A method and device for predicting user behavior
By analyzing user short video browsing records and comment area information, generating a collection of entries to be recommended, and updating the blacklist of entries based on user behavior, the problem of inaccurate prediction of user search needs in the existing technology is solved, and more accurate knowledge is achieved Search for recommendations.
Patent Information
- Application Number
- CN202411171062.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-26
AI Technical Summary
The prior art cannot accurately predict the search needs of users when browsing short videos.
By analyzing the user's short video browsing records within the preset time period, determining the user's portrait, and semantic analysis is performed based on the current browsing and comment area information, generating a collection of entries to be recommended, monitoring user behavior in real time, updating the entries blacklist, and filtering and expand the entries to be recommended and generate target recommended entries.
Accurate prediction of user knowledge search needs is achieved, and the recommended entries are more in line with the actual needs of users, improving the accuracy and relevance of recommendations.
Smart Images

Figure CN118690048B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of behavior prediction, and in particular, relates to a method and device for user behavior prediction. Background Art
[0002] The Internet has now entered the era of self-media, and short videos have become the main medium for online social networking. When users browse short videos, they usually initiate a search demand for certain knowledge based on the short videos. However, the existing technology cannot accurately predict the user's search needs based on their behavior. Summary of the invention
[0003] The purpose of the present invention is to provide a method and device for predicting user behavior, so as to solve the technical problem in the prior art that it is impossible to accurately predict the search needs of users based on their behavior.
[0004] A user behavior prediction method, characterized in that the method comprises:
[0005] S1: Determine a first user portrait according to a short video browsing record of the first user within a preset time period;
[0006] The S1 comprises the following sub-steps:
[0007] S11: Acquire A first short videos browsed by the first user within a preset time period, wherein each of the first short videos corresponds one-to-one to a first tag;
[0008] S12: Performing a first frequency analysis on the A first short videos using the first tag to obtain B first short videos and B first viewing times;
[0009] S13: performing cluster analysis on the B first short videos according to the B first viewing times to obtain C first cluster centers, and obtaining the first user portrait based on the C first cluster centers;
[0010] S2: Obtain a second short video currently browsed by the first user, and determine a first semantic analysis result according to the second short video;
[0011] S3: Obtain D pieces of first comment area information of the second short video, and determine a second semantic analysis result according to the D pieces of first comment area information;
[0012] S4: generating a first set of terms to be recommended according to the first semantic analysis result and the second semantic analysis result;
[0013] S5: monitoring the browsing behavior of the first user on each piece of the first comment area information in real time, and generating a second set of terms to be recommended according to the semantic analysis result of the first comment area information;
[0014] S6: generating a third set of terms to be recommended based on the first set of terms to be recommended and the second set of terms to be recommended, filtering the third set of terms to be recommended using the first user profile and the first term blacklist to obtain target recommended terms;
[0015] S7: According to the second operation of the first user on the target recommended term I, the first term blacklist is updated;
[0016] Among them, A, B, C, D and I are all natural numbers greater than 0.
[0017] Preferably, the S13 comprises the following sub-steps:
[0018] S131: performing semantic analysis on the B first short videos one by one to obtain B first semantic analysis information;
[0019] S132: weighting the B pieces of first semantic analysis information based on the B first viewing times to obtain B pieces of first weight information corresponding to the B first short videos;
[0020] S133: performing cluster analysis on the B first short videos according to the B first weight information and the B first semantic analysis information to obtain C first cluster centers;
[0021] S134: Determine a first user portrait based on the C first cluster centers.
[0022] Preferably, S2 comprises the following sub-steps:
[0023] S21: performing a first context analysis on the second short video to obtain a first context analysis result;
[0024] S22: performing a first picture analysis on the second short video to obtain a first picture analysis result;
[0025] S23: Determine a first semantic analysis result according to the first context analysis result and the first picture analysis result.
[0026] Preferably, the S21 includes the following sub-steps:
[0027] S211: extracting first text information and first subtitle information from the second short video;
[0028] S212: Inputting the first text information and the first subtitle information into a first context analysis model to obtain the first context analysis result.
