An intelligent user behavior analysis platform based on big data and AI

Abstract

The present invention relates to the field of data processing technology, and in particular to an intelligent user behavior analysis platform based on big data and AI, comprising a data acquisition module for collecting web page data and mobile terminal data; a data preprocessing module for preprocessing web page data and mobile terminal data to obtain a user behavior data set; a behavior association analysis module for calculating the item category support and target item association confidence based on the user behavior data set, and judging the user closeness; a user behavior prediction module for calculating and judging the purchase tendency of target platform items; and a copy push selection module for calculating the adaptation form of high-quality target item copy based on the purchase tendency of target platform items, and pushing it as a user-selected copy. The present invention comprehensively collects and processes data through big data processing and artificial intelligence, accurately analyzes user behavior, and recommends personalized copy, thereby achieving accurate grasp and efficient satisfaction of user needs.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an intelligent user behavior analysis platform based on big data and AI Background Art

[0002] Big data advertising recommendation technology collects and analyzes massive amounts of user behavior data to explore user preferences and needs. It uses advanced algorithm models to achieve personalized advertising push, improve advertising conversion rates and user satisfaction, and is a key technology in the field of modern digital marketing.

[0003] For example, a Chinese patent with publication number CN118505311A discloses an advertising information recommendation system based on big data, including: a management control module, an analysis and optimization module, and a transmission feedback module, characterized in that: the management control module is network-connected to the analysis and optimization module, the analysis and optimization module is network-connected to the transmission feedback module, the management control module is used to manage and control mailbox advertising recommendations, the analysis and optimization module is used to analyze and optimize advertising recommendations, and the transmission feedback module is used to perform transmission feedback control of data information. The management control module includes a graph construction module, a node management module, and a relationship mining module, the graph construction module is electrically connected to the node management module, the graph construction module is network-connected to the relationship mining module, and the graph construction module is used to construct and process a knowledge graph of mailbox user information; it can be seen that the advertising information recommendation system based on big data has the problem of insufficient personalization of user behavior analysis and advertising copy recommendation schemes during actual use. Summary of the Invention

[0004] To this end, the present invention provides an intelligent user behavior analysis platform based on big data and AI to overcome the problem of insufficient personalization of user behavior analysis and copy recommendation solutions in the prior art.

[0005] To achieve the above objectives, the present invention provides an intelligent user behavior analysis platform based on big data and AI.

[0006] Data collection module, used to collect web data and mobile data;

[0007] a data preprocessing module for cleaning the web page data and the mobile terminal data to obtain valid web page data and valid mobile terminal data, integrating the valid web page data and the valid mobile terminal data to obtain data to be converted, and converting the data to be converted into a user behavior data set;

[0008] a behavior association analysis module, configured to determine the target item category based on the user behavior dataset, calculate the target item association confidence based on the target item category and the target platform item set, and determine the user closeness based on the target item association confidence;

[0009] A user behavior prediction module is used to judge the purchase tendency of the target platform item based on the association confidence of the target item;

[0010] The copy push selection module is used to judge the high-quality target item copy based on the purchase tendency of the target platform items, and is also used to judge the adaptation form of the high-quality target item copy based on the high-quality target item copy form confidence and high-quality target item copy form support, and push the high-quality target item copy and the adaptation form as user selected copy, and is also used to compare the number of mobile terminal behaviors with the number of web terminal behaviors, and adjust the adaptation form judgment process according to the comparison results.

[0011] Furthermore, the data preprocessing module includes a data cleaning unit, which performs noise removal on the web data and mobile data, and performs missing value processing on the web data and mobile data after the noise removal processing. After completing the missing value processing, the data in the web data and mobile data that does not conform to the business logic is filtered to obtain valid web data and valid mobile data.

[0012] Furthermore, the data preprocessing module also includes a data integration unit, which performs data conflict elimination processing on the data in the valid web data and the valid mobile data, eliminates the conflicting data in the valid web data and the valid mobile data, and performs multi-source data fusion processing on the valid web data and the valid mobile data after the data conflict elimination processing to obtain the data to be converted.

