A search attribution method, device, computer device and storage medium

Abstract

The present disclosure provides a search attribution method and device, computer equipment and a storage medium, wherein the method comprises: obtaining target search data; determining a target demand scenario corresponding to the target search data based on target category information corresponding to the target search data under a preset multi-level category; and determining an attribution video corresponding to the target search data in an attribution manner matching the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of computer, and in particular, to a search attribution method, device, computer equipment and storage medium. BACKGROUND

[0002] In a video watching scenario, after seeing a target video, a user can initiate a new video search behavior, and then find other videos similar to the target video. At this time, it can be considered that the target video can bring the user search cognition that similar videos can be searched.

[0003] Such videos that can bring search cognition are very important for improving the user's search cognition of the search engine,

[0004] and better utilizing the search resources of the search engine. However, because the amount of videos watched by a user every day is large and the types and contents of videos are complex and diverse, how to find such videos has become a technical difficulty. SUMMARY

[0005] The embodiments of the present disclosure at least provide a search attribution method, device, computer equipment and storage medium.

[0006] In a first aspect, the embodiments of the present disclosure provide a search attribution method, comprising:

[0007] obtaining target search data;

[0008] determining a target demand scenario corresponding to the target search data based on target category information corresponding to the target search data under a preset multi-level category;

[0009] determining an attribution video corresponding to the target search data in an attribution manner matching the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition.

[0010] In a possible implementation, in a case where the target demand scenario is a generalization demand scenario, the attribution manner comprises determining the attribution video based on the multi-level category;

[0011] The determining of the attribution video corresponding to the target search data in the attribution manner matching the target demand scenario comprises:

[0012] determining a target leaf category in the target category information corresponding to the target search data;

[0013] matching a historical browsing video matching the target leaf category in historical browsing videos in a preset time period before the target search data is generated as the attribution video.

[0014] In a possible implementation, when the target demand scenario is a non-generalized demand scenario, the attribution manner comprises determining the attribution video based on text similarity;

[0015] The determining of the attribution video corresponding to the target search data according to the attribution manner matching the target demand scenario comprises:

[0016] The historical browsing video having text similarity greater than a set threshold with the target search data in a preset time period before the target search data is generated is taken as the attribution video.

[0017] In a possible implementation, the text similarity between the historical browsing video and the target search data is determined according to the following steps:

[0018] The video text data corresponding to the historical browsing video is obtained; the video text data at least comprises a search entity feature corresponding to the historical browsing video;

[0019] The text similarity between the historical browsing video and the target search data is obtained by calculating the similarity between the video text data and the target search data.

[0020] In a possible implementation, the method further comprises pre-clustering each target leaf category according to the following steps:

[0021] The video text data of each first browsing video corresponding to the generalized demand scenario in historical search results is pre-obtained;

[0022] Based on the video text data of each first browsing video, a first clustering model pre-trained is used to cluster each first browsing video, to obtain each interest point category, and each interest point category obtained by clustering is taken as a target leaf category.

[0023] In a possible implementation, the method further comprises pre-clustering each search entity feature according to the following steps:

[0024] The video text data of each second browsing video corresponding to the non-generalized demand scenario in historical search results is pre-obtained;

[0025] Based on the video text data of each second browsing video, a second clustering model pre-trained is used to cluster each second browsing video, to obtain a word sense clustering result, and one category in the word sense clustering result corresponds to one search entity feature.

[0026] In a possible implementation, after determining the attribution video corresponding to the target search data, the method further includes:

[0027] In the information push page, the associated videos related to the attribution video are distributed.

[0028] In a second aspect, the embodiments of the present disclosure further provide a search attribution device, comprising:

[0029] an acquisition module configured to acquire target search data;

[0030] a first determination module configured to determine a target demand scenario corresponding to the target search data based on target category information corresponding to the target search data under a preset multi-level category;

[0031] a second determination module configured to determine an attribution video corresponding to the target search data according to an attribution manner matched with the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition.

