A search engine based user behavior analysis method and device

Abstract

The application relates to the technical field of data analysis, in particular to a user behavior analysis method and device based on a search engine, which comprises the following steps: determining search input information of a user in a target search engine in a preset time period; inputting the search input information into a trained user preference model to obtain user hot information and user cold information predicted by the user preference model; inputting the user hot information and the user cold information into a trained user behavior analysis model to obtain search display information, search associated terms and associated display information associated with the search associated terms predicted by the user behavior analysis model; and showing recommended information to the user based on the search display information, the search associated terms and the associated display information. The application effectively improves the browsing experience of the user, thereby optimizing the recommended result of the engine and meeting the personalized needs of the user.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to a method and apparatus for user behavior analysis based on a search engine. Background Technology

[0002] To obtain more useful data, operators need to analyze the behavior of each user to ensure a more comfortable user experience and more accurate recommendations in subsequent product interactions, thereby improving the user experience.

[0003] Users input search query descriptions, such as keywords or images, and search engines return search results to users based on these descriptions. However, current technologies rely on relatively simplistic user behavior analysis, only filtering keywords that users prefer based on their historical search results. They fail to consider potential information security risks such as spam and fraudulent information. This not only fails to provide better information recommendations but may also lead to information security issues. Summary of the Invention

[0004] Therefore, it is necessary to provide a user behavior analysis method and device based on a search engine to address the shortcomings of existing user behavior analysis methods, which may lead to information security issues in order to provide better information recommendations to users.

[0005] To solve the problems of the prior art, the technical solution adopted by the present invention is as follows:

[0006] This invention provides a user behavior analysis method based on a search engine, the method comprising:

[0007] Determine the search input information of users on the target search engine within a preset time period;

[0008] The search input information is input into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model;

[0009] The user hot information and the user cold information are input into a trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value. The popularity is used to characterize the degree of information recommendation.

[0010] Recommended information is displayed to the user based on the search display information, the search related terms, and the related display information.

[0011] Optionally, the user preference model is trained through the following steps:

[0012] Determine a sample access information list of users on the target search engine; the sample access information list contains user access information within a sample time period;

[0013] Determine the sample return visit information list in the sample access information list; the sample return visit information list includes the website dwell time, return visit information, number of return visits, and time interval between return visits within the sample time period;

[0014] Determine the user's sample search input information and the engine information of the target search engine;

[0015] Using the sample access information list, the sample return visit information list, the sample search input information, and the engine information of the target search engine as input data for training, a user preference model for predicting user hot information and user cold information is obtained by clustering.

[0016] Optionally, the user behavior analysis model is trained through the following steps:

[0017] The hot spot information and cold spot information of the samples are input into the target feature extraction model to extract the feature information from the hot spot information and cold spot information;

[0018] Based on the matching results of the feature information and the historical database, the sample hotspot information and sample coldspot information are determined as sample danger information; the historical database stores the reporting information of several target search engines;

[0019] Based on the matching results between the feature information and the historical association database, sample association information in the sample hotspot information and the sample coldspot information is determined; the historical association database stores jump entries based on user search input information based on samples;

[0020] Using the sample hotspot information, sample cold information, sample danger information, and sample association information as input data for training, a clustering method is employed to obtain the user behavior analysis model used to predict the search display information, the search related terms, and the related display information.

[0021] Optionally, determining the sample danger information in the sample hotspot information and the sample cold information based on the matching of the feature information with the historical database specifically includes:

[0022] Determine the characteristic information in the reported information;

[0023] The feature information in the hot and cold information of the samples is compared and matched with the feature information in the report information;

[0024] Once a match is found, the corresponding hotspot information / coldspot information of the sample is marked as the dangerous information of the sample.

[0025] Optionally, determining the sample association information in the sample hotspot information and the sample coldspot information based on the matching results of the feature information and the historical association database specifically includes:

[0026] Determine the feature information in the jump terms;

[0027] The feature information in the hot and cold information of the samples is compared and matched with the feature information in the jump terms;

[0028] Once a match is confirmed, the redirected term is used as the corresponding sample association information for the hot / cold sample information.

[0029] Optionally, the method further includes the following steps:

[0030] If dangerous information is identified when a user browses the search display information and the associated display information, an alarm message is generated and displayed to the user; the dangerous information is information that poses a security risk.