[0013] Furthermore, the data preprocessing module also includes a data conversion unit, which discretizes various types of data in the data to be converted and standardizes the discretized data to be converted to obtain a user behavior data set, wherein the user behavior data set includes basic user behavior.

[0014] Furthermore, the behavior association analysis module calculates the item category support Support{X} based on the item category behavior item set X and the total item behavior item set N in the user behavior dataset, and sets The item category support Support{X} is compared with the preset minimum item category support Support{X}0, and the target item category is determined based on the comparison result, where:

[0015] When Support{X}≥Support{X}0, the behavior association analysis module determines that the item category behavior item set X is a frequent item set, and uses the item category corresponding to the item category behavior item set X as the target item category;

[0016] When Support{X}<Support{X}0, the behavior association analysis module determines that the item category behavior item set X is a non-frequent item set;

[0017] The count{X} refers to the number of times the item category behavior item set X appears in the user behavior dataset. The preset minimum item category support Support{X}0 refers to the preset support value for determining whether the item category behavior item set X is a frequent item set, and is set to 0.3<Support{X}0<1.

[0018] Furthermore, the behavior association analysis module obtains the item category behavior item set of the target item category, takes it as the target item category behavior item set Xa, and calculates the target item association confidence according to the support Support(Xa∪Y) of the union of the target item category behavior item set Xa and the target platform item set Y. set up Support(Xa∪Y) is the support of the target item category behavior item set Xa and the target platform item set Y. The behavior association analysis module converts the target item association confidence Minimum confidence level with preset association rules Perform a comparison and determine the user closeness based on the comparison results, where:

[0019] when When , the behavior association analysis module determines that the user closeness is high;

[0020] when When , the behavior association analysis module determines that the user closeness is low;

[0021] The minimum confidence of the preset association rule Refers to the preset value of the association rule for judging the closeness of users.

[0022] Furthermore, the user behavior prediction module predicts the target item according to its associated confidence. And the target platform item support Support (Y) calculates the target item promotion degree, set And judge the purchase tendency of the target platform items based on the calculation results, where:

[0023] when When the user behavior prediction module determines that the target platform item has a high purchase tendency, the user behavior prediction module recommends high-frequency copywriting for the target platform item;

[0024] when When the user behavior prediction module determines that the purchase tendency of the target platform item is average, the user behavior prediction module performs random copywriting recommendation of the target platform item;

[0025] when When the user behavior prediction module determines that the purchase tendency of the target platform item is low, no copy recommendation of the target platform item is performed.

[0026] Furthermore, the copy push selection module selects the target platform items based on the support of the copy A with high purchase tendency (Support(A), the support of the union of copy A and target user B (Support(A∪B)), the user copy confidence Calculate the target item copy acceptance set up And judge the high-quality target product copy based on the calculation results, among which:

[0027] when When the target article A is received, the article selection module determines that the target article A has a high acceptance rate and that the article A is a high-quality target article A, and increases the frequency of pushing the article A to the user B;

[0028] when When the copy push selection module determines that the acceptance of the target item copy is low, the copy A does not belong to the high-quality target item copy, and the copy A is not pushed to the target user B.

[0029] Furthermore, the copy push selection module selects the target product copy according to its confidence level. Calculate the improvement of the high-quality target item copywriting form by the support degree (D) of the high-quality target item copywriting form set up And based on the calculation results, the adaptation form of the high-quality target item copy is judged, where:

[0030] when When the copy push selection module determines that the adaptation form of the high-quality target item copy is appropriate, and pushes it;

[0031] when When the copy push selection module determines that the adaptation form of the high-quality target item copy is inappropriate, the push is cancelled.