[0032] In a third aspect, the optional implementation of the present disclosure further provides a computer device, a processor and a memory, the memory stores machine readable instructions executable by the processor, the processor is configured to execute the machine readable instructions stored in the memory, and the machine readable instructions are executed by the processor to execute the steps of the first aspect or any possible implementation of the first aspect.

[0033] In a fourth aspect, the optional implementation of the present disclosure further provides a computer readable storage medium, and the computer readable storage medium stores a computer program, and the computer program is executed to execute the steps of the first aspect or any possible implementation of the first aspect.

[0034] For the effects of the search attribution device, the computer device and the computer readable storage medium, refer to the description of the search attribution method, and details are not described herein.

[0035] The search attribution method and device, the computer device, and the storage medium provided by the embodiments of the present disclosure can be used to determine the attribution manner required for searching the target search data, according to the target category information corresponding to the target search data, and then perform attribution according to the attribution manner matched with the target demand scenario, so as to accurately determine the attribution video corresponding to the target search data, that is, accurately determine the video that can bring search cognition, thereby improving the search cognition of the user to the search engine, and finally achieving the purpose of fully utilizing the search resources of the search engine.

[0036] In order to make the above objectives, characteristics and advantages of the present disclosure more apparent, clear and easy to understand, the following will specifically describe the preferred embodiments in conjunction with the accompanying drawings, and the detailed description is as follows. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following will briefly introduce the drawings needed to be used in the embodiments, the drawings herein are incorporated into the description and form a part of the description, the drawings show the embodiments consistent with the present disclosure, and are used to illustrate the technical solutions of the present disclosure together with the description. It should be understood that the following drawings only show some embodiments of the present disclosure, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of the drawings.

[0038] Figure 1 A schematic diagram of a cognitive link provided by the embodiments of the present disclosure is shown;

[0039] Figure 2 A flowchart of a search attribution method provided by the embodiments of the present disclosure is shown;

[0040] Figure 3 A schematic diagram of a search attribution provided by the embodiments of the present disclosure is shown;

[0041] Figure 4 A schematic diagram of determining a target leaf category and a search entity word feature provided by the embodiments of the present disclosure is shown;

[0042] Figure 5 An attribution idea diagram provided by the embodiments of the present disclosure is shown;

[0043] Figure 6 A schematic diagram of a search attribution device provided by the embodiments of the present disclosure is shown;

[0044] Figure 7A structural diagram of a computer device is shown. DETAILED DESCRIPTION

[0045] To make the objectives, technical solutions and advantages of the embodiments of the present disclosure clearer, the following will be combined with the accompanying drawings for the embodiments of the present disclosure to make a clear and complete description of the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure and not all the embodiments. The components of the embodiments of the present disclosure described and shown herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure is not intended to limit the scope of the claimed present disclosure, but only represents selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present disclosure.

[0046] In addition, the terms "first", "second", and the like in the specification and claims of the embodiments of the present disclosure and the above-described accompanying drawings are used to distinguish similar objects and do not necessarily have to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein.

[0047] "Multiple or several" referred to herein means two or more. "And / or" describes the association between the associated objects, which means that there can be three relationships, for example, A and / or B, which can mean that A exists alone, A and B exist together, and B exists alone. The character " / " generally represents that the associated objects before and after are in an "or" relationship.

[0048] It is found through research that the formation of user search cognition often depends on the cognitive link as shown in Figure 1 That is, the user first sees the video content, then initiates a search and gets the search results that meet the search needs, so as to form search cognition related to the video content seen. Therefore, if the user's search cognition can be accurately attributed to the past video, it will be able to achieve the purpose of helping users improve search cognition. However, there are two major technical difficulties in the video attribution process:

[0049] 1. Attribution difficulty: Since the magnitude of videos seen by users every day is very large, it is difficult to accurately attribute the search behavior of a period of time (such as a day, a week, a month) to a certain video seen. Not only is the workload huge and the attribution difficult, but the accuracy of the attribution result cannot be grasped.

[0050] 2. Difficulty in grasping the granularity of attribution: After watching a video, users form search cognition, which can be generally divided into two scenarios: one is to find accurate content related to the theme of the video watched, achieving the purpose of fine-grained search; the other is to find generalized content similar to the theme of the video watched, achieving the purpose of coarse-grained search.

