User behavior analysis method and system based on big data

Abstract

The invention provides a user behavior analysis method and system based on big data. The method comprises the following steps: obtaining a query log of a target user on an Internet search service platform; extracting theme tags and long-tail query data; determining a plurality of behavior interestingness degrees according to the long-tail query data, and determining a plurality of query interest sub-tags based on the query interest index and all the behavior interestingness degrees; performing query directivity association on the query interest sub-tags and the theme tags to obtain a plurality of directivity association vectors, and determining behavior trend characteristics according to all the directivity association vectors; obtaining a query word group input by the target user, and performing query matching on the query word group and the behavior trend characteristics to obtain query behavior characteristics; and uploading the query behavior characteristics to a user behavior analysis center. According to the scheme, on the basis of the query behavior characteristics, the query phrases of the user can be accurately captured, the search interest directions of the query phrases are associated and classified, and therefore accurate analysis of the user search behavior mode is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of big data analysis, and more specifically, to a user behavior analysis method and system based on big data. Background Art

[0002] With the explosive growth of data and the rapid improvement of computing power in the information age, big data analysis has gradually become a key technology in modern scientific and technological society. Traditional data processing methods can no longer cope with massive, fast, complex and diverse data from various data sources. Among them, big data analysis mainly relies on distributed computing, cloud storage and advanced algorithm processing architecture for efficient storage and parallel processing of data. In addition, big data analysis also provides key driving force for promoting innovative fields such as personalized recommendations, business decision optimization and intelligent prediction.

[0003] Existing big data analysis uses a variety of technical means to process and analyze massive data. In query behavior analysis, natural language processing technology is used to understand and analyze the search terms and intentions entered by users, so as to accurately analyze the user's interests and behavior patterns. By analyzing these data, the Internet search service platform can realize personalized recommendations and user behavior predictions, and improve user experience and commercial value. However, when users conduct search behaviors, they often express their query needs from the user behavior analysis system based on big data through simple and vague descriptive query phrases. It is difficult for the query system to accurately capture the user's real focus and interest preferences. Secondly, the user's interest preferences when using query phrases to query are dynamically changing. For a key query topic, there will be multiple changes in interest query directions. Existing traditional query methods are difficult to adapt to these changes. Therefore, how to accurately capture the user's query phrases and make associated classifications of the search interest directions of the query phrases, so as to achieve accurate analysis of the user's search behavior patterns has become a difficult problem faced by the industry. Summary of the invention

[0004] The present application provides a user behavior analysis method and system based on big data, which can accurately capture the user's query phrases and make associated classifications of the search interest directions of the query phrases, thereby achieving accurate analysis of the user's search behavior patterns.

[0005] In a first aspect, the present application provides a user behavior analysis method based on big data, comprising the following steps:

[0006] Obtaining the query log of the target user on the Internet search service platform from the big data center of the Internet search service platform;

[0007] Extracting multiple topic tags when the target user performs a query and long-tail query data of each topic tag from the query log;

[0008] A topic tag is selected as the selected topic tag, and multiple behavioral interests of the target user when performing long-tail queries under the selected topic tag are determined based on the long-tail query data of the selected topic tag, and then multiple query interest subtags of the selected topic tag are determined based on the query interest index of the Internet search service platform and all the behavioral interests;

[0009] Associating each query interest sub-tag with the selected topic tag for query directionality, thereby obtaining multiple directionality association vectors corresponding to the selected topic tag, determining the behavior trend characteristics of the target user when performing long-tail queries under the selected topic tag based on all directionality association vectors, and continuing to determine the behavior trend characteristics of the target user when performing long-tail queries under the remaining topic tags;

[0010] Obtaining the query phrase entered by the target user on the Internet search service platform at the current moment, performing query matching on the query phrase and the behavior trend characteristics under each topic tag, and obtaining the query behavior characteristics of the target user;

[0011] The query behavior characteristics of the target user are uploaded to the user behavior analysis center.

[0012] In some embodiments, extracting multiple topic tags when a target user performs a query and long-tail query data of each topic tag from the query log specifically includes:

[0013] Extracting multiple query keywords from the query log;

[0014] All query keywords are clustered to obtain multiple topic tags when the target user performs a query;

[0015] Selecting a topic tag as a selected topic tag, and combining long-tail query records corresponding to each query keyword in the selected topic tag in the query log into long-tail query data of the selected topic tag;

[0016] Continue to identify long-tail query data for the remaining hashtags.

[0017] In some embodiments, all query keywords are clustered into tags to obtain multiple topic tags when the target user performs a query, specifically including:

[0018] Determine the semantic similarity between each query keyword;

[0019] Based on the semantic similarity between each query keyword, a clustering algorithm is used to group all query keywords by similarity to obtain multiple query similarity groups;

[0020] The topic labels of each query similarity group are determined, thereby obtaining multiple topic labels when the target user performs a query.

[0021] In some embodiments, determining multiple behavioral interests of a target user when performing a long-tail query under a selected topic tag based on the long-tail query data of the selected topic tag specifically includes:

[0022] Select a long-tail query term from the long-tail query data of the selected topic tag as the selected long-tail query term;

[0023] Determining the behavior pattern frequency of the selected long-tail query term based on the long-tail query data;

[0024] Determining an interest weight of a selected long-tail query term in the long-tail query data;

[0025] Determining the behavioral interest of the target user when searching for the selected long-tail query term under the selected topic tag according to the behavioral pattern frequency and the interest weight;

[0026] Continue to determine the target users' behavioral interest in the remaining long-tail search terms under the selected topic tag.

[0027] In some embodiments, determining multiple query interest sub-tags of a selected topic tag based on the query interest index of the Internet search service platform and all behavioral interests specifically includes:

[0028] Obtaining the query interest index of the Internet search service platform;

[0029] Performing query behavior analysis on the selected topic tag based on the query interest index and each behavior interest degree to obtain multiple query behavior viscosities of the selected topic tag;

[0030] Multiple query interest sub-tags of the selected topic tag are determined according to the viscosity of all query behaviors.