[0029] Preferably, S3 includes the following sub-steps:
[0030] S31: Obtain D pieces of first comment area information corresponding to the second short video;
[0031] S32: performing semantic clustering analysis on the D pieces of information in the first comment area to obtain E second clustering centers;
[0032] S33: Determine F second word information based on the E second cluster centers, and determine a second semantic analysis result according to the F second word information;
[0033] Wherein, E and F are both natural numbers greater than 0.
[0034] Preferably, S5 comprises the following sub-steps:
[0035] S51: Determine the second comment area information browsed by the first user in real time; wherein the first comment area information browsed by the first user in real time is used as the second comment area information;
[0036] S52: Determine whether the first user has performed a first operation when browsing the information in the second comment area, if not, proceed to S51, if yes, proceed to S53;
[0037] S53: performing semantic expansion analysis on the second comment area information to obtain G basic knowledge information and H expanded search information;
[0038] S54: obtaining a second set of to-be-recommended terms based on the G basic knowledge information and the H expanded search information;
[0039] Wherein, G and H are both natural numbers greater than 0.
[0040] Preferably, the S53 includes the following sub-steps:
[0041] S531: performing semantic analysis on the second comment area information to obtain G third word information;
[0042] S532: taking each of the third term information as a piece of basic knowledge information, thereby determining G pieces of basic knowledge information;
[0043] S533: Perform an extended search for each of the third term information, thereby determining H extended search information.
[0044] Preferably, the S6 comprises the following sub-steps:
[0045] S61: Integrate the first set of terms to be recommended and the second set of terms to be recommended to obtain a third set of terms to be recommended;
[0046] S62: Calculate the first similarity between each term in the third set of terms to be recommended and the first user portrait, and delete the terms whose first similarity is less than a first preset value from the third set of terms to be recommended to obtain a fourth set of terms to be recommended;
[0047] S63: Calculate the second similarity between each entry in the fourth set of recommended terms and the first set of blacklisted terms, and delete the entries whose second similarity is greater than the second preset value from the fourth set of recommended terms to obtain a fifth set of recommended terms;
[0048] S64: Determine I target recommended term according to the fifth set of terms to be recommended;
[0049] Wherein, I is a natural number greater than 0.
[0050] Preferably, the S7 comprises the following sub-steps:
[0051] S71: Acquire a second operation of the first user;
[0052] S72: if the second operation is a click operation on the first target recommended term, then the process proceeds to S73; if the second operation is a click operation on a "Change Batch" button, then the process proceeds to S74;
[0053] S73: calculating the third similarities between the first target recommended terms and the first target recommended terms respectively, and adding the target recommended terms whose third similarities are less than a third preset value to the first term blacklist;
[0054] S74: Add all the target recommended terms to the first term blacklist.
[0055] The present application also proposes a user behavior prediction system for implementing the above-mentioned user behavior prediction method.
[0056] The present application proposes a method and device for predicting user behavior. In order to accurately predict the knowledge search needs generated by users during the short video browsing process, the short video currently browsed by the user is first analyzed in two dimensions, namely, the meaning and the picture, so as to obtain a first semantic analysis result, and the information in the comment area is analyzed to obtain a second semantic analysis result, so as to obtain a first set of terms to be recommended. Then, based on the user's specific operation on a specific comment, basic knowledge information and expanded search information are determined, so as to obtain a second set of terms to be recommended. Finally, the first term blacklist is updated based on the user's operation information on the target recommended term. The present application determines the terms to be recommended from multiple dimensions, further expands the terms to be recommended according to the user's specific operation information, and filters the recommended terms based on the user's browsing profile and the term blacklist, so as to make the recommended terms more in line with the user's actual needs. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.
[0058] Figure 1 It is an execution flow chart of a user behavior prediction method of the present invention. DETAILED DESCRIPTION
[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0060] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, wherein the illustrative embodiments and descriptions are only used to explain the present invention but are not intended to limit the present invention.
[0061] A method and device for predicting user behavior of the present invention are described in detail below.
[0062] This embodiment proposes a user behavior prediction method for predicting the user's operation behavior in a short video APP. The method flow is as follows: Figure 1 As shown, the specific steps include:
[0063] S1: Determine a first user portrait based on the short video browsing record of the first user within a preset time period.
[0064] The first user refers to a user for whom user behavior prediction is to be performed.