[0031] The present invention also provides a user behavior analysis device based on a search engine, the device comprising:

[0032] The input determination module is used to determine the user's search input information on the target search engine within a preset time period;

[0033] The first prediction module is used to input the search input information into a trained user preference model to obtain user hot information and user cold information predicted by the user preference model; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value, and the popularity is used to characterize the recommendation degree of the information;

[0034] The second prediction module is used to input the user hot information and the user cold information into the trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms;

[0035] The third prediction module is used to display recommended information to the user based on the search display information, the search related terms, and the related display information.

[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of any of the above-described search engine-based user behavior analysis methods.

[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-described search engine-based user behavior analysis methods.

[0038] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of any of the above-described search engine-based user behavior analysis methods.

[0039] The advantages of this application compared to the prior art are:

[0040] Based on user behavior data, a user preference model and a user behavior analysis model are established. Both models include the correspondence between user search history data and user behavior, avoiding the incomplete analysis caused by using a single recommendation criterion to analyze user behavior. This improves the accuracy of the recommendation results presented to users, thereby enhancing the accuracy of the recommended content. By analyzing past user behavior, user interest feature patterns can be derived. Cluster centers are compared with the categories to which user interests belong, and clusters belonging to categories to which users are interested are ranked first. Other cluster results are ranked in descending order of their relevance to user interest, effectively improving the user browsing experience and optimizing the engine's recommendation results to meet users' personalized needs. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart illustrating the user behavior analysis method based on a search engine according to the present invention.

[0043] Figure 2 This is a schematic diagram of the user behavior analysis device based on a search engine according to the present invention;

[0044] Figure 3 This is a schematic diagram of the electronic device provided by the present invention. Detailed Implementation

[0045] To further understand the features, technical means, and specific objectives and functions achieved by the present invention, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0046] As a preferred embodiment of the present invention, the present invention provides a user behavior analysis method based on a search engine, the method comprising:

[0047] Determine the search input information of users on the target search engine within a preset time period;

[0048] The search input information is input into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model;

[0049] The user hot information and the user cold information are input into a trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value. The popularity is used to characterize the degree of information recommendation.

[0050] Recommended information is displayed to the user based on the search display information, the search related terms, and the related display information.

[0051] This invention also provides a user behavior analysis device based on a search engine, the device comprising:

[0052] The input determination module is used to determine the user's search input information on the target search engine within a preset time period;

[0053] The first prediction module is used to input the search input information into a trained user preference model to obtain user hot information and user cold information predicted by the user preference model; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value, and the popularity is used to characterize the recommendation degree of the information;

[0054] The second prediction module is used to input the user hot information and the user cold information into the trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms;

[0055] The third prediction module is used to display recommended information to the user based on the search display information, the search related terms, and the related display information.

[0056] The present invention provides a user behavior analysis method and apparatus based on a search engine. Based on user behavior data, it establishes a user preference model and a user behavior analysis model. Both models include the correspondence between user search history data and user behavior, avoiding the incomplete analysis caused by using a single recommendation condition to analyze user behavior. This improves the accuracy of the recommended content displayed to users. By analyzing past user behavior, it derives user interest feature patterns, compares cluster centers with the categories to which user interests belong, and ranks the clusters of categories the user is interested in at the top. Other cluster results are ranked in descending order of their relevance to the user's interest, thereby effectively improving the user's browsing experience and optimizing the engine's recommendation results to meet the user's personalized needs.

[0057] The preferred embodiments of the user behavior analysis method and apparatus based on a search engine of the present invention will be described below with reference to the accompanying drawings.

[0058] Please see Figure 1 This invention provides a user behavior analysis method based on a search engine, the method comprising:

[0059] S10. Determine the user's search input information on the target search engine within a preset time period.

[0060] First, the user searches for input information in the target search engine within a preset time period. The input information can be text, image, or voice. Both text and voice information can include keywords.

[0061] S20. Input the search input information into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model.

[0062] Hot user information and cold user information are both information that users are predicted to browse and that can be displayed through the target search engine.

