs shrinkage optimization on the collaborative knowledge graph of the target user. In particular, the server may leave the first knowledge member (representing user a) and the second knowledge member (representing user B having direct interest in contact with user a, such as user B) unchanged, while contracting the other knowledge members, which might appear to simplify the presentation of those knowledge members on the graph, or to aggregate them into a contracted node, to reduce the complexity of the graph and highlight the user directly related to user a.

For example, if the original map contains knowledge members corresponding to users C, D, E, etc. with low relevance to the interests of user a, then after shrink optimization, these knowledge members may be reduced to a shrink node to save space and highlight core relationships.

During the shrink optimization, the server may take special care of the connection between the second knowledge member (i.e. the knowledge member corresponding to the user with which user a has direct interest in contact) and the shrink knowledge member. If a certain second knowledge member (e.g., a knowledge member representing user B) has a connection relationship with a certain knowledge member in the contracted node (e.g., a knowledge member representing user C) originally, such connection relationship still needs to be preserved and displayed after contraction.

The server can determine which connection relations are important through data analysis, and clearly show the relations in the contracted map, so that a user can clearly see a user B which has direct interest in contact with the user A and the relations between the user B and other users in the contracted nodes.

When a user wants to further understand the relationship between a user in the contracted node (e.g., user C) and a direct interest contact user of user A (e.g., user B), knowledge members in the contracted node that represent user C may be selected through interface operations. Upon receiving this selection instruction, the server expands the contracted node and presents other knowledge members (referred to herein as third knowledge members) that have a first degree of interest correlation with user C.

These third knowledge members may be knowledge members corresponding to other users having direct collaborative interests with respect to user C. The server can determine the relationships through data analysis and clearly show the relationships in the map, so that the user can intuitively see the relationships between the user C and the user B and the interest correlation degree between the user C and the user B.

Thus, by shrinking and expanding specific parts in the knowledge graph, the user is helped to more clearly understand and explore the collaborative interests and connections between target users.

In one possible embodiment, the method further comprises:

And step B110, obtaining a filtering requirement based on a filtering instruction of the target user collaborative knowledge graph, wherein the filtering requirement comprises at least one of a user characteristic label, a threshold user interest correlation degree and an interest correlation level.

And step B120, weakening expression of knowledge members which do not meet the filtering requirement in the target user collaborative knowledge graph.

In this embodiment, the server receives a filtering instruction for the collaborative knowledge graph of the target user, where the filtering instruction may be from an operation of the user interface, and the user wants to view specific information in the graph more accurately through filtering. Upon receiving the filtering instruction, the server parses and obtains specific filtering requirements, where the filtering requirements may include at least one of a user feature tag (e.g., age, gender, geographic location, etc.), a threshold user interest correlation (i.e., a minimum criterion for interest correlation between users), and an interest correlation hierarchy (indicating a depth or breadth of interest correlation between users).

For example, a user may want to view a user population between ages 25-35 that has an interest-related degree with user A that is higher than 0.8. In this case, the server may obtain these specific filtering requirements.

After the filtering requirements are obtained, the server can traverse the collaborative knowledge graph of the target user to identify knowledge members which do not meet the filtering requirements, and the interest correlation degree of the users possibly represented by the knowledge members and the user A is lower than a set threshold value, or the user characteristic labels of the knowledge members and the appointed labels are not matched, or the knowledge members and the user characteristic labels are not in a user set interest correlation level.

For those knowledge members that do not meet the filtering requirements, the server may do the de-emphasis. The manner in which expression is impaired may be to reduce the visible prominence of these knowledge members in the atlas (e.g., reduce node size, fade color, etc.), or to temporarily hide them from the atlas, in order to allow the user to more clearly see the knowledge members meeting the filtering requirements and their relationships, and thus to more easily obtain and analyze the desired information.

For example, if the server finds that user B is connected to user a, but the age of user B is outside the range set by the user, the server weakens the expression of user B in the map so that the user can concentrate more on other users who meet his filtering criteria.