[0031] In some embodiments, performing query directionality association between each query interest sub-tag and a selected topic tag, and then obtaining a plurality of directionality association vectors corresponding to the selected topic tag specifically includes:

[0032] Selecting a query interest sub-tag as a selected query interest sub-tag;

[0033] Determine the tag co-occurrence rate between the selected query interest subtag and the selected topic tag in the historical search records of the target user;

[0034] Performing a directivity analysis on the selected query interest subtag to obtain the associated directivity of the selected query interest subtag under the selected topic tag;

[0035] Obtain the tag responsiveness of the target user to the selected query interest subtag;

[0036] Determine a directional association vector of the selected query interest sub-tag to the selected topic tag based on the tag co-occurrence rate, the association directivity and the tag responsiveness;

[0037] Continue to determine a plurality of pointing association vectors corresponding to the selected topic label.

[0038] In some embodiments, determining the behavior trend characteristics of the target user when performing a long-tail query under a selected topic tag based on all pointing association vectors specifically includes:

[0039] determining multiple association similarities of all pointing association vectors;

[0040] Perform feature decomposition on all associated similarities to obtain multiple behavioral trends and multiple behavioral features under the selected topic tags;

[0041] Based on all behavioral trends and all behavioral features, the behavioral trend features of the target user when performing long-tail queries under the selected topic tags are determined.

[0042] In a second aspect, the present application provides a user behavior analysis system based on big data, comprising:

[0043] An acquisition module is used to acquire the query log of the target user on the Internet search service platform from the big data center of the Internet search service platform;

[0044] A processing module, used to extract multiple topic tags when a target user performs a query and long-tail query data of each topic tag from the query log;

[0045] The processing module is further used to select a topic tag as a selected topic tag, determine multiple behavioral interest levels of the target user when performing a long-tail query under the selected topic tag based on the long-tail query data of the selected topic tag, and then determine multiple query interest sub-tags of the selected topic tag based on the query interest index of the Internet search service platform and all the behavioral interest levels;

[0046] The processing module is further used to associate each query interest sub-tag with the selected topic tag for query directionality, thereby obtaining multiple directionality association vectors corresponding to the selected topic tag, determining the behavior trend characteristics of the target user when performing long-tail queries under the selected topic tag based on all the directionality association vectors, and continuing to determine the behavior trend characteristics of the target user when performing long-tail queries under the remaining topic tags;

[0047] The processing module is further used to obtain the query phrase input by the target user on the Internet search service platform at the current moment, and perform query matching on the query phrase with the behavior trend characteristics under each topic tag to obtain the query behavior characteristics of the target user;

[0048] The execution module is used to upload the query behavior characteristics of the target user to the user behavior analysis center.

[0049] In a third aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores codes, and the processor is configured to obtain the codes and execute the above-mentioned user behavior analysis method based on big data.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned user behavior analysis method based on big data is implemented.

[0051] The technical solution provided by the embodiments disclosed in this application has the following beneficial effects:

[0052] In the user behavior analysis method and system based on big data provided by the present application, first, the query log of the target user on the Internet search service platform is obtained from the big data center of the Internet search service platform; secondly, multiple topic tags when the target user performs a query and the long-tail query data of each topic tag are extracted from the query log; further, a topic tag is selected as the selected topic tag, and multiple behavioral interests of the target user when performing a long-tail query under the selected topic tag are determined based on the long-tail query data of the selected topic tag, and multiple query interest sub-tags of the selected topic tag are determined based on the query interest index of the Internet search service platform and all behavioral interests; then Then, each query interest sub-tag is associated with the selected subject tag for query directionality, thereby obtaining multiple directionality association vectors corresponding to the selected subject tag, and the behavior trend characteristics of the target user when performing long-tail queries under the selected subject tag are determined based on all directionality association vectors, and the behavior trend characteristics of the target user when performing long-tail queries under the remaining subject tags are further determined; in addition, the query phrase entered by the target user on the Internet search service platform at the current moment is obtained, and the query phrase is query-matched with the behavior trend characteristics under each subject tag to obtain the query behavior characteristics of the target user; finally, the query behavior characteristics of the target user are uploaded to the user behavior analysis center.

[0053] It can be seen that the present application can accurately capture the user's query phrases and make associated classifications of the search interest directions of the query phrases, thereby realizing accurate analysis of the user's search behavior patterns; first, extract multiple topic tags when the target user makes a query, so as to highly summarize and condense the main intentions of the query keywords, so as to help the Internet search service platform better understand the query requests of the target user; secondly, determine the multiple behavioral interests of the target user when making long-tail queries under the selected topic tags based on the long-tail query data of the selected topic tags, so as to accurately capture the user's real focus and interest preferences, so as to optimize the correlation strength of the query results; further, determine multiple query interest sub-tags of the selected topic tags, so as to analyze the detailed query directions that the target user shows high interest in other attributes of the topic tags when performing search behaviors under the topic tags, so as to effectively identify changes in the interest query directions; then, each query interest The sub-tag is associated with the selected topic tag for query directionality so as to more clearly understand the strength of association between the two tags, thereby avoiding the problem of being unable to associate and classify query interests; in addition, the query phrase input by the target user is matched with the behavioral trend characteristics under each topic tag to obtain the query behavior characteristics of the target user, so as to reflect the target user's search intention and information needs at the current moment, so as to make an accurate analysis of the target user's potential behavioral trends, and thus better capture the dynamic changes of the user's query for interest query phrases; finally, the query behavior characteristics are uploaded to the user behavior analysis center to build a more complete user portrait, thereby realizing accurate identification of the target user's interest preferences and behavior patterns; in summary, the technical solution provided by the present application can accurately capture the user's query phrases, and associate and classify the search interest directions of the query phrases, thereby realizing accurate analysis of the user's search behavior patterns. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is an exemplary flow chart of a user behavior analysis method based on big data according to some embodiments of the present application;

[0055] Figure 2 is an exemplary flow chart of determining multiple query interest sub-tags of a selected topic tag according to some embodiments of the present application;

[0056] Figure 3 is an exemplary flow chart of determining a plurality of pointing association vectors corresponding to a selected topic tag according to some embodiments of the present application;

[0057] Figure 4 is a schematic diagram of exemplary hardware and / or software of a user behavior analysis system based on big data according to some embodiments of the present application;

[0058] Figure 5It is a structural diagram of a computer device for implementing a user behavior analysis method based on big data as shown in some embodiments of the present application. DETAILED DESCRIPTION