[0063] In this method, the popularity of user hot information exceeds a preset value, while the popularity of user cold information does not exceed a preset value. Popularity is used to characterize the degree of information recommendation. More specifically, popularity is generated based on historical information such as the user's historical website dwell time, return visit information, number of return visits, and time interval between return visits. This reflects the user's interests, domain, and search tendencies. By summarizing the user's interests, domain, and search tendencies, the user's historical behavioral characteristics can be derived, and user information can be distinguished through popularity.

[0064] Understandably, the preset value is a threshold related to popularity, or a preset popularity value. User hotspot information represents web page information that a user is highly likely to browse. The higher the popularity value of user hotspot information, the higher the probability, which is predicted based on the user's previous behaviors. User hotspot information also represents web page information that a user is unlikely to browse. However, for different users, a low probability of browsing a web page does not mean they will not browse it, because each user has unique individual needs. Only some niche information can satisfy a user's individual needs. Directly ignoring this potential information would lead to the inability to meet the usage needs of some users.

[0065] S30. Input user hot information and user cold information into the trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, search related terms, and related display information associated with search related terms.

[0066] In this application, the search display information is the content generated and displayed on the target search engine based on the user's current search input. Depending on the target search engine's basic settings, page layout, etc., it can be displayed in pages. In this embodiment, the search display information is sorted based on popularity, and in descending order. That is, the higher the popularity value, the earlier it appears in the list, requiring fewer page turns; conversely, the lower the popularity value, the later it appears in the list, requiring more page turns.

[0067] Search keyword entries are generated along with search display information, for example, below the input field of the target search engine. Search display information can be displayed below the search keyword entries. Each search keyword entry corresponds to at least one related display message. When a user triggers a search keyword entry, for example, by clicking on one of the search keyword entries, they are redirected to another set of display pages (at least one page). This set of display pages is used to display related display information. This set of display pages is similar to the webpage displaying search display information. The related display information can also be displayed in pages according to the target search engine's basic settings, page layout, etc. In this embodiment, the related display information is also sorted based on popularity, and in descending order. That is, the higher the popularity value, the earlier it appears in the list, requiring fewer page turns; the lower the popularity value, the later it appears in the list, requiring more page turns.

[0068] S40. Display recommended information to users based on search display information, related search terms, and related display information.

[0069] After a user enters search information into the target search engine, the webpage of the browser that runs the target search engine will generate and display search results, related search terms, and related information associated with those terms. When the user triggers the search results or related information, the specific content will be displayed as recommended information for the user to browse.

[0070] The user behavior analysis method based on a search engine of this invention establishes a user preference model and a user behavior analysis model based on user behavior data. Both models include the correspondence between user search history data and user behavior, avoiding the incomplete analysis caused by using a single recommendation condition to analyze user behavior. This improves the accuracy of the recommendation results displayed to users, thereby enhancing the accuracy of the recommended content. By analyzing past user behavior, user interest feature patterns can be derived. Cluster centers are compared with the categories to which user interests belong, and the clusters to which the user's interests belong are ranked first. Other cluster results are ranked in descending order of their relevance to the user's interests, thus effectively improving the user's browsing experience and optimizing the engine's recommendation results to meet the user's personalized needs.

[0071] In this application, the user preference model is trained through the following steps:

[0072] A10. Determine a sample access information list of users on the target search engine. In this method, the sample access information list contains user access information within a sample time period.

[0073] A20. Determine the sample return visit information list in the sample access information list. In this method, the sample return visit information list includes the website dwell time, return visit information, number of return visits, and time interval between return visits within the sample time period.

[0074] A30. Determine the user's sample search input information and the target search engine's engine information.

[0075] A40. Using the sample access information list, sample return visit information list, sample search input information, and target search engine engine information as input data for training, a user preference model for predicting user hot information and user cold information is obtained by using clustering.

[0076] When building a user preference model, the first step is to determine a list of sample access information for users based on their historical visits (triggered information) on the target search engine, and then obtain a list of sample return visit information. Next, based on the obtained sample search input information and the target search engine's engine information (including basic information about the target search engine), user behavior is learned. Based on information such as website dwell time, return visit information, the number of return visits, and the time interval between return visits, clustering is used to classify the displayed information generated based on the sample search input information into user hot information and user cold information.