[0032] Furthermore, when the adaptation form is the copy length, the copy push selection module compares the number of mobile terminal behaviors P with the number of web page behaviors Q, and adjusts the adaptation form determination process according to the comparison result, wherein:

[0033] When P < Q, no adjustment is made to the adaptation form determination process;

[0034] When P≥Q, adjust the adaptation form determination process and set the adjustment coefficient V=0.6+e -(P-Q)-1 ,e is the base of natural logarithm, and the degree of improvement of the copywriting form of high-quality target items is improved according to the adjustment coefficient V After adjustment, the quality target item copywriting form improvement degree is W, set

[0035] Compared with the existing technology, the beneficial effect of the present invention is that the platform comprehensively collects data from the web and mobile terminals through the data collection module, ensuring the breadth and diversity of the data, and providing a solid foundation for subsequent data analysis and user behavior prediction. The platform data preprocessing module cleans and integrates the collected data, effectively removes noise and redundant information, and improves the accuracy and availability of the data. At the same time, the processed data is converted into a user behavior data set, providing direct and effective data support for subsequent behavior association analysis. Through the analysis of the user behavior data set by the behavior association analysis module, the support of the item category is calculated, and the target item category is accurately judged. Furthermore, the association confidence is calculated according to the target item category and the target platform item set, and the closeness between the user and the item is effectively evaluated. In addition, the calculation order of the association confidence is adjusted according to the length of the target item category, which optimizes the analysis process and improves the analysis efficiency. The user behavior prediction module calculates the target item promotion degree based on the results of the behavior association analysis and accurately predicts the user's purchase tendency on the target platform. This step provides a strong basis for the subsequent copy push and helps to achieve precision marketing. The copy push selection module calculates the target item copy acceptance according to the user's purchase tendency and accurately judges the high-quality target item copy. At the same time, the promotion degree is calculated based on the copy form confidence and support, and the copy adaptation form is optimized. The high-quality target item copy and the adaptation form are pushed as user-selected copy, creating personalized copy design for customers, improving the pertinence and attractiveness of the copy, and helping to improve user satisfaction and purchase conversion rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of the intelligent user behavior analysis platform based on big data and AI in this embodiment. DETAILED DESCRIPTION

[0037] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0038] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0039] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0040] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0041] See also Figure 1 As shown in FIG, it is a flow chart of the intelligent user behavior analysis platform based on big data and AI in this embodiment, including:

[0042] Data collection module, used to collect web data and mobile data;

[0043] a data preprocessing module for cleaning the web page data and the mobile terminal data to obtain valid web page data and valid mobile terminal data, integrating the valid web page data and the valid mobile terminal data to obtain data to be converted, and converting the data to be converted into a user behavior data set; the data preprocessing module is connected to the data acquisition module;

[0044] a behavior association analysis module, configured to determine the target item category based on the user behavior dataset, calculate the target item association confidence based on the target item category and the target platform item set, and determine the user closeness based on the target item association confidence. The behavior association analysis module is connected to the data preprocessing module;

[0045] A user behavior prediction module, used to judge the purchase tendency of the target platform item based on the target item association confidence, and the user behavior prediction module is connected to the behavior association analysis module;

[0046] The copy push selection module is used to judge the high-quality target item copy according to the purchase tendency of the target platform items, and is also used to judge the adaptation form of the high-quality target item copy according to the high-quality target item copy form confidence and high-quality target item copy form support, and push the high-quality target item copy and the adaptation form as user selected copy, and is also used to compare the number of mobile terminal behaviors with the number of web page behaviors, and adjust the adaptation form judgment process according to the comparison results. The push copy selection module is connected to the user behavior prediction module.