[0065] The S1 comprises the following sub-steps:
[0066] S11: Acquire A first short videos viewed by the first user within a preset time period, wherein each of the first short videos corresponds one-to-one to a first tag.
[0067] The preset time period refers to the time period when the user is relatively active on the short video platform. For example, if the first user is only active on the short video platform for one hour a day, the preset time period can be taken as the one hour when the first user is active on the short video platform. If the first user is active on the short video platform for more than six hours a day, the preset time period can be taken as the six hours when the first user is active on the short video platform.
[0068] S12: Perform a first frequency analysis on the A first short videos using the first tag to obtain B first short videos and B first viewing times.
[0069] When a user browses short videos on a short video platform, if he encounters a video of interest, he may browse it repeatedly for multiple times and may even play it back. Therefore, statistical analysis is performed on the playback frequency of the first short video of the first user to further analyze and derive the browsing preference of the first user.
[0070] S13: Perform cluster analysis on the B first short videos according to the B first viewing times to obtain C first cluster centers, and obtain the first user portrait based on the C first cluster centers.
[0071] Among them, A, B and C are all natural numbers greater than 0, and A>B>C.
[0072] The cluster analysis is performed based on the semantic information of the B first short videos.
[0073] The S13 comprises the following sub-steps:
[0074] S131: Perform semantic analysis on the B first short videos one by one to obtain B first semantic analysis information.
[0075] S132: weighting the B pieces of first semantic analysis information based on the B first viewing times to obtain B pieces of first weight information corresponding to the B first short videos.
[0076] When the first viewing number is larger, it means that the first short video matches the first user's preference more closely, and therefore the weight is higher; when the first viewing number is smaller, it means that the first short video matches the first user's preference less closely, and therefore the weight is lower.
[0077] The specific weighting rule may assign 1 to the first short video with the largest number of first views, and normalize and weight other first short videos according to their corresponding first view counts.
[0078] S133: Perform cluster analysis on the B first short videos according to the B first weight information and the B first semantic analysis information to obtain C first cluster centers.
[0079] The C first clustering centers are semantic label sets obtained by semantically clustering the B first short videos.
[0080] S134: Determine a first user portrait based on the C first cluster centers.
[0081] Among them, each of the first cluster centers includes a number of semantic tags, and the first user portrait can be determined by extracting the semantic tags whose occurrence frequencies meet the preset requirements.
[0082] The first user portrait is used to characterize the browsing preference information of the first user within a preset time period.
[0083] S2: Obtain a second short video currently browsed by the first user, and determine a first semantic analysis result according to the second short video.
[0084] The first user portrait is obtained based on the short video browsing record of the first user within a preset time period, and can be used as a basic basis for entry recommendation.
[0085] However, users often suddenly have the need to search for certain knowledge after viewing a short video, so it is also necessary to recommend terms based on users' real-time search needs.
[0086] The second short video in this step refers to the most recent short video generated by the first user while browsing short videos.
[0087] The S2 comprises the following sub-steps:
[0088] S21: Perform a first contextual analysis on the second short video to obtain a first contextual analysis result.
[0089] The S21 includes the following sub-steps:
[0090] S211: Extracting first text information and first subtitle information from the second short video.
[0091] The first copy information refers to the copy information corresponding to the second short video, which is usually edited by the publisher of the short video.
[0092] The first subtitle information refers to the subtitle information corresponding to each frame of the short video during the playback of the second short video.
[0093] S212: Inputting the first text information and the first subtitle information into a first context analysis model to obtain the first context analysis result.
[0094] The first context analysis model is obtained based on the training of the convolutional neural network model. In the training sample, the text information and subtitle information of the short video are input data, and the context analysis results displayed in the form of entries are output data. After training the convolutional neural network with the sample data, the first context analysis model can be obtained.
[0095] The first context analysis result includes a plurality of terms.
[0096] S22: Perform a first picture analysis on the second short video to obtain a first picture analysis result.
[0097] The first picture analysis is implemented based on a first picture analysis model, and the first picture analysis model is obtained based on training of a convolutional neural network model. In the training sample, the picture information in the short video sample is used as input data, and the description information corresponding to the short video sample picture is used as output data. After training the convolutional neural network model with the short video sample data, the first picture analysis model can be obtained.