In one possible embodiment, the method further comprises:

and step C110, determining the optimization parameter number of the target user collaborative knowledge graph according to the redundancy optimization weight represented by the redundancy optimization instruction based on the redundancy optimization instruction of the target user collaborative knowledge graph.

And step C120, optimizing the feature vectors generated by the knowledge members and/or the knowledge links according to the optimization parameter, wherein the information which can be displayed by the knowledge members comprises user marks, user names and the number of collaborative groups, and the information which can be displayed by the knowledge links comprises user interest correlation degrees.

In this embodiment, the server may receive a redundancy optimization instruction for the collaborative knowledge graph of the target user, where the purpose of the redundancy optimization instruction is to reduce redundancy information in the graph and improve efficiency and accuracy of the graph.

The redundancy optimization instruction includes a redundancy optimization weight that indicates how much redundancy should be reduced by the server during the optimization process. The server may determine specific optimization parameters based on this weight, which may include the number of redundant knowledge members that need to be deleted or merged, the number of knowledge links that need to be simplified, etc.

For example, if the redundancy optimization weight is set higher, the server may determine that more redundant knowledge members need to be deleted and more knowledge links simplified to achieve a higher optimization effect.

After determining the optimization parameters, the server may begin optimizing the knowledge members and/or knowledge links, which may include deleting redundant knowledge members, merging similar knowledge members, simplifying complex knowledge links, and so forth.

During the optimization process, the server may recalculate and optimize feature vectors generated by knowledge members and knowledge links. A feature vector is a set of values representing knowledge members or knowledge link characteristics that contains information about user identity, user name, number of co-groups (for knowledge members), and user interest relevance (for knowledge links).

For example, for knowledge members, the server may delete or merge some redundant users based on the optimization parameters and recalculate the feature vectors of the remaining users to ensure that they reflect the user's characteristics more accurately. For knowledge links, the server may simplify some complex links and recalculate feature vectors for the links to more accurately represent the interest correlation between users.

After the optimization is completed, the server can update the collaborative knowledge graph of the target user, so that the collaborative knowledge graph is simpler and more efficient, and analysis and decision-making of the user can be better supported.

In one possible implementation, step S120 may include:

step S121, determining a reference user set, where the reference user set is composed of a plurality of the target users.

Step S122, determining collaborative feature data of the reference user set according to the user clothing order behavior big data of each target user in the reference user set.

Step S123, when the collaborative feature data meets a collaborative matching requirement, determining that a plurality of the target users in the reference user set form the user collaborative group.

Step S124, determining the user interest correlation degree of the user collaboration group according to the collaboration feature data.

In this embodiment, the server will first screen out a part of users from the massive user data as target users, which are usually active users, i.e. users with a certain amount of clothing order behavior data. The server may then further pick a group of users from the target users according to specific conditions or algorithms, such as the frequency of the users' purchases, the categories of purchases, etc., to form a reference user set.

For example, the server may select users who purchased clothing at least three times in the past month, forming a reference user set containing 100 users.

The server will then analyze the garment order performance big data for each target user in the reference user set, which may include the user's purchase time, purchase category, purchase price, purchase frequency, etc. Through in-depth analysis and mining of these data, the server is able to extract common features of these users, namely collaborative feature data.

For example, the server finds that 80 of these 100 users purchased winter thermal jackets in the last month, and most of them opt to place orders in the evening, which is a significant synergy feature.

The server may preset some collaborative matching requirements, such as that the collaborative feature data needs to reach a certain level of significance, or that the number of collaborative users needs to exceed a certain threshold, etc. When the extracted collaborative feature data meets the requirements, the server may determine that a collaborative interest state exists between the users.

For the previous example, if the server sets a cooperative matching requirement that at least 70% of the users have the same purchasing behavior, the cooperative feature of purchasing winter thermal jackets by 80 users meets the cooperative matching requirement.