[0051] The existence of the above two technical difficulties leads to low difficulty and accuracy of video attribution, therefore, how to realize accurate video attribution has become a problem to be solved.

[0052] Based on the above research, the present disclosure provides a search attribution scheme. Since the attribution methods and attribution granularity are different under different target demand scenarios, the target demand scenario corresponding to the target search data is determined according to the target category information corresponding to the target search data, so as to quickly and accurately determine the attribution method required when the target search data is searched and attributed; then attribution is performed according to the attribution method matched with the target demand scenario, which can accurately determine the attribution video corresponding to the target search data, that is, the video that can bring search cognition can be accurately determined, so as to improve the search cognition of users to the search engine by using these videos, and finally achieve the purpose of fully utilizing the search resources of the search engine.

[0053] The defects of the above scheme are the result of the inventors' careful research after practice, therefore, the discovery process of the above problems and the solutions proposed by the present disclosure to solve the above problems should be the contribution of the inventors to the present disclosure in the process of the present disclosure.

[0054] It should be noted that similar reference numbers and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0055] It can be understood that before using the technical solutions disclosed in the embodiments of the present disclosure, the type, scope of use, use scenario, etc. of the personal information involved in the present disclosure should be informed to the user and the authorization of the user should be obtained in accordance with relevant laws and regulations.

[0056] To facilitate understanding of this embodiment, a search attribution method disclosed in this disclosure will first be described in detail. The execution subject of the search attribution method provided in this disclosure is generally a terminal device or other processing device with certain computing power. The terminal device can be a user equipment (UE), mobile device, user terminal, terminal, personal digital assistant (PDA), handheld device, computer device, etc. In some possible implementations, the search attribution method can be implemented by the processor calling computer-readable instructions stored in the memory.

[0057] The search attribution method provided in this disclosure is illustrated below using a computer device as an example of the executing entity.

[0058] like Figure 2 The flowchart shown is a search attribution method provided in an embodiment of this disclosure, which may include the following steps:

[0059] S201: Obtain target search data.

[0060] Here, target search data can be search behavior data initiated by users based on their formed target search cognition, which can be a re-search behavior. For example, target search data can be search terms or search phrases entered by the user. Target search cognition can be the cognition a user forms after watching a video related to that video; the video the user watched can be called the attribution video that brings about the target search cognition. Target search cognition can be understood as: the user's perception that videos satisfying their target needs can be found through searching.

[0061] For example, target search cognition can be the user's belief that they can find content related to video content within a video application (APP). For instance, target search cognition could be the user's belief that they can find film / TV show A within a video APP; or, for another example, the user's belief that they can find food videos within a video APP.

[0062] In practice, the acquired target search data may include at least one search data point from one or more users. If there are multiple target search data points, each search data point can be processed to determine the attribution video corresponding to each search data point.

[0063] S202: Based on the target category information corresponding to the target search data under the preset multi-level categories, determine the target demand scenario corresponding to the target search data.

[0064] Here, a plurality of levels of categories can be pre-set, different levels of categories include different category information, and the granularity of the category information under different levels of categories is different. For some category information, the category information under a high level category can be the parent category information of the category information under an adjacent low level category. For example, six levels of categories can be pre-set, the category information under the first level category can include fashion category information, the category information under the second level category can include fashion product information and fashion tutorial information, the category information under the third level category can include clothing and accessory product information, beauty and skin care product information, other fashion product information under fashion product information, and clothing and accessory tutorial information, beauty and skin care tutorial information, and other tutorial information under fashion tutorial information; the category information under the fourth level category can include clothing and accessory brand information, clothing and hat product information, fashion bag product information, jewelry product information, fashion glasses product information, fashion watch product information, and other clothing and accessory product information under clothing and accessory product information, beauty and skin care brand information, skin care product information, beauty product information, nail product information, hair styling information, fragrance product information, and hand / body care product information under beauty and skin care product information, clothing and hat tutorial information, fashion bag knowledge information, jewelry knowledge information, fashion glasses knowledge information, fashion watch knowledge information, and other clothing and accessory tutorial information under clothing and accessory tutorial information, and skin care tutorial information, beauty tutorial information, nail tutorial information, hair styling tutorial information, and other beauty and skin care tutorial information under beauty and skin care tutorial information; the fifth level category can include clothing and hat tutorial information under clothing and hat tutorial information, and other clothing and hat tutorial information; and the sixth level category can include casual type hat and clothing tutorial information, business type hat and clothing tutorial information, and sports type hat and clothing tutorial information, and mix tutorial information under clothing and hat tutorial information.