[0098] The first screen analysis result also includes multiple entries.
[0099] S23: Determine a first semantic analysis result according to the first context analysis result and the first picture analysis result.
[0100] In this step, the first semantic analysis result is obtained based on the integration of the first context analysis result and the first picture analysis result, and the first semantic analysis result includes multiple entries.
[0101] S3: Obtain D pieces of first comment area information of the second short video, and determine a second semantic analysis result according to the D pieces of first comment area information.
[0102] The S3 comprises the following sub-steps:
[0103] S31: Obtain D pieces of first comment area information corresponding to the second short video.
[0104] The D pieces of first comment area information refer to all comment information corresponding to the second short video.
[0105] S32: Perform semantic clustering analysis on the D pieces of information in the first comment area to obtain E second cluster centers.
[0106] S33: Determine F second term information based on the E second cluster centers, and determine a second semantic analysis result according to the F second term information.
[0107] D, E and F are natural numbers greater than or equal to 0, and F>D>E.
[0108] S4: Generate a first set of terms to be recommended according to the first semantic analysis result and the second semantic analysis result.
[0109] In this step, after integrating the plurality of term information in the first semantic analysis result and the second semantic analysis result, the first set of terms to be recommended can be obtained.
[0110] The integration refers to the operation of merging and removing duplicate entries.
[0111] S5: monitoring the browsing behavior of the first user on each piece of the first comment area information in real time, and generating a second set of terms to be recommended based on the semantic analysis result of the first comment area information.
[0112] The S5 comprises the following sub-steps:
[0113] S51: Determine the second comment area information browsed by the first user in real time.
[0114] The first comment area information browsed in real time by the first user is used as the second comment area information.
[0115] In this step, the user's gaze point and the user's click position can be obtained in real time through the front camera, and the second comment area information currently being browsed by the first user can be determined by combining the target gaze point and the click position.
[0116] S52: Determine whether the first user has performed a first operation when browsing the information in the second comment area. If not, proceed to S51; if yes, proceed to S53.
[0117] When a user is browsing the comment area, if a certain comment information stimulates the user's search need for a certain piece of knowledge, the user will usually perform a return operation to search for the knowledge in the search bar, so the return operation is used as the first operation.
[0118] S53: Perform semantic expansion analysis on the second comment area information to obtain G basic knowledge information and H expanded search information.
[0119] The S53 comprises the following sub-steps:
[0120] S531: Perform semantic analysis on the second comment area information to obtain G third term information.
[0121] The semantic analysis refers to performing semantic analysis on the second comment area information and the sub-comment information of the second comment area information, so as to extract G third term information.
[0122] S532: Taking each of the third term information as a piece of basic knowledge information, thereby determining G basic knowledge information.
[0123] S533: Perform an extended search for each of the third term information, thereby determining H extended search information.
[0124] The extended search information is obtained through an extended search based on the basic knowledge information. The extended search refers to a multi-stage search. For example, B can be obtained by searching through A, and C can be obtained by searching through B. In this case, A can be used as the basic knowledge information, and B and C can be used as the extended search information.
[0125] S54: Based on the G basic knowledge information and the H expanded search information, a second set of terms to be recommended is obtained.
[0126] The second set of terms to be recommended is obtained by integrating the G basic knowledge information and the H expanded search information.
[0127] S6: Generate a third set of terms to be recommended based on the first set of terms to be recommended and the second set of terms to be recommended, and filter the third set of terms to be recommended using the first user portrait and the first term blacklist to obtain target recommended terms.
[0128] The S6 comprises the following sub-steps:
[0129] S61: Integrate the first set of terms to be recommended and the second set of terms to be recommended to obtain a third set of terms to be recommended.
[0130] The integration refers to merging and removing duplicate entries in the first set of terms to be recommended and the second set of terms to be recommended.
[0131] S62: Calculate the first similarity between each term in the third set of terms to be recommended and the first user portrait, and delete the terms whose first similarity is less than a first preset value from the third set of terms to be recommended to obtain a fourth set of terms to be recommended.