[0047] Specifically, the platform is applied to short video software. The platform comprehensively collects data from web pages and mobile terminals through the data collection module, ensuring the breadth and diversity of the data, and providing a solid foundation for subsequent data analysis and user behavior prediction. The platform data preprocessing module cleans and integrates the collected data, effectively removes noise and redundant information, and improves the accuracy and availability of the data. At the same time, the processed data is converted into a user behavior data set, providing direct and effective data support for subsequent behavior association analysis. Through the analysis of the user behavior data set by the behavior association analysis module, the support of the item category is calculated, and the target item category is accurately judged. Furthermore, the association confidence is calculated according to the target item category and the target platform item set, and the closeness between the user and the item is effectively evaluated. In addition, the calculation order of the association confidence is adjusted according to the length of the target item category, which optimizes the analysis process and improves the analysis efficiency. The user behavior prediction module calculates the target item promotion degree based on the results of the behavior association analysis and accurately predicts the user's purchase tendency on the target platform. This step provides a strong basis for the subsequent copy push and helps to achieve precision marketing. The copy push selection module calculates the target item copy acceptance according to the user's purchase tendency and accurately judges the high-quality target item copy. At the same time, the promotion degree is calculated based on the copy form confidence and support, and the copy adaptation form is optimized. The high-quality target item copy and the adaptation form are pushed as user-selected copy, creating personalized copy design for customers, improving the pertinence and attractiveness of the copy, and helping to improve user satisfaction and purchase conversion rate.

[0048] Specifically, the web-side data refers to various types of data generated, collected and stored on the web browser side. The web-side data includes search behavior, browsing behavior, interactive behavior and purchase behavior generated by the web browser side. The data collection module collects the web-side data through JavaScript tracking code and Web log analysis. The mobile-side data refers to various types of data generated, collected and stored by mobile phones and tablets. The mobile-side data includes search behavior, browsing behavior, interactive behavior and purchase behavior generated by mobile phones and tablets. The data collection module collects the mobile-side data through DK integration, Wi-Fi positioning and probe technology, and membership card and payment data association.

[0049] Specifically, the data preprocessing module includes a data cleaning unit, which performs noise removal on the web data and mobile data, and performs missing value processing on the web data and mobile data after noise removal. After completing the missing value processing, the data in the web data and mobile data that does not conform to the business logic is filtered to obtain valid web data and valid mobile data.

[0050] Specifically, the noise removal data processing refers to automatically identifying and filtering out duplicate records caused by network fluctuations and abnormal behavior data caused by user misoperation in web data and mobile data by setting threshold rules. The threshold rule refers to setting a limit for the data. When the data exceeds the limit, the data is considered abnormal and filtered out. The threshold rule includes a repetition frequency threshold. The repetition frequency threshold means that when the number of times a certain data appears repeatedly within a period of time exceeds this threshold, it is considered to be a duplicate record and should be filtered. The data that does not conform to business logic refers to records in e-commerce order data with negative order amounts or zero item quantities.

[0051] Specifically, by removing noise data processing, it is possible to automatically identify and filter out duplicate records caused by network fluctuations and abnormal behavior data caused by user misoperation. Removing these noise data can significantly improve the quality of the data. Missing value processing is performed on the data after noise data removal, and missing data is filled in to make the data set more complete, which can ensure the accuracy and stability of data analysis. By filtering invalid data processing, data that does not conform to business logic can be excluded. After the above processing steps, the valid web data and valid mobile data obtained provide a solid foundation for subsequent data analysis.

[0052] Specifically, the data preprocessing module also includes a data integration unit, which performs data conflict elimination processing on the data in the valid web data and the valid mobile data, eliminates the conflicting data in the valid web data and the valid mobile data, and performs multi-source data fusion processing on the valid web data and the valid mobile data after the data conflict elimination processing to obtain the data to be converted.

[0053] Specifically, the data conflict elimination process refers to resolving conflicts by setting data update time priority when the same user has different classification labels for the same item in different systems, recording the last update time of the data, and giving priority to data with a later update time in the event of a conflict. The multi-source data fusion process refers to integrating user data from multiple different channels and systems, establishing a unified data relationship model, and clarifying the entities, attributes, and relationships in the unified data relationship model. The data relationship model refers to a structure used to organize and represent user data and their relationships, including entities and their attributes, as well as relationships between entities.

[0054] Specifically, data conflict elimination processing can identify and correct conflicting data in valid web data and valid mobile data to ensure data accuracy and consistency. Multi-source data fusion processing can integrate valid web data and valid mobile data from different channels and systems to form a comprehensive data set. Through this data set, more valuable information can be mined to provide assistance for subsequent decision-making.