[0059] The core of the present application is to obtain the query log of the target user on the Internet search service platform from the big data center; extract multiple topic tags when the target user performs queries and long-tail query data of each topic tag from the query log; select a topic tag as the selected topic tag, determine multiple behavioral interests of the target user when performing long-tail queries under the selected topic tag according to the long-tail query data of the selected topic tag, and determine multiple query interest sub-tags of the selected topic tag based on the query interest index of the Internet search service platform and all behavioral interests; associate each query interest sub-tag with the selected topic tag for query directionality, and then obtain multiple directional association vectors corresponding to the selected topic tag, determine the behavioral trend characteristics of the target user when performing long-tail queries under the selected topic tag according to all directional association vectors, and continue to determine the behavioral trend characteristics of the target user when performing long-tail queries under the remaining topic tags; obtain the query phrase entered by the target user on the Internet search service platform at the current moment, query and match the query phrase with the behavioral trend characteristics under each topic tag, and obtain the query behavior characteristics of the target user; upload the query behavior characteristics of the target user to the user behavior analysis center. The above scheme is based on query behavior characteristics, can accurately capture the user's query phrases, and make associated classifications of the search interest directions of the query phrases, thereby achieving accurate analysis of the user's search behavior patterns.

[0060] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods. Figure 1 , which is an exemplary flow chart of a user behavior analysis method based on big data according to some embodiments of the present application. The user behavior analysis method 100 based on big data mainly includes the following steps:

[0061] In step 101, the query log of the target user on the Internet search service platform is obtained from the big data center of the Internet search service platform.

[0062] In the specific implementation, the query log of the target user on the Internet search service platform is obtained from the big data center of the Internet search service platform, that is: the target user information is queried on the big data center of the Internet search service platform, and then the query log of the target user on the Internet search service platform is obtained. The Internet search platform is a platform that provides search services through a search engine, and usually stores the original search data used to analyze and process user search behavior through the big data center. Specifically, the big data center of the Internet search service platform of the present application is an integrated system of various data entries and user query logs generated and collected in a big data environment, and a large number of log records are stored thereon, including big data such as user query logs. The query log in the present application records the data generated when the target user queries for information on the Internet search service platform, wherein the query log includes query keywords, long-tail query records, etc. In other embodiments, the query log may also include other data content, which is not limited here. By obtaining the query log, the search behavior of the target user can be analyzed, thereby optimizing the search algorithm, thereby improving the search experience of the target user.

[0063] It should be noted that the query keyword refers to the keyword phrase input by the target user, and the long-tail query record refers to the phrase combination of the long-tail query words. The long-tail query words refer to the query phrases that expand the content and supplement the query keywords. One query keyword corresponds to one long-tail query record, and one long-tail query record contains multiple long-tail query words.

[0064] In step 102, multiple topic tags and long-tail query data of each topic tag when the target user performs a query are extracted from the query log.

[0065] In some embodiments, the following method may be used to extract multiple topic tags and long-tail query data of each topic tag when the target user performs a query from the query log, namely:

[0066] Extracting multiple query keywords from the query log;

[0067] All query keywords are clustered to obtain multiple topic tags when the target user performs a query;

[0068] Selecting a topic tag as a selected topic tag, and combining long-tail query records corresponding to each query keyword in the selected topic tag in the query log into long-tail query data of the selected topic tag;

[0069] Continue to identify long-tail query data for the remaining hashtags.

[0070] In some embodiments, all query keywords are clustered into tags to obtain multiple topic tags when the target user performs a query, which can be done in the following manner, namely:

[0071] Determine the semantic similarity between each query keyword;

[0072] Based on the semantic similarity between each query keyword, a clustering algorithm is used to group all query keywords by similarity to obtain multiple query similarity groups;

[0073] The topic labels of each query similarity group are determined, thereby obtaining multiple topic labels when the target user performs a query.

[0074] In the specific implementation, the semantic similarity between each query keyword is determined, and the semantic similarity model can be used to capture the semantic similarity between each query keyword. Specifically, Word2Vec in the semantic similarity model can be used to capture the semantic similarity between each query keyword. In other embodiments, other semantic similarity models can also be used to capture the semantic similarity between each query keyword, for example, GloVe is not limited here.

[0075] In specific implementation, a clustering algorithm is used to perform similarity grouping on all query keywords based on the semantic similarity between each query keyword to obtain multiple query similarity groups, that is, a K-Means clustering algorithm can be used to perform similarity grouping on all query keywords, and each grouping result is used as a query similarity group to obtain multiple query similarity groups. In addition, in other embodiments, other clustering algorithms can also be used to perform similarity grouping on all query keywords, such as a Hierarchical Clustering clustering algorithm, which is not limited here.

[0076] It should be noted that, in this embodiment, the query similarity group represents a set of multiple query keywords with a high degree of semantic similarity. By determining the query similarity group, the query subjects of the target user can be better classified and sorted, which is conducive to analyzing their behavior patterns.

[0077] In specific implementation, the topic label of each query similarity group can be determined by the topic modeling method, and then multiple topic labels when the target user performs a query can be obtained. In addition, in other embodiments, other methods can also be used to determine the topic label of each query similarity group, such as word frequency analysis method, co-occurrence matrix method, etc., which are not limited here.

[0078] It should be noted that the semantic similarity model in this embodiment represents a model used to capture the semantic similarity between query keywords, the semantic similarity represents a measure of the similarity between query keywords in meaning or concept, the similarity grouping represents the process of grouping all query keywords according to the semantic similarity measure, and in addition, label clustering in this embodiment represents the process of extracting and calibrating core topics that are highly relevant to the query behavior of the target user, which will not be repeated here.

[0079] It should also be noted that the topic tags in this application represent tags of core topics that are highly relevant to the target user's query behavior. A topic tag contains multiple query keywords. By determining the topic tags, the main intention of the query keywords can be highly condensed, which is conducive to the Internet search service platform to better understand the target user's query request.

[0080] In addition, it should be noted that the long-tail query data in this application represents a collection of long-tail query records. One long-tail query data contains multiple long-tail query records. By determining the long-tail query data, the special needs of the target users can be more completely reflected, thereby optimizing the query of the search service.

[0081] In step 103, a topic tag is selected as the selected topic tag, and multiple behavioral interests of the target user when performing long-tail queries under the selected topic tag are determined based on the long-tail query data of the selected topic tag, and then multiple query interest sub-tags of the selected topic tag are determined based on the query interest index of the Internet search service platform and all behavioral interests.