[0132] The term and the first user portrait are respectively expressed as vectors, and the cosine similarity between the vectors is calculated, and the cosine similarity is the first similarity.
[0133] S63: Calculate the second similarity between each term in the fourth set of terms to be recommended and the first set of terms blacklist, and delete the terms whose second similarity is greater than a second preset value from the fourth set of terms to be recommended to obtain a fifth set of terms to be recommended.
[0134] The first term blacklist refers to terms that are obviously inconsistent with the preferences of the first user. When the similarity between the term to be recommended and the first term blacklist is greater than the second preset value, the term can be deleted.
[0135] S64: Determine I target recommended terms according to the fifth set of terms to be recommended.
[0136] The fifth set of terms to be recommended includes multiple terms. According to the requirement on the number of terms recommended to the user, I target recommended terms can be determined from the fifth set of terms to be recommended and displayed to the user.
[0137] S7: According to the second operation of the first user on the I target recommended terms, the first term blacklist is updated.
[0138] The S7 comprises the following sub-steps:
[0139] S71: Obtain a second operation of the first user.
[0140] The second operation includes clicking on a certain target recommended term or clicking on a "change batch" button. The "change batch" button means that if the currently displayed target recommended terms are not of interest to the first user, clicking the button can change to another batch of target recommended terms.
[0141] S72: If the second operation is a click operation on the first target recommended term, then go to S73; if the second operation is a click operation on a "Change Batch" button, then go to S74.
[0142] Among them, the first target recommended term is one of the I target recommended terms.
[0143] S73: Calculate the third similarities between the I target recommended terms and the first target recommended terms respectively, and add the target recommended terms whose third similarities are less than a third preset value to the first term blacklist.
[0144] S74: Add all the target recommended terms to the first term blacklist
[0145] The present application also proposes a user behavior prediction system for executing the above-mentioned user behavior prediction method.
[0146] The present application proposes a method and device for predicting user behavior. In order to accurately predict the knowledge search needs generated by users during the short video browsing process, the short video currently browsed by the user is first analyzed in two dimensions, namely, the meaning and the picture, so as to obtain a first semantic analysis result, and the information in the comment area is analyzed to obtain a second semantic analysis result, so as to obtain a first set of terms to be recommended. Then, based on the user's specific operation on a specific comment, basic knowledge information and expanded search information are determined, so as to obtain a second set of terms to be recommended. Finally, the first term blacklist is updated based on the user's operation information on the target recommended term. The present application determines the terms to be recommended from multiple dimensions, further expands the terms to be recommended according to the user's specific operation information, and filters the recommended terms based on the user's browsing profile and the term blacklist, so as to make the recommended terms more in line with the user's actual needs.
[0147] The above description is only a preferred embodiment of the present invention, so all equivalent changes or modifications made according to the structure, characteristics and principles described in the scope of the patent application of the present invention are included in the scope of the patent application of the present invention.
Claims
1. A user behavior prediction method, characterized in that: The method includes: S1: Determine a first user portrait according to a short video browsing record of the first user within a preset time period; The S1 comprises the following sub-steps: S11: Acquire A first short videos browsed by the first user within a preset time period, wherein each of the first short videos corresponds one-to-one to a first tag; S12: Performing a first frequency analysis on the A first short videos using the first tag to obtain B first short videos and B first viewing times; S13: performing cluster analysis on the B first short videos according to the B first viewing times to obtain C first cluster centers, and obtaining the first user portrait based on the C first cluster centers; S2: Obtain a second short video currently browsed by the first user, and determine a first semantic analysis result according to the second short video; S3: Obtain D pieces of information in the first comment area of the second short video, and determine a second semantic analysis result according to the D pieces of information in the first comment area; comprising the following sub-steps: S31: Obtain D pieces of first comment area information corresponding to the second short video; The D pieces of first comment area information refer to all comment information corresponding to the second short video; S32: performing semantic clustering analysis on the D pieces of information in the first comment area to obtain E second clustering centers; S33: Determine F second word information based on the E second cluster centers, and determine a second semantic analysis result according to the F second word information; Among them, E and F are both natural numbers greater than 0; S4: generating a first set of terms to be recommended according to the first semantic analysis result and the second semantic analysis result; S5: monitoring the browsing behavior of the first user on each piece of the first comment area information in real time, and generating a second set of terms to be recommended according to the semantic analysis result of the first comment area information; comprising the following sub-steps: S51: Determine the second comment area information browsed by the first user in real time; wherein the first comment area information browsed by the first user in real time is used as the second comment area information; S52: Determine whether the first user has performed a first operation when browsing the information in the second comment area, if not, proceed to S51, if yes, proceed to S53; S53: performing semantic expansion analysis on the second comment area information to obtain G basic knowledge information and H expanded search information; S54: obtaining a second set of to-be-recommended terms based on the G basic knowledge information and the H expanded search information; Wherein, G and H are both natural numbers greater than 0; S6: generating a third set of terms to be recommended based on the first set of terms to be recommended and the second set of terms to be recommended, filtering the third set of terms to be recommended using the first user profile and the first term blacklist to obtain target recommended terms; S7: According to the second operation of the first user on the target recommended term I, the first term blacklist is updated; Among them, A, B, C, D and I are all natural numbers greater than 0.