[0055] Specifically, the data preprocessing module also includes a data conversion unit, which discretizes various types of data in the data to be converted and standardizes the discretized data to be converted to obtain a user behavior data set, which includes basic user behavior.

[0056] Specifically, the discretization processing refers to dividing the continuous user behavior data in the data to be converted into equal-width intervals and classifying the data to be converted. The standardization processing refers to converting the text data and web page addresses in the data to be converted into digital data corresponding to the user behavior type according to the item category.

[0057] Specifically, the data to be converted is classified through discretization processing, and the text data in the data to be converted is converted into digital data using standardization processing, so that the subsequent user behavior analysis module can use the data.

[0058] Specifically, the behavior association analysis module calculates the item category support Support{X} based on the item category behavior item set X and the total item behavior item set N in the user behavior dataset, setting The item category support Support{X} is compared with the preset minimum item category support Support{X}0, and the target item category is determined based on the comparison result, where:

[0059] When Support{X}≥Support{X}0, the behavior association analysis module determines that the item category behavior item set X is a frequent item set, and uses the item category corresponding to the item category behavior item set X as the target item category;

[0060] When Support{X}<Support{X}0, the behavior association analysis module determines that the item category behavior item set X is a non-frequent item set;

[0061] The count{X} refers to the number of times the item category behavior item set X appears in the user behavior dataset. The preset minimum item category support Support{X}0 refers to the preset support value for determining whether the item category behavior item set X is a frequent item set, and is set to 0.3<Support{X}0<1.

[0062] Specifically, the item category behavior item set X refers to the behavior item set made by users in the user behavior dataset for a certain type of items. The behavior item set includes the number of search behaviors, the number of browsing behaviors, the number of interactive behaviors, and the number of purchase behaviors. The total item behavior item set N refers to the sum of the behavior item sets made by users in the user behavior dataset for all categories of items. The frequent item set refers to the item category behavior item set whose item category support is greater than the preset minimum item support.

[0063] Specifically, by setting a preset minimum item support, frequent item sets can be quickly screened out, reducing unnecessary calculations, thereby improving the overall efficiency of data mining. By screening out frequent item sets, the foundation can be laid for subsequent confidence and lift calculations, thereby discovering more valuable association rules.

[0064] Specifically, the behavior association analysis module obtains the item category behavior item set of the target item category, takes it as the target item category behavior item set Xa, and calculates the target item association confidence according to the support Support(Xa∪Y) of the union of the target item category behavior item set Xa and the target platform item set Y set up Support(Xa∪Y) is the support of the target item category behavior item set Xa and the target platform item set Y. The behavior association analysis module converts the target item association confidence Minimum confidence level with preset association rules Perform a comparison and determine the user closeness based on the comparison results, where:

[0065] when When , the behavior association analysis module determines that the user closeness is high;

[0066] when When , the behavior association analysis module determines that the user closeness is low;

[0067] The preset minimum confidence of the association rule Refers to the preset value of the association rule for judging the closeness of users.

[0068] Specifically, the target item category behavior item set Xa refers to the item category behavior item set of the same item category in the frequent item set corresponding to the target item category, the target platform item set Y refers to the sum of items corresponding to the target item category owned by the platform, the user closeness refers to the closeness between platform items and user preferences, that is, the degree of association between platform items and user preferences, the support degree Support(Xa∪Y) of the union of the target item category behavior item set Xa and the target platform item set Y is calculated based on the target item category behavior item set Xa, the target platform item set Y, the total item behavior item set N and the platform item set Yx, and set

[0069] Specifically, by calculating the target item association confidence and combining it with the preset minimum confidence of the association rule, we can screen out item set combinations with high user closeness, thereby improving the accuracy of predicting future user behavior.