[0077] In this application, the user behavior analysis model is trained through the following steps:

[0078] B10. Input the hot and cold information of the samples into the target feature extraction model to extract the feature information from the hot and cold information.

[0079] B20. Based on the matching of feature information with historical database, the dangerous information of samples in sample hotspot information and sample cold information is determined. In this method, the historical database stores the reporting information of several target search engines.

[0080] B30. Based on the matching of feature information and historical association database, determine the sample association information of hot and cold sample information. In this method, the historical association database stores the jump terms of user search input information based on samples.

[0081] B40. Using sample hotspot information, sample cold information, sample dangerous information, and sample association information as input data for training, a clustering method is used to obtain a user behavior analysis model for predicting search display information, search related terms, and related display information.

[0082] Feature extraction of hot and cold information from samples can be achieved through keyword extraction and word segmentation, improving the accuracy of feature extraction and reducing excessive search results irrelevant to the user's search query, thus facilitating user browsing. By analyzing past user behavior to derive user interest patterns, clusters belonging to categories of user interest are ranked first, with other clusters arranged in descending order of their relevance to the user's interest, effectively enhancing the user's browsing experience.

[0083] More specifically, step B20 includes:

[0084] B21. Identify the key features in the reported information.

[0085] In some cases, when other users or the user feels that the content recommended by the target search engine contains inappropriate or fraudulent content, they will report the aforementioned content, thereby obtaining the report information. Similarly, feature information is obtained by extracting features from the report information through methods such as extracting keywords and word segmentation information.

[0086] B22. Compare and match the feature information in the hot and cold information of the samples with the feature information in the report information.

[0087] B23. Once a match is found, mark the corresponding hot / cold sample information as dangerous sample information.

[0088] If the feature information in the sample information, including hot and cold sample information, is highly similar to the feature information in the reported information, then these matching hot and cold sample information are marked as dangerous sample information.

[0089] More specifically, step B30 includes:

[0090] B31. Determine the feature information in the jump entry.

[0091] Currently, most search engines generate not only recommended information but also relevant redirect keywords after a user enters relevant information.

[0092] B32. Compare and match the feature information in the hot and cold information of the samples with the feature information in the jump terms.

[0093] Step B32 is similar to step B22, both involving matching based on feature information.

[0094] B33. Once a match is confirmed, the redirected term is used as the sample association information for the corresponding hot / cold information.

[0095] Step B33 involves matching sample information that is not marked as dangerous sample information. If the feature information in the sample information, including hot sample information and cold sample information, is highly similar to the feature information of the jump term, then these matching hot sample information and cold sample information are marked as sample association information.

[0096] The method also includes the following steps:

[0097] S50. Determine the dangerous information in the user's browsing search display information and related display information, generate alarm information and display it to the user. In this method, dangerous information refers to information that poses a security risk.

[0098] Dangerous information is derived from sample dangerous information and refers to spam, fraudulent information, and information that poses security risks during payment. When a user triggers a search or related information display, if the information is marked as dangerous, an alert is generated and displayed to the user, thereby preventing security issues on the information on the smart terminal.

[0099] The following describes the user behavior analysis device based on a search engine provided by the present invention. The user behavior analysis device based on a search engine described below can be referred to in correspondence with the user behavior analysis method based on a search engine described above.

[0100] Please see Figure 2 The present invention provides a user behavior analysis device based on a search engine, the device comprising:

[0101] The input determination module 10 is used to determine the user's search input information on the target search engine within a preset time period.

[0102] First, the user searches for input information in the target search engine within a preset time period. The input information can be text, image, or voice. Both text and voice information can include keywords.

[0103] The first prediction module 20 is used to input the search input information into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model.

[0104] Hot user information and cold user information are both information that users are predicted to browse and that can be displayed through the target search engine.

[0105] In this device, the popularity of user-popular information exceeds a preset value, while the popularity of user-unpopular information does not exceed a preset value. Popularity is used to characterize the degree of information recommendation. More specifically, popularity is generated based on historical information such as the user's historical website dwell time, return visit information, number of return visits, and time interval between return visits. This reflects the user's interests, domain, and search tendencies. By summarizing the user's interests, domain, and search tendencies, the user's historical behavioral characteristics can be derived, and user information can be differentiated through popularity.