[0082] In some embodiments, the following method may be used to determine the multiple behavioral interests of the target user when performing long-tail queries under the selected topic tag based on the long-tail query data of the selected topic tag, namely:

[0083] Select a long-tail query term from the long-tail query data of the selected topic tag as the selected long-tail query term;

[0084] Determining the behavior pattern frequency of the selected long-tail query term based on the long-tail query data;

[0085] Determining an interest weight of a selected long-tail query term in the long-tail query data;

[0086] Determining the behavioral interest of the target user when searching for the selected long-tail query term under the selected topic tag according to the behavioral pattern frequency and the interest weight;

[0087] Continue to determine the target users' behavioral interest in the remaining long-tail search terms under the selected topic tag.

[0088] In a specific implementation, the behavior pattern frequency of the selected long-tail query word is determined based on the long-tail query data, that is, the total vocabulary of the long-tail query data and the occurrence frequency of the selected long-tail query word in the long-tail query data are respectively obtained, and the ratio of the occurrence frequency to the total vocabulary is used as the behavior pattern frequency of the selected long-tail query word. In addition, in other embodiments, other calculation methods can also be used to calculate the behavior pattern frequency, which is not limited here.

[0089] It should be noted that the behavior pattern frequency in this embodiment represents a statistical indicator used to describe the frequency of target users' queries for long-tail query terms. Determining the behavior pattern frequency can help analyze the interest preferences of target users, thereby optimizing the user's search experience.

[0090] In specific implementation, the interest weight of the selected long-tail query term in the long-tail query data is determined, that is, the total number of long-tail query records in the long-tail query data is obtained, the ratio of the total number of records to the number of long-tail query records containing the selected long-tail query term is used as a weight factor, and the logarithm of the weight factor is used as the interest weight of the selected long-tail query term in the long-tail query data. In addition, in other embodiments, other calculation methods can also be used to calculate the interest weight, which is not limited here.

[0091] It should be noted that the interest weight in this embodiment is used to measure the query importance of the target user when searching for long-tail query terms. By determining the interest weight, the target user's interest direction when searching can be grasped, so as to make more reasonable query retrieval. In addition, the weight factor in this embodiment represents a calculation intermediate variable, which has no practical significance and will not be repeated here.

[0092] In specific implementation, the behavioral interest of the target user when searching for the selected long-tail query word under the selected topic tag is determined based on the behavioral pattern frequency and the interest weight, that is, the product of the behavioral pattern frequency and the interest weight is used as the behavioral interest of the target user when searching for the selected long-tail query word under the selected topic tag. In addition, in other embodiments, other calculation methods can also be used to calculate the behavioral interest, which is not limited here.

[0093] It should be noted that the behavioral interest degree in the present application indicates the preference tendency of the target user for the interest direction represented by the long-tail query word under the subject tag when issuing a query request. The greater the behavioral interest degree, the more prominent the preference tendency of the target user for the interest direction represented by the long-tail query word under the subject tag when issuing a query request. The smaller the behavioral interest degree, the weaker the preference tendency of the target user for the interest direction represented by the long-tail query word under the subject tag when issuing a query request. One behavioral interest degree corresponds to one long-tail query word. By determining the behavioral interest degree, the behavioral motivation of the target user can be better understood, thereby optimizing the correlation strength of the query result.

[0094] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flow chart of determining multiple query interest sub-tags of a selected topic tag according to some embodiments of the present application. In this embodiment, determining multiple query interest sub-tags of a selected topic tag based on the query interest index of the Internet search service platform and all behavioral interests can be implemented by the following steps:

[0095] First, in step 1031, the query interest index of the Internet search service platform is obtained;

[0096] Then, in step 1032, query behavior analysis is performed on the selected topic tag based on the query interest index and each behavior interest degree to obtain multiple query behavior viscosities of the selected topic tag;

[0097] Finally, in step 1033, multiple query interest sub-tags of the selected topic tag are determined according to all query behavior viscosities.

[0098] In specific implementation, the query interest index of the Internet search service platform can be obtained in the search server of the Internet search platform. The query interest index represents an adaptive coefficient generated by the Internet search service platform based on the historical search data of the target user. The query interest index reflects the potential interests of the target user and will not be elaborated here.

[0099] In specific implementation, query behavior analysis is performed on the selected topic tag based on the query interest index and each behavioral interest degree to obtain multiple query behavior viscosities of the selected topic tag, that is, the query interest index is multiplied by each behavioral interest degree respectively, and each product result is used as the query behavior viscosity of the selected topic tag, thereby obtaining multiple query behavior viscosities of the selected topic tag. In addition, in other embodiments, other calculation methods can also be used to calculate the query behavior viscosity, which is not limited here.

[0100] It should be noted that, in this embodiment, the query behavior viscosity represents the target user's dependence on a specific query interest direction when performing a search behavior. The greater the query behavior viscosity, the greater the target user's dependence on the specific query interest direction when performing a search behavior. The smaller the query behavior viscosity, the smaller the target user's dependence on the specific query interest direction when performing a search behavior. One query behavior viscosity corresponds to a long-tail query term. By determining the query behavior viscosity, the target user's specific query interest direction can be effectively analyzed. In addition, in this embodiment, the query behavior analysis represents the process of analyzing the target user's specific query interest direction, which will not be repeated here.

[0101] In some embodiments, the following method may be used to determine multiple query interest sub-tags of a selected topic tag based on all query behavior viscosities, namely:

[0102] Compare each query behavior viscosity with a preset behavior viscosity, extract all query behavior viscosities greater than the preset behavior viscosity, and use the long-tail query words corresponding to each extracted query behavior viscosity as the interest subject words, otherwise, do not process;

[0103] A plurality of query interest sub-tags of the selected topic tag are determined based on all the interest topic words.

[0104] In the specific implementation, multiple query interest sub-tags of the selected topic label are determined based on all interest keywords, that is: an interest keyword is selected as the selected interest keyword, the semantic similarity between the remaining long-tail query words under the selected topic label and the selected interest keyword is captured, and the remaining long-tail query words under the selected topic label are similarly grouped based on all semantic similarities, and a query interest sub-tag with the selected interest keyword as the core is determined, thereby obtaining multiple query interest sub-tags of the selected topic label.