[0070] Specifically, the user behavior prediction module is based on the target item association confidence And the target platform item support Support (Y) calculates the target item promotion degree, set And judge the purchase tendency of the target platform items based on the calculation results, where:

[0071] when When the user behavior prediction module determines that the target platform item has a high purchase tendency, the user behavior prediction module recommends high-frequency copywriting for the target platform item;

[0072] when When the user behavior prediction module determines that the purchase tendency of the target platform item is average, the user behavior prediction module performs random copywriting recommendation of the target platform item;

[0073] when When the user behavior prediction module determines that the purchase tendency of the target platform item is low, no copy recommendation of the target platform item is performed.

[0074] Specifically, the target item improvement degree It is a key metric used to evaluate the strength of the association between two item sets in the association rule, namely the target item category behavior item set Xa and the target platform item item set Y. The target platform item support is an indicator used to reflect the degree of user preference for the target platform item. The target platform item support (Y) is calculated based on the target platform item item set Y and the platform item total item set Yx.

[0075] Specifically, by calculating the lift degree of the target item, the user's purchasing desire can be understood, and different copywriting recommendation plans can be implemented for the user. When the lift degree is greater than 1, it is possible to identify users with high purchasing intention for specific products and recommend them high-frequency copywriting, thereby improving the conversion rate of copywriting and user satisfaction. When the lift degree is equal to 1, random frequency copywriting recommendations are used to test user reactions to different products, providing data support for subsequent personalized marketing strategies. When the lift degree is less than 1, it is possible to identify users who are not interested in specific products, thereby avoiding pushing related copywriting to them, saving copywriting costs and reducing user dissatisfaction.

[0076] Specifically, the copy push selection module selects the target platform items based on the support of the copy A with high purchase tendency Support (A), the support of the union of copy A and target user B Support (A∪B), the user copy confidence Calculate the target item copy acceptance set up And judge the high-quality target product copy based on the calculation results, among which:

[0077] when When the target article A is received, the article selection module determines that the target article A has a high acceptance rate, and the article A is a high-quality target article A, and increases the frequency of pushing the article A to the user B;

[0078] when When the copy push selection module determines that the acceptance of the target item copy is low, the copy A does not belong to the high-quality target item copy, and the copy A is not pushed to the target user B.

[0079] Specifically, the user copy confidence refers to the platform's assessment of the strength of the association between the user and the copy. The high-quality target item copy refers to the copy with high user acceptance. The copy A support (A) is calculated based on the item set A of the target platform item corresponding copy and the total type item set Ax of the target platform copy. The support degree of the union of document A and target user B, Support(A∪B), is calculated based on the union of the item set A and target user B. Where Bx is the sum itemset of all user types.

[0080] Specifically, by comprehensively considering the purchase tendency of items on the target platform, the support between the copy and the target users, and the user's copy confidence, the solution can more accurately determine which copy is more likely to be accepted and liked by the target users, helping to reduce irrelevant and unwelcome information push, thereby improving users' reading experience and satisfaction.

[0081] Specifically, the copy push selection module selects high-quality target items based on their copy form confidence. Calculate the improvement of the high-quality target item copywriting form by the support degree (D) of the high-quality target item copywriting form set up And based on the calculation results, the adaptation form of the high-quality target item copy is judged, where:

[0082] when When the copy push selection module determines that the adaptation form of the high-quality target item copy is appropriate, and pushes it;

[0083] when When the copy push selection module determines that the adaptation form of the high-quality target item copy is inappropriate, the push is cancelled.

[0084] Specifically, the high-quality target item copy form support is an indicator that reflects the user's support for the high-quality target item copy form, the high-quality target item copy form confidence is an indicator that reflects the degree of correlation between the user and the high-quality target item copy form, the high-quality target item copy form improvement is a measure of the advantage of a specific form of high-quality target item copy over other forms in attracting user attention, conveying product information and stimulating purchasing desire, the adaptation form refers to the formal adjustment and optimization of the high-quality target item copy according to the characteristics of the target user group, the characteristics of the target platform and the needs of item dissemination, so as to ensure that it can be presented to the user in the best state, the adaptation form includes the copy length, copy content and copy format, the high-quality target item copy form support Support (D) is calculated based on the high-quality target item copy form item set D and the total copy form item set Dx, and is set Confidence level of the high-quality target product copywriting format It is based on the support degree of the high-quality target item copy form Support (D), the support degree of the union of the high-quality target item copy form item set D and the target user B (D∪B), and sets The support degree Support (D∪B) of the union of the high-quality target item copy form item set D and the target user B is calculated based on the high-quality target item copy form item set D, the target user B, the sum item set Bx of all user types and the total copy form item set Dx.