[0065] The target category information is the category information to which the target search data corresponds. In the implementation, when S202 is performed, the target search data can be determined to correspond to the target category information under each level of category. In this case, there can be one or more categories that do not have target category information corresponding to the target search data. Alternatively, the target search data can be determined to correspond to the target category information under any category, such as the target category information under the lowest level. Of course, the target category information under the target category corresponding to the target search data can be determined first, and the target category information under the target category corresponding to the target search data can be determined to be the final target category information corresponding to the target search data. In terms of the height of the level, the first level is higher than the second level, which is higher than the third level, which is higher than the fourth level, which is higher than the fifth level, which is higher than the sixth level.

[0066] The demand scenario can be specifically divided into a generalized demand scenario and a non-generalized demand scenario. The generalized demand scenario refers to a scenario in which target search data needs to be generalized to determine target category information. The generalized demand scenario can be, for example, a food scenario, a recipe search scenario, a travel strategy scenario, and the like. For example, a user has a cooking demand, which can search for the method of cooking food A on the first day and the method of cooking food B on the second day, and food A and food B can be generalized to recipe category information or food category information. In the generalized demand scenario, it is sufficient to help the user search for an attribution video corresponding to the target search data and belonging to the same category.

[0067] The non-generalized demand scenario refers to a scenario in which the user has a precise search. In this scenario, the target search data can be directly matched with target category information under a multi-level category, such as the target search data being directly equal to the target category information or the target search data including the target category information. The non-generalized demand scenario can be, for example, a movie and television scenario, a game scenario, a novel scenario, and the like. For example, in the case where the target search data is information about a movie and television work A, such as the movie and television work A, the ending of the movie and television work A, and a task A in the movie and television work A, the target search data can be directly matched with the target category information (i.e., the movie and television work A), and the content matched with the movie and television work A can be accurately found.

[0068] Different target category information can correspond to different demand scenarios. For example, food category information and recipe category information belong to a generalized demand scenario, and novel A category information and movie and television work B category information belong to a non-generalized demand scenario. The target demand scenario is a demand scenario corresponding to the target search data.

[0069] In specific implementation, an association relationship between category information under a multi-level category and a demand scenario can be established in advance. After the target search data is obtained, the target category information corresponding to the target search data under the preset multi-level category can be determined, and then the target demand scenario corresponding to the target search data can be determined according to the association relationship established in advance.

[0070] S203: determining an attribution video corresponding to the target search data in a manner matched with the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition.

[0071] Here, different demand scenarios correspond to different attribution manners. Specifically, for the generalization demand scenario, since it is a scenario with a larger range, the search granularity is larger, and therefore the video can be attributed according to a large search granularity. For the non-generalization demand scenario, since it is a scenario with a small range and a clear purpose, the search granularity is smaller, and therefore the video can be attributed according to a small search granularity.

[0072] The attributed video is a video that helps the user to have target search cognition. For example, if the attributed video is a food video, the user can have target search cognition about food; if the attributed video is a car video, the user can have target search cognition about cars; if the attributed video is a movie A, the user can have target search cognition about movie A; and if the attributed video is a novel A, the user can have target search cognition about novel A.

[0073] For example, in the case of giving the user target search cognition, it is equivalent to giving the user cognition of searching for a video in the video APP that matches the target search cognition, that is, the user knows that the video APP can provide a video that matches the target search cognition, and in the case of needing to find a video that matches the target search cognition, the user can initiate a search behavior in the video APP based on the target search cognition, so as to obtain search behavior data.