[0105] It should be noted that the process of capturing semantic similarity and the process of similarity grouping in this embodiment have been described in detail in step 102 and will not be repeated here. In addition, the preset behavior viscosity in this embodiment is a pre-set threshold of query behavior viscosity. The specific value can be set according to actual application requirements and is not limited here.

[0106] It should also be noted that, in this embodiment, the interest subject words represent the detailed query directions in which the target user shows high interest during the search process. The query interest sub-tags in this application represent the detailed query directions in which the target user shows high interest for other attributes of the subject tag when performing search behavior under the subject tag. A query interest sub-tag contains multiple long-tail query words. By determining the query interest sub-tags, the special needs of the target user can be better understood, which is beneficial for the search engine to conduct behavioral analysis of the target user's search habits.

[0107] In step 104, query directionality association is performed on each query interest sub-tag and the selected topic tag, thereby obtaining multiple directionality association vectors corresponding to the selected topic tag, and the behavioral trend characteristics of the target user when performing long-tail queries under the selected topic tag are determined based on all directionality association vectors, and the behavioral trend characteristics of the target user when performing long-tail queries under the remaining topic tags are further determined.

[0108] In some embodiments, reference Figure 3 As shown in FIG. 1 , this figure is an exemplary flow chart of determining multiple directional association vectors corresponding to a selected topic tag according to some embodiments of the present application. In this embodiment, query directionality association is performed on each query interest subtag with the selected topic tag, and then multiple directional association vectors corresponding to the selected topic tag are obtained, which can be implemented by the following steps:

[0109] First, in step 1041, a query interest subtag is selected as a selected query interest subtag;

[0110] Secondly, in step 1042, the tag co-occurrence rate between the selected query interest subtag and the selected topic tag in the historical search records of the target user is determined;

[0111] Then, in step 1043, the selected query interest sub-tag is subjected to a directionality analysis to obtain the correlation directionality of the selected query interest sub-tag under the selected topic tag;

[0112] Then, in step 1044, the tag responsiveness of the target user to the selected query interest subtag is obtained;

[0113] In addition, in step 1045, a directional association vector of the selected query interest sub-tag to the selected topic tag is determined based on the tag co-occurrence rate, the association directivity, and the tag responsiveness;

[0114] Finally, in step 1046, a plurality of directional association vectors corresponding to the selected topic tag are continuously determined.

[0115] In some embodiments, the tag co-occurrence rate of the selected query interest subtag and the selected topic tag in the historical search record of the target user may be determined in the following manner, namely:

[0116] Extract all query keywords and all long-tail query words in the query logs of target users;

[0117] Map all query keywords and all long-tail query words to each topic tag and each query interest sub-tag respectively, obtain the number of matches of the query keyword mapping match and the number of matches of the long-tail query word mapping match, add the number of matches of the query keyword mapping match and the number of matches of the long-tail query word mapping match, and use the added result as the total number of times the tag appears in the historical search record of the target user;

[0118] The number of times the selected query interest subtag and the selected topic tag co-occur in the historical search records of the target user is obtained, and the ratio of the number of times the tag co-occurs to the total number of times the tag appears is used as the tag co-occurrence rate of the selected query interest subtag and the selected topic tag in the historical search records of the target user.

[0119] In the specific implementation, all query keywords and all long-tail query words are mapped to each topic tag and each query interest sub-tag respectively, that is: determine whether each query keyword belongs to one of the topic tags, if so, the mapping match is successful, otherwise the mapping match is unsuccessful; determine whether each long-tail query word belongs to one of the query interest sub-tags, if so, the mapping match is successful, otherwise the mapping match is unsuccessful, which will not be repeated here. In addition, the number of matches in this embodiment represents the number of successful mapping matches.

[0120] It should be noted that, in this embodiment, the tag co-occurrence rate is used to measure the degree of correlation between the query interest sub-tag and the subject tag. The higher the tag co-occurrence rate, the better the correlation between the query interest sub-tag and the subject tag. The lower the tag co-occurrence rate, the worse the correlation between the query interest sub-tag and the subject tag.

[0121] In specific implementation, a directional analysis is performed on the selected query interest subtag to obtain the associated directional nature of the selected query interest subtag under the selected subject tag, that is, the TF-IDF algorithm can be used to perform a directional analysis on the selected query interest subtag to obtain the associated directional nature of the selected query interest subtag under the selected subject tag. The specific directional analysis process will not be repeated here. In addition, in other embodiments, other methods can also be used to perform directional analysis on the selected query interest subtag, such as cosine similarity algorithm, etc., which are not limited here.

[0122] It should be noted that the directional analysis in this embodiment represents an analysis method for calculating the text correlation between the query interest sub-tag and the topic tag, which will not be described in detail here.

[0123] It should also be noted that in this embodiment, the correlation directionality represents the correlation between the query interest sub-tag and the topic tag. By determining the correlation directionality, it is helpful to analyze the correlation strength between the two tags, thereby effectively improving the accuracy of the search service platform's analysis of user behavior.

[0124] In specific implementation, the tag responsiveness of the target user to the selected query interest sub-tag can be obtained through the search server of the Internet search platform. The tag responsiveness indicates the interest level of the target user when the Internet search platform pushes the relevant search results according to the selected query interest sub-tag after receiving the search request. Obtaining the tag responsiveness is conducive to a more specific analysis of the specific demand direction of the target user.

[0125] In specific implementation, the pointing association vector of the selected query interest sub-tag for the selected topic tag is determined based on the tag co-occurrence rate, the association directionality and the tag responsiveness, that is: the tag co-occurrence rate, the association directionality and the tag responsiveness are vector-converted to obtain the pointing association vector of the selected query interest sub-tag for the selected topic tag. Specifically, the vector conversion can be achieved through the numpy component in Python. The specific conversion process will not be repeated here. In addition, other vector conversion methods can also be used for conversion, which are not limited here.

[0126] It should be noted that the pointing association vector in the present application represents the vectorized representation of the association pointing indicator of the query interest sub-tag for the selected topic tag. Each query interest sub-tag corresponds to a pointing association vector. By determining the pointing association vector, we can more clearly understand the association strength between the two tags, thereby more accurately analyzing the user search behavior.