Once the collaborative feature data meets the collaborative matching requirement, the server forms the target users with collaborative interest states into a user collaborative group. Meanwhile, the server calculates the user interest correlation degree of the user collaboration group according to the factors such as the significance and consistency of the collaboration feature data.

In the above example, the 80 users having the same purchase behavior are grouped into a user collaborative group, and their user interest correlation degree may be calculated according to factors such as specific purchase category, purchase time, etc., for example, a value between 0 and 1 may be provided, and a higher value indicates a greater interest correlation degree between users.

Therefore, the server can accurately determine the collaborative interest state and the user interest correlation degree between target users according to the big data of the clothing order behaviors of the users, and powerful data support is provided for follow-up personalized recommendation, marketing and the like.

In one possible implementation manner, the collaborative feature data includes a collaborative purchase rate promotion vector, a collaborative browsing rate promotion vector, and a user preference matching degree, where the collaborative purchase rate promotion vector is used to characterize an influence condition of the reference user set on a purchase rate, the collaborative browsing rate promotion vector is used to characterize an influence of the reference user set on a commodity browsing rate, and the user preference matching degree is used to characterize whether a similar clothing preference exists between at least two target users.

The method comprises the following steps:

and when the collaborative purchase rate lifting vector or the collaborative browsing rate lifting vector is a negative lifting vector, determining that the collaborative matching requirement is not met. Or alternatively, the first and second heat exchangers may be,

And determining to meet the collaborative matching requirement when the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors, the user preference matching degree represents similar preference, and the significant lifting vector value of the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector is larger than a threshold value. Or alternatively, the first and second heat exchangers may be,

And when the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors and the user preference matching degree representation has similar preference, determining that the collaborative matching requirement is not met when the significant lifting vector values of the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are smaller than a threshold value. Or alternatively, the first and second heat exchangers may be,

And when the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors, and the user preference matching degree representation does not have similar preference, but the significant lifting vector value of the collaborative purchase rate lifting vector is larger than a threshold value, determining to meet the collaborative matching requirement. Or alternatively, the first and second heat exchangers may be,

In this embodiment, the method involves detailed analysis of collaborative feature data to determine whether collaborative matching requirements are met, where the collaborative feature data includes a collaborative purchase rate boost vector, a collaborative browse rate boost vector, and a user preference matching degree. The following are detailed scenarios for each case:

case one: the collaborative purchase rate boost vector or the collaborative browse rate boost vector is a negative boost vector

The server analyzes a reference user set consisting of 100 users. During the observation period, the cooperative buying rate of the user set for a new one-piece dress on the market is improved by-5%, and the cooperative browsing rate is improved by-3%, which means that the frequency of buying and browsing the one-piece dress is reduced compared with other user groups.

Because the co-purchase rate boost vector and the co-browse rate boost vector are both negative boost vectors, the server determines that the set of users does not meet the co-match requirement, which may be because the dress does not meet the general aesthetic or demand of the set of users.

And a second case: the collaborative purchase rate promotion vector and the collaborative browsing rate promotion vector are forward promotion vectors, the user preference matching degree indicates that similar preference exists, and the significant promotion vector value is larger than the threshold value

The server observes another reference user set consisting of 200 users loving for outdoor exercises. When a new outdoor jacket is marketed in spring, the collaborative purchase rate lifting vector of the user set is 15%, the collaborative browsing rate lifting vector is 12%, and the user preference matching degree shows that the users are generally highly interested in outdoor sports equipment.

Because the collaborative purchase rate promotion vector and the collaborative browse rate promotion vector are both forward promotion vectors, and the user preference matching degree indicates that similar preference exists, and the significant promotion vector values exceed a threshold value (for example, 10%) set by the server, the server determines that the user set meets the collaborative matching requirement, which indicates that the new outdoor jacket meets the interests and requirements of the user set very well.