[0085]

[0086] Specifically, by comprehensively considering the degree of match between the copy form and the target users and target platforms (i.e., copy form support) as well as the reliability and effectiveness of the copy form itself (i.e., copy form confidence), this solution can more accurately determine which copy form is most suitable for the target users, thereby improving the accuracy and effectiveness of the push.

[0087] Specifically, when the adaptation form is the copy length, the copy push selection module compares the number of mobile terminal behaviors P with the number of web page behaviors Q, and adjusts the adaptation form determination process according to the comparison result, wherein:

[0088] When P < Q, no adjustment is made to the adaptation form determination process;

[0089] When P≥Q, adjust the adaptation form determination process and set the adjustment coefficient V=0.6+e -(P-Q)-1 ,e is the base of natural logarithm, and the degree of improvement of the copywriting form of high-quality target items is improved according to the adjustment coefficient V After adjustment, the quality target item copywriting form improvement degree is W, set

[0090] Specifically, the number of mobile terminal behaviors refers to the number of mobile terminal behaviors on a certain category of items in the user behavior data set, and the number of web page behaviors refers to the number of web page behaviors on a certain category of items in the user behavior data set.

[0091] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. An intelligent user behavior analysis platform based on big data and AI, characterized by: include: Data collection module, used to collect web data and mobile data; a data preprocessing module for cleaning the web page data and the mobile terminal data to obtain valid web page data and valid mobile terminal data, integrating the valid web page data and the valid mobile terminal data to obtain data to be converted, and converting the data to be converted into a user behavior data set; a behavior association analysis module, configured to determine the target item category based on the user behavior dataset, calculate the target item association confidence based on the target item category and the target platform item set, and determine the user closeness based on the target item association confidence; A user behavior prediction module is used to judge the purchase tendency of the target platform item based on the association confidence of the target item; The copy push selection module is used to judge the high-quality target item copy based on the purchase tendency of the target platform items, and is also used to judge the adaptation form of the high-quality target item copy based on the high-quality target item copy form confidence and high-quality target item copy form support, and push the high-quality target item copy and the adaptation form as user selected copy, and is also used to compare the number of mobile terminal behaviors with the number of web terminal behaviors, and adjust the adaptation form judgment process according to the comparison results.

2. The intelligent user behavior analysis platform based on big data and AI according to claim 1 is characterized in that: The data preprocessing module includes a data cleaning unit, which removes noise from the web page data and the mobile terminal data, and performs missing value processing on the web page data and the mobile terminal data after the noise is removed. After completing the missing value processing, the data in the web page data and the mobile terminal data that does not conform to the business logic is filtered to obtain valid web page data and valid mobile terminal data.

3. The intelligent user behavior analysis platform based on big data and AI according to claim 2 is characterized in that: The data preprocessing module also includes a data integration unit, which performs data conflict elimination processing on the data in the valid web data and the valid mobile data, eliminates the conflicting data in the valid web data and the valid mobile data, and performs multi-source data fusion processing on the valid web data and the valid mobile data after the data conflict elimination processing to obtain the data to be converted.

4. The intelligent user behavior analysis platform based on big data and AI according to claim 3 is characterized in that: The data preprocessing module also includes a data conversion unit, which discretizes various types of data in the data to be converted and standardizes the discretized data to be converted to obtain a user behavior data set, wherein the user behavior data set includes basic user behaviors.