[0106] Understandably, the preset value is a threshold related to popularity, or a preset popularity value. User hotspot information represents web page information that a user is highly likely to browse. The higher the popularity value of user hotspot information, the higher the probability, which is predicted based on the user's previous behaviors. User hotspot information also represents web page information that a user is unlikely to browse. However, for different users, a low probability of browsing a web page does not mean they will not browse it, because each user has unique individual needs. Only some niche information can satisfy a user's individual needs. Directly ignoring this potential information would lead to the inability to meet the usage needs of some users.

[0107] The second prediction module 30 is used to input user hot information and user cold information into the trained user behavior analysis model to obtain the search display information, search related terms, and related display information associated with the search related terms predicted by the user behavior analysis model.

[0108] In this application, the search display information is the content generated and displayed on the target search engine based on the user's current search input. Depending on the target search engine's basic settings, page layout, etc., it can be displayed in pages. In this embodiment, the search display information is sorted based on popularity, and in descending order. That is, the higher the popularity value, the earlier it appears in the list, requiring fewer page turns; conversely, the lower the popularity value, the later it appears in the list, requiring more page turns.

[0109] Search keyword entries are generated along with search display information, for example, below the input field of the target search engine. Search display information can be displayed below the search keyword entries. Each search keyword entry corresponds to at least one related display message. When a user triggers a search keyword entry, for example, by clicking on one of the search keyword entries, they are redirected to another set of display pages (at least one page). This set of display pages is used to display related display information. This set of display pages is similar to the webpage displaying search display information. The related display information can also be displayed in pages according to the target search engine's basic settings, page layout, etc. In this embodiment, the related display information is also sorted based on popularity, and in descending order. That is, the higher the popularity value, the earlier it appears in the list, requiring fewer page turns; the lower the popularity value, the later it appears in the list, requiring more page turns.

[0110] The third prediction module 40 is used to display recommended information to users based on search display information, search related terms, and related display information.

[0111] After a user enters search information into the target search engine, the webpage of the browser that runs the target search engine will generate and display search results, related search terms, and related information associated with those terms. When the user triggers the search results or related information, the specific content will be displayed as recommended information for the user to browse.

[0112] The user behavior analysis device based on a search engine of the present invention establishes a user preference model and a user behavior analysis model based on user behavior data. Both models include the correspondence between user search history data and user behavior, avoiding the problem of incomplete analysis caused by using a single recommendation condition to analyze user behavior. This improves the accuracy of the recommendation results displayed to users and enhances the accuracy of the recommended content. By analyzing the user's past behavior, user interest feature patterns can be derived. Cluster centers are compared with the categories to which the user's interests belong, and the clusters to which the user's interests belong are ranked first. Other cluster results are ranked in descending order of their relevance to the user's interests, thereby effectively improving the user's browsing experience and optimizing the engine's recommendation results to meet the user's personalized needs.

[0113] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical commands stored in the memory 830 to execute a user behavior analysis method based on a search engine, the method including:

[0114] Determine the search input information of users on the target search engine within a preset time period;

[0115] The search input information is input into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model;

[0116] The user hot information and the user cold information are input into a trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value. The popularity is used to characterize the degree of information recommendation.

[0117] Recommended information is displayed to the user based on the search display information, the search related terms, and the related display information.

[0118] Furthermore, the logical commands in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several commands to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0119] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute the user behavior analysis method based on a search engine provided by the above methods, the method comprising:

[0120] Determine the search input information of users on the target search engine within a preset time period;

[0121] The search input information is input into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model;

[0122] The user hot information and the user cold information are input into a trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value. The popularity is used to characterize the degree of information recommendation.

[0123] Recommended information is displayed to the user based on the search display information, the search related terms, and the related display information.

[0124] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the search engine-based user behavior analysis method provided by the methods described above, the method comprising:

[0125] Determine the search input information of users on the target search engine within a preset time period;

[0126] The search input information is input into the trained user preference model to obtain the user hotspot information and user coldspot information predicted by the user preference model;

[0127] The user hot information and the user cold information are input into a trained user behavior analysis model to obtain the search display information predicted by the user behavior analysis model, the search related terms, and the related display information associated with the search related terms; the popularity of the user hot information exceeds a preset value, and the popularity of the user cold information does not exceed a preset value. The popularity is used to characterize the degree of information recommendation.