And a third case: the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors, but the significant lifting vector value is smaller than the threshold value

For a reference user set of 150 young female users, the server notices a collaborative purchase rate boost vector of 5% and a collaborative browse rate boost vector of 4% for a newly released fashion handbag. Although these are forward boost vectors and the user preference matching degree shows some interest in the user's fashion, these significant boost vector values do not reach the threshold value of the server (e.g., 8%).

Although the set of users has shown some increased interest in the handbag, because the significantly boost vector value is less than the threshold value, the server determines that the set of users does not meet the collaborative matching requirement, which may be because the design or price of the handbag does not fully capture the popular preferences of the set of users.

Case four and case five: user preference matching characterizes a determination when no similar preferences exist

The two cases mainly consider that even though the collaborative purchase rate boost vector and the collaborative browse rate boost vector are forward boost vectors, if there is no similar clothing preference between users, the judgment of the collaborative matching requirement will mainly depend on whether the significant boost vector value of the collaborative purchase rate boost vector is greater than the threshold value.

Scenario example (case four):

A reference user set consisting of users of different ages and professional contexts shows a forward collaborative purchase rate boost vector (10%) and a collaborative browse rate boost vector (7%) when purchasing a multi-function wristwatch, but the user preference matches show that there is no obvious similar clothing or accessory preference between them. However, since the significant boost vector value of the collaborative purchase rate boost vector exceeds the threshold value (e.g., 8%), the server still determines that this set of users meets the collaborative matching requirement.

Scenario example (case five):

Another diverse set of users exhibited forward but smaller collaborative purchase rate boost vectors (3%) and collaborative browse rate boost vectors (2%) when a new style of athletic shoe was marketed, with no obvious similarity preference between users. Because the significant boost vector value of the collaborative purchase rate boost vector does not reach the threshold value, the server determines that the set of users does not meet the collaborative matching requirement.

In one possible implementation, step S124 includes: and determining the user interest correlation degree of the user collaborative group according to the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector, wherein the user interest correlation degree and the collaborative purchase rate lifting vector are in a forward association relationship, and the user interest correlation degree and the collaborative browsing rate lifting vector are in a forward association relationship.

In this embodiment, the server first collects and sorts the data of the collaborative purchase rate promotion vector and the collaborative browsing rate promotion vector, where the data is obtained by analyzing the big data of the user's clothing order, and reflects the overall trend of the user collaborative group in purchasing and browsing clothing.

For example, suppose that the server has determined a collaborative group of 100 users by previous steps, the users in the group have significantly improved purchase and browse rates for certain types of outdoor sports clothing over the past month. The server collects purchase and browsing data of the users in the period of time, and calculates a cooperative purchase rate promotion vector and a cooperative browsing rate promotion vector according to the purchase and browsing data.

Next, the server may calculate a user interest correlation of the user collaborative group according to the collaborative purchase rate boost vector and the collaborative browse rate boost vector. Because the user interest correlation is in a forward association with the collaborative purchase rate promotion vector and the collaborative browsing rate promotion vector, when the two vectors are increased, the user interest correlation is also increased.

For example, the server uses a particular algorithm (e.g., weighted summation, machine learning model, etc.) to combine the co-purchase rate boost vector and the co-browse rate boost vector to derive the user interest relevance. For example, if the collaborative purchase rate boost vector is 10% and the collaborative browse rate boost vector is 8%, the server may assign different weights (e.g., purchase behavior is more important and thus higher) to the two vectors and then calculate an integrated user interest correlation.

Finally, the server may determine a specific value for the relevance of the user's interests, which value will be used for subsequent personalized services such as recommendation, advertisement placement, etc. For example, the server obtains that the user interest correlation degree of the user cooperative group is 0.85 (the value range of the user interest correlation degree is 0 to 1 is assumed, wherein 1 indicates that the interests are completely correlated), and the value indicates that the user cooperative group has higher interests in outdoor sports and is an important reference for subsequent personalized services.

Therefore, by collecting the collaborative feature data, calculating the interest correlation degree of the user and determining the specific value thereof, the interest preference of the collaborative group of the user can be more accurately understood, and thus more accurate personalized service can be provided.