5. The intelligent user behavior analysis platform based on big data and AI according to claim 4 is characterized in that: The behavior association analysis module calculates the item category support Support{X} based on the item category behavior item set X and the total item behavior item set N in the user behavior dataset, and sets The item category support Support{X} is compared with the preset minimum item category support Support{X}0, and the target item category is determined based on the comparison result, where: When Support{X}≥Support{X}0, the behavior association analysis module determines that the item category behavior item set X is a frequent item set, and uses the item category corresponding to the item category behavior item set X as the target item category; When Support{X}<Support{X}0, the behavior association analysis module determines that the item category behavior item set X is a non-frequent item set; The count{X} refers to the number of times the item category behavior item set X appears in the user behavior dataset. The preset minimum item category support Support{X}0 refers to the preset support value for determining whether the item category behavior item set X is a frequent item set, and is set to 0.3<Support{X}0<1.

6. The intelligent user behavior analysis platform based on big data and AI according to claim 5 is characterized in that: The behavior association analysis module obtains the item category behavior item set of the target item category and uses it as the target item category behavior item set Xa, and calculates the target item association confidence based on the support Support(Xa∪Y) of the union of the target item category behavior item set Xa and the target platform item set Y. set up Support(Xa∪Y) is the support of the target item category behavior item set Xa and the target platform item set Y. The behavior association analysis module converts the target item association confidence Minimum confidence level with preset association rules Perform a comparison and determine the user closeness based on the comparison results, where: when When , the behavior association analysis module determines that the user closeness is high; when When , the behavior association analysis module determines that the user closeness is low; The preset minimum confidence of the association rule Refers to the preset value of the association rule for judging the closeness of users.

7. The intelligent user behavior analysis platform based on big data and AI according to claim 6 is characterized in that: The user behavior prediction module is based on the target item association confidence And the target platform item support Support (Y) calculates the target item promotion degree, set And judge the purchase tendency of the target platform items based on the calculation results, where: when When the user behavior prediction module determines that the target platform item has a high purchase tendency, the user behavior prediction module recommends high-frequency copywriting for the target platform item; when When the user behavior prediction module determines that the purchase tendency of the target platform item is average, the user behavior prediction module performs random copywriting recommendation of the target platform item; when When the user behavior prediction module determines that the purchase tendency of the target platform item is low, no copy recommendation of the target platform item is performed.

8. The intelligent user behavior analysis platform based on big data and AI according to claim 7 is characterized in that: The copy push selection module selects the target platform items based on the support of the copy A with high purchase tendency Support (A), the support of the union of copy A and target user B Support (A∪B), the user copy confidence Calculate the target item copy acceptance set up And judge the high-quality target product copy based on the calculation results, among which: when When the target article A is received, the article selection module determines that the target article A has a high acceptance rate and that the article A is a high-quality target article A, and increases the frequency of pushing the article A to the user B; when When the copy push selection module determines that the acceptance of the target item copy is low, the copy A does not belong to the high-quality target item copy, and the copy A is not pushed to the target user B.

9. The intelligent user behavior analysis platform based on big data and AI according to claim 8 is characterized in that: The copy push selection module is based on the confidence level of the copy of the high-quality target item. Calculate the improvement of the high-quality target item copywriting form by the support degree (D) of the high-quality target item copywriting form set up And based on the calculation results, the adaptation form of the high-quality target item copy is judged, where: when When the copy push selection module determines that the adaptation form of the high-quality target item copy is appropriate, and pushes it; when When the copy push selection module determines that the adaptation form of the high-quality target item copy is inappropriate, the push is cancelled.

10. The intelligent user behavior analysis platform based on big data and AI according to claim 9 is characterized in that: When the adaptation form is the copy length, the copy push selection module compares the number of mobile terminal behaviors P with the number of web page behaviors Q, and adjusts the adaptation form determination process according to the comparison result, wherein: When P < Q, no adjustment is made to the adaptation form determination process; When P≥Q, adjust the adaptation form determination process and set the adjustment coefficient V=0.6+e -(P-Q)-1 ,e is the base of natural logarithm, and the degree of improvement of the copywriting form of high-quality target items is improved according to the adjustment coefficient V After adjustment, the quality target item copywriting form improvement degree is W, set

Images