[0074] In specific implementation, after determining the target demand scenario corresponding to the target search data, a target attribution manner that matches the target demand scenario can be determined, the target search data is attributed according to the attribution manner, and one or more attributed videos corresponding to the target search data are determined. In the case where the attributed video includes multiple videos, the multiple attributed videos correspond to the same target search cognition.

[0075] In an embodiment, in the case where the target demand scenario is a generalization demand scenario, the attribution manner includes determining the attributed video based on a multi-level category, that is, the multi-level category can be used for attribution to obtain the attributed video. Specifically, for S203, the following steps can be implemented:

[0076] S203-1: Determine a target leaf category in the target category information corresponding to the target search data.

[0077] Here, the target category information determined based on S202 can be target category information under a target level category, where the target level category can be a first level category, or a first level category with specific target category information. For example, in the case where the first level category has target category information, the target level category is the first level category; in the case where the second level category has target category information, the target level category is the second level category, and in the case where the sixth level category has target category information, the target level category is the sixth level category. The sixth level category can also be referred to as a leaf category, and the category information under the leaf category can be referred to as a leaf category. The leaf category can be a pre-constructed point of interest category.

[0078] In specific implementation, in the case where the target category information is higher than the leaf category, each target leaf category included under the leaf category corresponding to the target category information can be determined. For example, in the case where the target category information is food category information, the target leaf category under the leaf category can be recipe category.

[0079] S203-2: Among the historical browsing videos in the preset time period before the target search data is generated, the historical browsing videos matching the target leaf category are determined as attribution videos.

[0080] Here, the historical browsing videos are each video browsed by the user in the preset time period, and different historical browsing videos correspond to different leaf categories, i.e., different historical browsing videos correspond to different points of interest, and one historical browsing video can correspond to at least one leaf category.

[0081] In specific implementation, after the target leaf category is determined, each historical browsing video browsed in the preset time period before the target search data is generated can be obtained, and the leaf category corresponding to each historical browsing video is determined. Then, according to the leaf category corresponding to each historical browsing video and the target leaf category, the historical browsing videos corresponding to the leaf category including the target leaf category are filtered from the historical browsing videos, and the filtered historical browsing videos are determined as attribution videos.

[0082] In another embodiment, in the case where the target demand scenario is a non-generalized demand scenario, the attribution manner includes determining attribution videos based on text similarity, i.e., attribution can be performed using text similarity. Here, the text similarity can be the similarity between the text information corresponding to the target search data and the historical browsing videos. Specifically, for S203, the following steps can be implemented:

[0083] Among the historical browsing videos in the preset time period before the target search data is generated, the historical browsing videos having a text similarity with the target search data greater than a set threshold are determined as attribution videos.

[0084] Here, the setting threshold value can be set according to the required video attribution accuracy, the higher the required video attribution accuracy, the higher the setting threshold value, and the embodiment of the present disclosure does not specifically limit the setting threshold value.

[0085] For example, for each historical browsing video in a preset time period before the target search data is generated, the text information corresponding to each historical browsing video can be determined first. Here, the text information can be video title information in the historical browsing video, text information included in the video content, etc.

[0086] In specific implementation, for each historical browsing video, the text similarity between the historical browsing video and the target search data can be calculated. Then the historical browsing video with a text similarity greater than a setting threshold value can be taken as an attribution video.

[0087] In an embodiment, the text similarity between the historical browsing video and the target search data can be determined according to the following steps:

[0088] Step 1, obtain video text data corresponding to the historical browsing video; the video text data at least includes search entity word features corresponding to the historical browsing video.

[0089] Here, the search entity word features are used to represent the feature information of the historical browsing video. Specifically, the search entity word can be a feature obtained by hashing the search entity word corresponding to the historical browsing video, wherein one search entity word can be a label corresponding to the historical browsing video, denoted by hashtag.