In one detailed description, the user collaborative interest state is determined by analyzing user clothing order performance big data, based mainly on collaborative feature data. The collaborative feature data includes a collaborative purchase rate boost vector, a collaborative browse rate boost vector, and a user preference match.

Collaborative purchase rate boost vector: representing the impact of the user collaboration group on the purchase rate. The calculation formula may be: collaborative purchase rate boost vector = (average purchase rate of users within collaborative group-average purchase rate of total users)/average purchase rate of total users × 100%.

Collaborative browsing rate boost vector: the influence of the user collaboration group on the commodity browsing rate is represented. The calculation formula may be: collaborative browsing rate boost vector = (average browsing rate of users within collaborative group-average browsing rate of total users)/average browsing rate of total users × 100%.

User preference matching degree: may be calculated by comparing the clothing items, styles, colors, etc. purchased and viewed by the user. The specific algorithm may be cosine similarity, pearson correlation coefficient, etc. for quantifying interest similarity between users.

By integrating the three pieces of collaborative feature data, whether the user collaborative group meets the collaborative matching requirement can be judged based on a series of rules of the example, so that the user collaborative interest state is determined. For example, when the collaborative matching requirement is met, the state value corresponding to the user collaborative interest state is 1, and when the collaborative matching requirement is not met, the state value corresponding to the user collaborative interest state is 0.

The user interest correlation degree is used for quantifying the interest degree of the user in the user collaborative group in a specific clothing class or style, and is calculated mainly according to the collaborative purchase rate promotion vector and the collaborative browsing rate promotion vector.

One possible calculation formula is:

User interest correlation = w1 x collaborative purchase rate boost vector + w2 x collaborative browse rate boost vector

Wherein w1 and w2 are weight coefficients, which can be adjusted according to practical situations. For example, if the purchase behavior is considered to reflect user interest more than the browse behavior, w1 > w2 may be set.

The formula is used for carrying out weighted summation on the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector to obtain a comprehensive user interest correlation index. The higher the index value of this user interest correlation, the more interesting the user collaboration group is for a particular clothing class or style.

In one possible implementation, step S140 includes:

And step S141, carrying out liveness analysis on each knowledge member in the target user collaborative knowledge graph to generate liveness of each knowledge member, wherein the liveness is measured according to the interaction frequency, the order number and the evaluation times of the target user corresponding to each knowledge member on the clothing e-commerce platform.

In this embodiment, the server first performs liveness analysis on each knowledge member in the collaborative knowledge graph of the target user. The liveness is measured according to indexes such as interaction frequency, order number, evaluation frequency and the like of target users corresponding to each knowledge member on the clothing e-commerce platform.

For example, assuming that knowledge member a has a high frequency of interaction, a large number of orders, and a frequent number of evaluations in the past month, the server may determine that knowledge member a has a high liveness. In contrast, knowledge member B is relatively low in liveness if it appears flat on these indicators.

Step S142, selecting knowledge members with liveness meeting preset conditions as initial knowledge members of a collaborative interest path, starting from the initial knowledge members, expanding along a knowledge link, and primarily identifying the collaborative interest path associated with a target user, wherein the collaborative interest path is composed of a plurality of knowledge members and knowledge links which are associated with each other, and the interest transfer and influence relationship among the target users are reflected.

In this embodiment, the server selects a knowledge member whose activity meets a preset condition as the initial knowledge member of the collaborative interest path. Starting from the initial knowledge members, expanding along the knowledge links, and primarily identifying collaborative interest paths associated with target users.

For example, with knowledge member A having high activity as a starting point, the server finds that A frequently interacts with knowledge member C, D, E to form a collaborative interest path A-C-D-E, which reflects interest transfer and influence relationships between target users.

And step S143, carrying out weight assignment on the preliminarily identified collaborative interest paths based on the interest correlation degree between target users, the similarity of purchasing behavior and the consistency of user evaluation, and optimizing and pruning the collaborative interest paths according to the assigned weights.