[0127] It should also be noted that the query directionality association in the present application represents the analysis process of the directionality association of the query interest subtag with the selected topic tag, wherein each query interest subtag is query-directivity associated with the selected topic tag, namely: selecting a query interest subtag as the selected query interest subtag; determining the tag co-occurrence rate of the selected query interest subtag and the selected topic tag in the historical search records of the target user; performing directionality analysis on the selected query interest subtag to obtain the association directionality of the selected query interest subtag under the selected topic tag; obtaining the tag responsiveness of the target user to the selected query interest subtag; determining the directional association vector of the selected query interest subtag with respect to the selected topic tag based on the tag co-occurrence rate, the association directionality and the tag responsiveness; and continuing to determine multiple directional association vectors corresponding to the selected topic tag, i.e. completing the query directionality association between each query interest subtag and the selected topic tag.

[0128] In some embodiments, the following method may be used to determine the behavior trend characteristics of the target user when performing a long-tail query under a selected topic tag based on all pointing association vectors, namely:

[0129] determining multiple association similarities of all pointing association vectors;

[0130] Perform feature decomposition on all associated similarities to obtain multiple behavioral trends and multiple behavioral features under the selected topic tags;

[0131] Based on all behavioral trends and all behavioral features, the behavioral trend features of the target user when performing long-tail queries under the selected topic tags are determined.

[0132] In the specific implementation, multiple correlation similarities of all pointing correlation vectors are determined, that is, the Euclidean distance between every two pointing correlation vectors is used as the correlation similarity between every two pointing correlation vectors, and then multiple correlation similarities of all pointing correlation vectors are obtained. In addition, in other embodiments, other calculation methods can also be used to obtain the correlation similarity, which is not limited here.

[0133] It should be noted that, in this embodiment, the correlation similarity represents the correlation similarity between two query interest subtags. The higher the correlation similarity, the higher the correlation similarity between the two query interest subtags. The lower the correlation similarity, the lower the correlation similarity between the two query interest subtags.

[0134] In some embodiments, feature decomposition is performed on all associated similarities to obtain multiple behavior trend degrees and multiple behavior features under the selected topic tag in the following manner, namely:

[0135] Construct a similarity matrix through all the associated similarities;

[0136] A plurality of behavioral trends and a plurality of behavioral features under the selected topic tag are determined by the similarity matrix.

[0137] In the specific implementation, a similarity matrix is ​​constructed through all the associated similarities, that is: a matrix dimension is preset, and based on the matrix dimension, each query interest sub-tag is correspondingly used as a row and column of the matrix, and the associated similarities between every two query interest sub-tags are correspondingly filled as matrix elements, thereby obtaining a similarity matrix, wherein a matrix dimension can be preset by the total number of query interest sub-tags. For example, if the total number of query interest sub-tags is 3, the matrix dimension is set to 3*3. The specific setting can be made according to actual needs and is not limited here.

[0138] In a specific implementation, the similarity matrix is ​​used to determine multiple behavioral trend degrees and multiple behavioral features under a selected topic label, that is, multiple matrix eigenvalues ​​of the similarity matrix and matrix eigenvectors corresponding to each matrix eigenvalue are obtained through eigendecomposition calculation, each matrix eigenvalue is used as a behavioral trend degree under the selected topic label, and the matrix eigenvector corresponding to each behavioral trend degree is used as a behavioral feature under the selected topic label, thereby obtaining multiple behavioral trend degrees and multiple behavioral features under the selected topic label.

[0139] It should be noted that, in this embodiment, the behavioral characteristics represent the typical behavioral models and search preferences of the target users when searching on the search service platform, and the behavioral trend represents the probability of the corresponding behavioral characteristics occurring when the target users search under the selected topic tags. By determining the behavioral trend and behavioral characteristics, it is helpful to understand the search preferences of the target users, thereby making a more accurate analysis of the search behavior patterns of the target users, and thus improving the user experience.

[0140] In specific implementation, the behavioral trend features of the target user when performing a long-tail query under the selected topic tag are determined based on all behavioral trend degrees and all behavioral features, that is, the largest behavioral trend degree among all behavioral trend degrees is extracted, and the behavioral features corresponding to the behavioral trend degree are used as the behavioral trend features of the target user when performing a long-tail query under the selected topic tag.

[0141] It should be noted that the behavioral trend characteristics in the present application represent the changing patterns of the target user's search behavior and search interests over a period of time. The behavioral trend characteristics can reflect the target user's search transition process from broad search to detailed search, so that the target user's behavior pattern can be analyzed more accurately and the search engine can be optimized, thereby enabling the search service platform to better adapt to the changes in the target user's needs.

[0142] In specific implementation, through the above-mentioned steps of determining the behavioral trend characteristics of the target user when performing long-tail queries under the selected topic tags based on all pointing association vectors, the behavioral trend characteristics of the target user when performing long-tail queries under the remaining topic tags can be further determined, which will not be repeated here.

[0143] In step 105, the query phrase input by the target user on the Internet search service platform at the current moment is obtained, and the query phrase is matched with the behavior trend characteristics under each topic tag to obtain the query behavior characteristics of the target user.

[0144] In specific implementation, the query phrase input by the target user on the Internet search service platform at the current moment can be obtained by embedding the front-end capture code. In other embodiments, other methods can also be used to obtain the query phrase, which is not limited here.

[0145] In some embodiments, the query phrase is matched with the behavior trend characteristics under each topic tag to obtain the query behavior characteristics of the target user in the following manner, namely:

[0146] Performing semantic analysis on the query phrase to obtain key subject words and auxiliary long-tail words when searching for the query phrase;

[0147] Determining semantic association features between the key subject words and the auxiliary long-tail words;

[0148] The semantic association features are matched with the behavior trend features under each topic tag to obtain the query behavior features of the target user.

[0149] In a specific implementation, a semantic analysis is performed on the query phrase to obtain key subject words and auxiliary long-tail words when querying the query phrase, that is, a word embedding analysis method in natural language processing technology is used to perform a semantic analysis on the query phrase to obtain key subject words and auxiliary long-tail words when querying the query phrase. The specific analysis process will not be repeated here. In addition, in other embodiments, other processing technologies can also be used to perform semantic analysis on the query phrase, such as topic modeling method, text similarity calculation method, etc., which are not limited here.

[0150] It should be noted that, in this embodiment, the key subject words represent the search topics when the target user searches, and the auxiliary long-tail words represent the auxiliary words that supplement the content and expand the direction of the search topics when the target user searches.