[0090] Optionally, in addition to the search entity word features corresponding to the historical browsing video, the video text data can also include first search data and second search data. The first search data is each search data that can recall the historical browsing video, and / or user search data initiated by the authorized collection user after viewing the historical browsing video, and / or comment data for the historical browsing video, and the first search data can be represented by click query. The second search data is user search data initiated by the authorized collection user after viewing the historical browsing video, and the second search data can be represented by query.

[0091] For example, in the case of a non-generalization demand scenario, each historical browsing video in a preset time period before the target search data is generated can be obtained, and the corresponding video text data is obtained respectively.

[0092] Step 2: Calculate the text similarity between historically viewed videos and the target search data by calculating the similarity between the video text data and the target search data.

[0093] In practice, any text similarity algorithm can be used to determine the similarity between the video text data corresponding to each historically viewed video and the target search data. Here, when the video text data includes search entity word features, first search data, and second search data, the first similarity between the target search data and the search entity word features, the second similarity between the target search data and the first search data, and the third similarity between the target search data and the second search data can be determined separately. Then, the maximum similarity among the first, second, and third similarities can be used as the text similarity between the historically viewed video and the target search data. Alternatively, the average of the first, second, and third similarities can be used as the text similarity between the historically viewed video and the target search data. Or, different weights can be pre-assigned to the search entity word features, the first search data, and the second search data. After determining the first, second, and third similarities, a weighted sum can be calculated according to the corresponding weights, and the sum can be used as the text similarity between the historically viewed video and the target search data.

[0094] like Figure 3 The diagram illustrates a search attribution method provided in this embodiment. After acquiring target search data, the target category information corresponding to the target search data under preset multi-level categories can be determined first. Then, the target demand scenario corresponding to the target search data can be determined based on the target category information. In the case of a generalized demand scenario, the target leaf category in the target category information can be used as the point of interest category. Then, the historical browsing videos matching the target leaf category can be used as the attribution videos corresponding to the target search data. In the case of a non-generalized demand scenario, the similarity between the target search data and the search entity word features in the video text data corresponding to each historical browsing video can be determined. Then, the search entity word feature with the highest similarity is selected, and the historical browsing video corresponding to the search entity word feature with the highest similarity is used as the attribution video corresponding to the target search data. That is, the historical browsing videos whose corresponding video text data includes the search entity word feature with the highest similarity are used as the attribution videos corresponding to the target search data.

[0095] Based on the above embodiments, since the attribution manner and attribution granularity are different in different target demand scenarios, the target demand scenario corresponding to the target search data is determined according to the target category information corresponding to the target search data, so that the attribution manner required when the target search data is attributed is quickly and accurately determined; then, attribution is performed according to the attribution manner matched with the target demand scenario, so that the attribution video corresponding to the target search data can be accurately determined, that is, the video that can bring search cognition can be accurately determined, so that these videos can be used to improve the search cognition of the user to the search engine, and finally the purpose of fully utilizing the search resources of the search engine is achieved.

[0096] In an embodiment, the determination of each target leaf category (i.e., point of interest category) can be determined according to the browsing videos in the generalized demand scenario. Specifically, a plurality of historical search results can be obtained. The plurality of historical search results can be search results respectively determined by a plurality of users in a plurality of search behaviors. For example, the plurality of historical search results can be a plurality of browsing videos.

[0097] Then, for the plurality of historical search results, each first browsing video corresponding to the generalized demand scenario can be selected from the plurality of historical search results, that is, each browsing video belonging to the generalized demand scenario can be selected from the plurality of historical search results. Then, according to the video text data of each first browsing video, a first clustering model pre-trained can be used to cluster each first browsing video, and each point of interest category obtained by clustering can be used as a target leaf category.

[0098] Here, the video text data of the first browsing video can include search entity word features (i.e., hashtags) and / or first search data (i.e., click queries) and / or all-user-after-viewing queries (i.e., all-user-after-viewing queries). The first clustering model is a model pre-trained for clustering from the perspective of leaf category.

[0099] For example, the video text data of each first browsing video can be input into the first clustering model, and the first clustering model can be used to cluster the video text data of each first browsing video from the perspective of leaf category, so as to obtain each point of interest category. Then, each point of interest category obtained by clustering can be used as a target leaf category.