In this embodiment, the server performs weight assignment on the preliminarily identified collaborative interest path based on the interest correlation between target users, the similarity of purchase behaviors, and the consistency of user evaluation. And optimizing and pruning the collaborative interest path according to the assigned weight.

For example, in path A-C-D-E, if C is highly correlated with A's interests, the purchasing behavior is similar, and the user's ratings are consistent, then C will get a higher weight. Conversely, if E performs poorly in these respects, its weight will be relatively low. According to the weight, the server may prune the path, e.g. remove E with lower weight, and the optimized path becomes a-C-D.

And step S144, identifying key knowledge members in the optimized and pruned collaborative interest path, and acquiring the position characteristics of each key knowledge member in the collaborative interest path, wherein the position characteristics comprise the specific position of each key knowledge member in the collaborative interest path and the path cost relation between each key knowledge member and other knowledge members in the collaborative interest path.

In this embodiment, in the collaborative interest path after optimization and pruning, the server identifies key knowledge members, and obtains the position feature of each key knowledge member in the collaborative interest path.

For example, in path A-C-D, C may be a key knowledge member, as it connects A and D, functioning as a bridge. The location characteristics of C include its intermediate position in the path, and path cost relationships (e.g., distance, number of interactions, etc.) with a and D.

Step S145, identifying character features of the key knowledge members in the collaborative interest path according to the user behaviors and collaborative influence of the key knowledge members in the collaborative interest path, wherein the character features comprise one of leading characters, transferring characters and following characters.

In this embodiment, the server identifies character features in the collaborative interest path according to the user behavior and collaborative impact of the key knowledge members in the collaborative interest path.

For example, if C often leads to new shopping trends and affects purchase decisions of other users, C may play a leading role. If C is primarily a transfer message, diffusing the effect of A to D, it may play a transfer role. If C is primarily a shopping trend that follows A, it may play a following role.

And step S146, mining target interest preference data from the collaborative behavior data between each key knowledge member and other knowledge members according to the position characteristics and the role characteristics of each key knowledge member in the collaborative interest path, and generating personalized push content in collaborative association with other target users according to the target interest preference data.

Finally, the server extracts target interest preference data from the collaborative behavior data between the key knowledge members and other knowledge members according to the position characteristics and the role characteristics of each key knowledge member in the collaborative interest path. From this data, personalized push content is generated that is cooperatively associated with other target users.

For example, if C plays a leading role and its location characteristics indicate that it has an important impact on D, then the server may push to D a similar fashion or trend of the trend of clothing to C, such pushing taking into account both the personal interests of D and the leading role of C, improving the accuracy and effectiveness of the pushing.

Fig. 2 illustrates a hardware structural intent of the big data based user clothing order performance analysis system 100 for implementing the big data based user clothing order performance analysis method according to an embodiment of the application, as shown in fig. 2, the big data based user clothing order performance analysis system 100 may include a processor 110, a machine readable storage medium 120, a bus 130, and a communication unit 140.

In one possible design, the big data based user apparel order behavior analysis system 100 may be a single server or a group of servers. The server set may be centralized or distributed (e.g., big data based user apparel order behavior analysis system 100 may be a distributed system). In some embodiments, big data based user apparel order behavior analysis system 100 may be local or remote. For example, big data based user apparel order behavior analysis system 100 may access information and/or data stored in machine-readable storage medium 120 via a network. As another example, big data based user apparel order performance analysis system 100 may be directly connected to machine-readable storage medium 120 to access stored information and/or data. In some embodiments, the big data based user apparel order behavior analysis system 100 may be implemented on a big data based user apparel order behavior analysis system. For example only, the big data based user clothing order behavior analysis system may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distribution cloud, an internal cloud, a multi-layer cloud, or the like, or any combination thereof.