[0151] In the specific implementation, the semantic association features between the key subject words and the auxiliary long-tail words are determined, that is, the key subject words and the auxiliary long-tail words are input into a pre-trained semantic association model to obtain the semantic association features between the key subject words and the auxiliary long-tail words.

[0152] It should be noted that the pre-trained semantic association model in this embodiment is a model obtained by semantic association training using a large number of query keyword samples and long-tail query word samples. This application adopts a word embedding model. In addition, in other embodiments, other models can also be used, such as context embedding models, etc., which will not be repeated here.

[0153] It should also be noted that in this embodiment, the semantic association feature represents the association between the target user's search behavior pattern and the key subject words and auxiliary long-tail words. By determining the semantic association feature, the target user's search interest direction and search behavior trend can be analyzed.

[0154] In some embodiments, the semantic association feature is matched with the behavior trend feature under each topic tag to obtain the query behavior feature of the target user in the following manner, namely:

[0155] Determine the similarity matching degree between the semantic association feature and the behavior trend feature under each topic tag;

[0156] The maximum similarity matching degree is extracted from all similarity matching degrees, and the behavior trend feature corresponding to the similarity matching degree is used as the query behavior feature of the target user.

[0157] In specific implementation, the Bayesian classification method can be used to determine the similarity matching degree between the semantic association features and the behavioral trend features under each topic tag. In addition, in other embodiments, other calculation methods can also be used to calculate the similarity matching degree, for example, the Jaccard similarity method, the Pearson correlation coefficient method, etc., which are not limited here.

[0158] It should be noted that, in this embodiment, the similarity matching degree represents the correlation between the semantic association feature and the behavioral trend feature. One similarity matching degree corresponds to one behavioral trend feature. By determining the similarity matching degree, the behavioral similarity between the target user's query phrase and each topic tag can be reflected, so that the target user's search behavior can be accurately inferred and analyzed.

[0159] It should also be noted that the query behavior characteristics in this application represent the specific search behavior patterns and search direction preferences exhibited by the target user when using the search service platform to query information. By determining the query behavior characteristics, the target user's search intention and information needs can be reflected, thereby making an accurate analysis of the target user's potential behavioral trends.

[0160] In step 106, the query behavior characteristics of the target user are uploaded to a user behavior analysis center.

[0161] In some embodiments, the query behavior characteristics of the target user may be uploaded to the user behavior analysis center in the following manner, namely:

[0162] The query behavior characteristics of the target user are processed in a transmission format, and the query behavior characteristics after the transmission format processing are sent to a user behavior analysis center through a standard transmission protocol, and the user behavior analysis center outputs the user's query behavior.

[0163] In specific implementation, the query behavior characteristics of the target user may be processed in a transmission format by the format conversion tool Apache NiFi according to the interface format of the user behavior analysis center. In addition, in other embodiments, other format conversion tools may be used to process the query behavior characteristics in a transmission format, such as Talend, etc., which is not limited here.

[0164] In specific implementation, the query behavior characteristics after transmission format processing are sent to the user behavior analysis center through the standard transmission protocol, and the user behavior analysis center outputs the user's query behavior, that is: the user management center further processes and analyzes the query behavior characteristics to build a more complete user behavior portrait and output the user's query behavior.

[0165] It should be noted that the user behavior analysis center in this embodiment represents an integrated system platform for collecting, storing, analyzing and managing user behavior data. Through the user behavior analysis center, we can gain an in-depth understanding of user behavior patterns, interest preferences and interaction history, thereby optimizing the user experience.

[0166] In addition, in another aspect of the present application, in some embodiments, the present application provides a user behavior analysis system based on big data, referring to Figure 4 , which is a schematic diagram of exemplary hardware and / or software of a user behavior analysis system based on big data according to some embodiments of the present application, the user behavior analysis system based on big data 200 includes: an acquisition module 201, a processing module 202 and an execution module 203, which are described as follows:

[0167] Acquisition module 201, in this application, acquisition module 201 is mainly used to obtain the query log of the target user on the Internet search service platform from the big data center of the Internet search service platform;

[0168] Processing module 202, in this application, processing module 202 is mainly used to extract multiple topic tags when the target user performs a query and long-tail query data of each topic tag from the query log;

[0169] The processing module 202 is further configured to select a topic tag as a selected topic tag, determine multiple behavioral interest levels of the target user when performing a long-tail query under the selected topic tag based on the long-tail query data of the selected topic tag, and further determine multiple query interest sub-tags of the selected topic tag based on the query interest index of the Internet search service platform and all the behavioral interest levels;

[0170] In addition, the processing module 202 is further used to associate each query interest sub-tag with the selected topic tag for query directionality, thereby obtaining multiple directionality association vectors corresponding to the selected topic tag, determining the behavior trend characteristics of the target user when performing long-tail queries under the selected topic tag based on all the directionality association vectors, and continuing to determine the behavior trend characteristics of the target user when performing long-tail queries under the remaining topic tags;

[0171] In addition, the processing module 202 is also used to obtain the query phrase input by the target user on the Internet search service platform at the current moment, and perform query matching on the query phrase with the behavior trend characteristics under each topic tag to obtain the query behavior characteristics of the target user;

[0172] Execution module 203, in this application, execution module 203 is mainly used to upload the query behavior characteristics of the target user to the user behavior analysis center.

[0173] In addition, the present application also provides a computer device, which includes a memory and a processor, the memory stores code, and the processor is configured to obtain the code and execute the above-mentioned user behavior analysis method based on big data.

[0174] In some embodiments, reference Figure 5 , which is a schematic diagram of the structure of a computer device for implementing a user behavior analysis method based on big data according to some embodiments of the present application. The user behavior analysis method based on big data in the above embodiment can be Figure 5 The computer device 300 shown in the figure is implemented, and the computer device 300 includes at least one processor 301, a communication bus 302, a memory 303 and at least one communication interface 304.

[0175] Processor 301 can be a general-purpose central processing unit (CPU), or an application-specific integrated circuit (ASIC) or one or more for controlling the execution of the user behavior analysis method based on big data in the present application.

[0176] The communication bus 302 may be used to transmit information between the above-mentioned components.

[0177] The memory 303 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a disk or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0178] Among them, the memory 303 is used to store the program code for executing the solution of the present application, and the execution is controlled by the processor 301. The processor 301 is used to execute the program code stored in the memory 303. The program code may include one or more software modules. The determination of the user behavior analysis method based on big data in the above embodiment can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.