[0128] Recommended information is displayed to the user based on the search display information, the search related terms, and the related display information.

[0129] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0130] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several commands to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A search engine based user behavior analysis method, characterized by, The method comprises: determining search input information of a user in a target search engine in a preset time period; inputting the search input information into a trained user preference model to obtain user hot information and user cold information predicted by the user preference model; inputting the user hot information and the user cold information into a trained user behavior analysis model to obtain search display information, search associated terms and associated display information associated with the search associated terms predicted by the user behavior analysis model; the hotness of the user hot information exceeds a preset value, and the hotness of the user cold information does not exceed a preset value, and the hotness is used to represent the recommendation degree of information; showing recommended information to the user based on the search display information, the search associated terms and the associated display information; The user behavior analysis model is trained by the following steps: inputting sample hot information and sample cold information into a target feature extraction model to extract feature information in the sample hot information and the sample cold information; determining sample dangerous information in the sample hot information and the sample cold information based on the matching of the feature information and a historical database; the historical database stores report information of a plurality of target search engines; determining sample associated information in the sample hot information and the sample cold information based on the matching of the feature information and a historical association library; the historical association library stores jump terms based on sample search input information of a user; using the sample hot information, the sample cold information, the sample dangerous information and the sample associated information as input data for training, and using a clustering method to obtain the user behavior analysis model for predicting the search display information, the search associated terms and the associated display information; wherein, based on the matching of the feature information and the historical database, determining the sample dangerous information in the sample hot information and the sample cold information, specifically comprising: determining the feature information in the report information; comparing and matching the feature information in the sample hot information and the sample cold information with the feature information in the report information; determining that the corresponding sample hot information / sample cold information can be matched, and marking the corresponding sample hot information / sample cold information as the sample dangerous information; based on the matching of the feature information and the historical association library, determining the sample associated information in the sample hot information and the sample cold information, specifically comprising: determining the feature information in the jump terms; comparing and matching the feature information in the sample hot information and the sample cold information with the feature information in the jump terms; determining that the jump terms can be matched, and taking the jump terms as the sample associated information of the corresponding sample hot information / sample cold information.

2. The method of claim 1, wherein, The user preference model is trained by the following steps: determining a sample access information list of a user in a target search engine; the sample access information list contains user access information in a sample time period; determining a sample revisit information list in the sample access information list; the sample revisit information list includes a website staying time, revisit information, revisit times of the revisit information, and revisit interval time of the website in a sample time period; determining sample search input information of a user and engine information of the target search engine; using the sample access information list, the sample revisit information list, the sample search input information, and the engine information of the target search engine as input data for training, and using a clustering method to obtain the user preference model for predicting user hot information and user cold information.

3. The search engine based user behavior analysis method of claim 1, wherein, The method further includes the following steps: determining dangerous information in the search display information and the associated display information browsed by the user, generating an alarm information, and displaying the alarm information to the user; the dangerous information is information with a security risk.

4. A search engine based user behavior analysis apparatus for implementing the search engine based user behavior analysis method according to any one of claims 1 to 3, characterized in that, The device includes: an input determination module configured to determine search input information of a user in a target search engine in a preset time period; a first prediction module configured to input the search input information into a trained user preference model to obtain user hot information and user cold information predicted by the user preference model; the hotness of the user hot information exceeds a preset value, the hotness of the user cold information does not exceed the preset value, and the hotness is used to represent a recommendation degree of information; a second prediction module configured to input the user hot information and the user cold information into a trained user behavior analysis model to obtain search display information, search associated terms, and associated display information associated with the search associated terms predicted by the user behavior analysis model; a third prediction module configured to display recommendation information to a user based on the search display information, the search associated terms, and the associated display information.

5. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the search engine-based user behavior analysis method according to any one of claims 1 to 3 when executing the program. 6.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program implements the steps of the search engine-based user behavior analysis method according to any one of claims 1 to 3 when executed by the processor.

7. A computer program product comprising a computer program, characterized in that, The computer program implements the steps of the search engine-based user behavior analysis method according to any one of claims 1 to 3 when executed by the processor.

Images