The machine-readable storage medium 120 may store data and/or instructions. In some embodiments, the machine-readable storage medium 120 may store data acquired from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and/or instructions that are used by the big data based user clothing order performance analysis system 100 to perform or use to complete the exemplary methods described in this disclosure.

In a specific implementation, the one or more processors 110 execute computer executable instructions stored by the machine-readable storage medium 120, so that the processor 110 may execute the big data based user clothing order behavior analysis method of the method embodiment, where the processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 may be used to control the transceiving actions of the communication unit 140.

The specific implementation process of the processor 110 may refer to the above-mentioned embodiments of the method executed by the user clothing order behavior analysis system 100 based on big data, and the implementation principle and technical effects are similar, which are not repeated herein.

In addition, the embodiment of the application also provides a readable storage medium, wherein computer executable instructions are preset in the readable storage medium, and when a processor executes the computer executable instructions, the user clothing order behavior analysis method based on big data is realized.

It should be noted that in order to simplify the presentation of the disclosure and thereby aid in understanding one or more embodiments of the application, various features are sometimes grouped together in a single embodiment, figure, or description thereof.

Claims

1. A method for analyzing customer clothing order behavior based on big data, the method comprising:

acquiring user clothing order behavior big data of a target user;

Determining a collaborative interest path of each target user based on the target user collaborative knowledge map, and pushing information to each target user based on the collaborative interest path of each target user;

generating a target user collaborative knowledge graph according to the user collaborative interest state and the user interest correlation degree comprises the following steps:

Generating the target user collaborative knowledge graph composed of the knowledge members and the knowledge links;

after generating a target user collaborative knowledge graph according to the user collaborative interest state and the user interest correlation degree, the method comprises the following steps:

carrying out reinforced expression on knowledge links between the first knowledge member and the second knowledge member and knowledge links between the second knowledge member;

The method further comprises the steps of:

based on the selected instruction of the contracted knowledge member, presenting a third knowledge member with a first-level interest correlation degree with the target knowledge member;

The method further comprises the steps of:

weakening expression is carried out on knowledge members which do not meet the filtering requirement in the target user collaborative knowledge graph;

The method further comprises the steps of:

Optimizing the feature vector generated by the knowledge member and/or the knowledge link according to the optimization parameter, wherein the information which can be displayed by the knowledge member comprises a user mark, a user name and the number of collaborative groups, and the information which can be displayed by the knowledge link comprises a user interest correlation degree;

the determining the user collaborative interest state and the user interest correlation degree between the target users according to the user clothing order behavior big data comprises the following steps:

The collaborative feature data comprises a collaborative purchase rate promotion vector, a collaborative browsing rate promotion vector and a user preference matching degree, wherein the collaborative purchase rate promotion vector is used for representing the influence condition of the reference user set on the purchase rate, the collaborative browsing rate promotion vector is used for representing the influence of the reference user set on the commodity browsing rate, and the user preference matching degree is used for representing whether similar clothing preference exists between at least two target users;

the method comprises the following steps:

When the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector are forward lifting vectors, the user preference matching degree indicates that similar preference does not exist, and the significant lifting vector value of the collaborative purchase rate lifting vector is smaller than a threshold value, determining that the collaborative matching requirement is not met;

The determining the user interest correlation degree of the user cooperative group according to the cooperative characteristic data includes:

Determining the user interest correlation degree of the user collaborative group according to the collaborative purchase rate lifting vector and the collaborative browsing rate lifting vector, wherein the user interest correlation degree and the collaborative purchase rate lifting vector are in a forward association relationship, and the user interest correlation degree and the collaborative browsing rate lifting vector are in a forward association relationship;

The step of determining a collaborative interest path of each target user based on the target user collaborative knowledge graph and pushing information to each target user based on the collaborative interest path of each target user comprises the following steps:

2. A big data based user clothing order performance analysis system, characterized in that the big data based user clothing order performance analysis system comprises a processor and a memory, the memory is connected with the processor, the memory is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the memory so as to realize the big data based user clothing order performance analysis method according to claim 1.