2. A user behavior prediction method according to claim 1, characterized in that: The S13 comprises the following sub-steps: S131: performing semantic analysis on the B first short videos one by one to obtain B first semantic analysis information; S132: weighting the B pieces of first semantic analysis information based on the B first viewing times to obtain B pieces of first weight information corresponding to the B first short videos; S133: performing cluster analysis on the B first short videos according to the B first weight information and the B first semantic analysis information to obtain C first cluster centers; S134: Determine a first user portrait based on the C first cluster centers.
3. A user behavior prediction method according to claim 2, characterized in that: The S2 comprises the following sub-steps: S21: performing a first context analysis on the second short video to obtain a first context analysis result; S22: performing a first picture analysis on the second short video to obtain a first picture analysis result; S23: Determine a first semantic analysis result according to the first context analysis result and the first picture analysis result.
4. A user behavior prediction method according to claim 3, characterized in that: The S21 includes the following sub-steps: S211: extracting first text information and first subtitle information from the second short video; S212: Inputting the first text information and the first subtitle information into a first context analysis model to obtain the first context analysis result.
5. A user behavior prediction method according to claim 1, characterized in that: The S53 comprises the following sub-steps: S531: performing semantic analysis on the second comment area information to obtain G third word information; S532: taking each of the third term information as a piece of basic knowledge information, thereby determining G pieces of basic knowledge information; S533: Perform an extended search for each of the third term information, thereby determining H extended search information.
6. A user behavior prediction method according to claim 5, characterized in that: The S6 comprises the following sub-steps: S61: Integrate the first set of terms to be recommended and the second set of terms to be recommended to obtain a third set of terms to be recommended; S62: Calculate the first similarity between each term in the third set of terms to be recommended and the first user portrait, and delete the terms whose first similarity is less than a first preset value from the third set of terms to be recommended to obtain a fourth set of terms to be recommended; S63: Calculate the second similarity between each entry in the fourth set of recommended terms and the first set of blacklisted terms, and delete the entries whose second similarity is greater than the second preset value from the fourth set of recommended terms to obtain a fifth set of recommended terms; S64: Determine I target recommended term according to the fifth set of terms to be recommended; Wherein, I is a natural number greater than 0.
7. A user behavior prediction method according to claim 6, characterized in that: The S7 comprises the following sub-steps: S71: Acquire a second operation of the first user; S72: if the second operation is a click operation on the first target recommended term, then the process proceeds to S73; if the second operation is a click operation on a "change batch" button, then the process proceeds to S74; S73: calculating the third similarities between the first target recommended terms and the first target recommended terms respectively, and adding the target recommended terms whose third similarities are less than a third preset value to the first term blacklist; S74: Add all the target recommended terms to the first term blacklist.
8. A user behavior prediction system, used to implement a user behavior prediction method according to any one of claims 1 to 7.
Citation Information
Patent Citations
Clustering method and device of search terms in website
CN106708901A
Recommendation method and device, equipment and computer storage medium
CN112395489A
Information display method and device and computer storage medium
CN113157970A
Search term display method and device, storage medium and computer program product
CN113779365A
Short video recommendation method and system based on improved bald eagle search algorithm
CN115878841A