[0179] The communication interface 304 uses any transceiver or other device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0180] In a specific implementation, as an embodiment, a computer device may include multiple processors, each of which may be a single-CPU processor or a multi-CPU processor. The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0181] The above-mentioned computer device can be a general-purpose computer device or a special-purpose computer device. In a specific implementation, the computer device can be a desktop computer, a portable computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device or an embedded device. The embodiment of the present application does not limit the type of computer device.

[0182] In addition, the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the above-mentioned user behavior analysis method based on big data is implemented.

[0183] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0184] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A user behavior analysis method based on big data, characterized in that: The steps include: Obtaining the query log of the target user on the Internet search service platform from the big data center of the Internet search service platform; Extracting multiple topic tags when the target user performs a query and long-tail query data of each topic tag from the query log; A topic tag is selected as the selected topic tag, and multiple behavioral interests of the target user when performing long-tail queries under the selected topic tag are determined based on the long-tail query data of the selected topic tag, and then multiple query interest subtags of the selected topic tag are determined based on the query interest index of the Internet search service platform and all the behavioral interests; Associating each query interest sub-tag with the selected topic tag for query directionality, thereby obtaining multiple directionality association vectors corresponding to the selected topic tag, determining the behavior trend characteristics of the target user when performing long-tail queries under the selected topic tag based on all directionality association vectors, and continuing to determine the behavior trend characteristics of the target user when performing long-tail queries under the remaining topic tags; Obtaining the query phrase entered by the target user on the Internet search service platform at the current moment, performing query matching on the query phrase and the behavior trend characteristics under each topic tag, and obtaining the query behavior characteristics of the target user; The query behavior characteristics of the target user are uploaded to the user behavior analysis center.

2. The method according to claim 1, characterized in that Extracting multiple topic tags when a target user performs a query and long-tail query data of each topic tag from the query log specifically includes: Extracting multiple query keywords from the query log; All query keywords are clustered to obtain multiple topic tags when the target user performs a query; Selecting a topic tag as a selected topic tag, and combining long-tail query records corresponding to each query keyword in the selected topic tag in the query log into long-tail query data of the selected topic tag; Continue to identify long-tail query data for the remaining hashtags.

3. The method according to claim 2, characterized in that All query keywords are clustered by tags to obtain multiple topic tags when the target user performs a query, including: Determine the semantic similarity between each query keyword; Based on the semantic similarity between each query keyword, a clustering algorithm is used to group all query keywords by similarity to obtain multiple query similarity groups; The topic labels of each query similarity group are determined, thereby obtaining multiple topic labels when the target user performs a query.

4. The method according to claim 1, characterized in that Determine the target user's multiple behavioral interests when performing long-tail queries under the selected topic tags based on the long-tail query data of the selected topic tags, including: Select a long-tail query term from the long-tail query data of the selected topic tag as the selected long-tail query term; Determining the behavior pattern frequency of the selected long-tail query term based on the long-tail query data; Determining an interest weight of a selected long-tail query term in the long-tail query data; Determining the behavioral interest of the target user when searching for the selected long-tail query term under the selected topic tag according to the behavioral pattern frequency and the interest weight; Continue to determine the target users' behavioral interest in the remaining long-tail search terms under the selected topic tag.

5. The method according to claim 1, characterized in that The multiple query interest sub-tags of the selected topic tag are determined based on the query interest index and all behavioral interest of the Internet search service platform, including: Obtaining the query interest index of the Internet search service platform; Performing query behavior analysis on the selected topic tag based on the query interest index and each behavior interest degree to obtain multiple query behavior viscosities of the selected topic tag; Multiple query interest sub-tags of the selected topic tag are determined according to the viscosity of all query behaviors.

6. The method according to claim 1, characterized in that The query directionality association of each query interest sub-tag with the selected topic tag is performed, and then multiple directionality association vectors corresponding to the selected topic tag are obtained, specifically including: Selecting a query interest sub-tag as a selected query interest sub-tag; Determine the tag co-occurrence rate between the selected query interest subtag and the selected topic tag in the historical search records of the target user; Performing a directivity analysis on the selected query interest subtag to obtain the associated directivity of the selected query interest subtag under the selected topic tag; Obtain the tag responsiveness of the target user to the selected query interest subtag; Determine a directional association vector of the selected query interest sub-tag to the selected topic tag based on the tag co-occurrence rate, the association directivity and the tag responsiveness; Continue to determine a plurality of pointing association vectors corresponding to the selected topic label.

7. The method according to claim 1, characterized in that The behavioral trend characteristics of the target user when performing long-tail queries under the selected topic tags are determined based on all pointing association vectors, including: determining multiple association similarities of all pointing association vectors; Perform feature decomposition on all associated similarities to obtain multiple behavioral trends and multiple behavioral features under the selected topic tags; Based on all behavioral trends and all behavioral features, the behavioral trend features of the target user when performing long-tail queries under the selected topic tags are determined.

8. A user behavior analysis system based on big data, characterized in that: include: An acquisition module is used to acquire the query log of the target user on the Internet search service platform from the big data center of the Internet search service platform; A processing module, used to extract multiple topic tags when a target user performs a query and long-tail query data of each topic tag from the query log; The processing module is further used to select a topic tag as a selected topic tag, determine multiple behavioral interest levels of the target user when performing a long-tail query under the selected topic tag based on the long-tail query data of the selected topic tag, and then determine multiple query interest sub-tags of the selected topic tag based on the query interest index of the Internet search service platform and all the behavioral interest levels; The processing module is further used to associate each query interest sub-tag with the selected topic tag for query directionality, thereby obtaining multiple directionality association vectors corresponding to the selected topic tag, determining the behavior trend characteristics of the target user when performing long-tail queries under the selected topic tag based on all the directionality association vectors, and continuing to determine the behavior trend characteristics of the target user when performing long-tail queries under the remaining topic tags; The processing module is further used to obtain the query phrase input by the target user on the Internet search service platform at the current moment, and perform query matching on the query phrase with the behavior trend characteristics under each topic tag to obtain the query behavior characteristics of the target user; The execution module is used to upload the query behavior characteristics of the target user to the user behavior analysis center.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores codes, and the processor is configured to obtain the codes and execute the user behavior analysis method based on big data as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the user behavior analysis method based on big data as described in any one of claims 1 to 7 is implemented.

Images