[0100] In another embodiment, for the search entity word feature, the browsing videos under the non-generalized demand scenario can be determined. Specifically, a plurality of historical search results can be obtained in advance, and then in the plurality of historical search results obtained in advance, each second browsing video corresponding to the non-generalized demand scenario is filtered, that is, each browsing video belonging to the non-generalized demand scenario is filtered from the plurality of historical search results, and then the video text data of each second browsing video can be obtained.

[0101] Here, the video text data of the second browsing video can include the search entity word feature (i.e. hashtag) and / or the second search data (i.e. click query) and / or the full user search query after watching (i.e. full user search query after watching). The second clustering model is a pre-trained model for clustering from the perspective of search entity word, that is, the second clustering model can cluster the entity words belonging to the same category in the video text data together, and determine the category corresponding to the clustered entity words.

[0102] Then, based on the video text data of each second browsing video, the pre-trained second clustering model can be used to cluster each second browsing video to obtain a word sense clustering result, and each category in the word sense clustering result corresponds to a search entity word feature.

[0103] Here, the word sense clustering result can include one or more categories, one category corresponds to one or more video text data clustered into a category, and one category corresponds to one search entity word feature, that is, one search entity word feature can be determined according to one category. Each video can correspond to one or more initial search entity words, and after obtaining the word sense clustering result, the final search entity word feature can be determined by using the initial search entity word matched with each category in the word sense clustering result.

[0104] For example, the video text data of each second browsing video can be input into the second clustering model, and the second clustering model can be used to cluster the video text data of each second browsing video according to the search entity word, so as to obtain each category. Then, each search entity word feature can be determined according to each category obtained by clustering.

[0105] As Figure 4The diagram illustrates a method for determining target leaf categories and search entity word features according to an embodiment of this disclosure. It involves acquiring at least one search data point from at least one user, then determining the historical search results corresponding to each search data point, and determining the video text data (including initial search entity words and / or second search data (i.e., click query) and / or full-user post-view search query) corresponding to each historical search result. Then, for each first viewed video in the historical search results corresponding to a generalized demand scenario, a pre-trained first clustering model is used to cluster the first viewed videos based on their video text data to obtain each target leaf category. For each second viewed video in the historical search results corresponding to a non-generalized demand scenario, a pre-trained second clustering model is used to cluster the second viewed videos based on their video text data to obtain each search entity word feature.

[0106] In one embodiment, Figure 4 The process shown for classifying target leaf categories and search entity word features is based on, for example... Figure 5 The attribution mind map shown identifies the underlying causes. For example... Figure 5 As shown, after a user sees the content of a target video, regardless of whether they immediately engage in a post-viewing search, they may develop a subsequent need to obtain search results related to the target video's content within a preset time period after watching the video, thus generating proactive search behavior. To determine whether this proactive search behavior is based on the target video's content, we can analyze the context of the search data corresponding to the proactive search behavior and the target video's text data to determine if the proactive search behavior corresponds to the same point of interest as the target video's text data. If so, it indicates that the proactive search behavior is an action generated after the user has developed a target search awareness, which must be based on the target video's content; therefore, it can be considered that the target video can generate target search awareness. If not, it indicates that it is impossible to determine whether the target video can generate target search awareness based solely on the proactive search behavior. Based on the attribution strategy diagram above, the corresponding attribution strategies can be generated... Figure 4 The division process is shown below.

[0107] In one embodiment, after determining the attribution video corresponding to the target search data, related videos associated with the attribution video can also be distributed on the information push page.

[0108] Here, the information push page can be the video playback page corresponding to a video app, and the associated videos are those belonging to the same or similar categories as the target video. For example, if the attributed video is a video of film or television work A, the associated videos could be introductory videos of film or television work A, videos of characters in film or television work A, narration videos of film or television work A, videos of a specific part of film or television work A, etc. As another example, if the attributed video is a video of dish A, the associated videos could be videos of dish B, dish C, or videos related to dish A from different authors, etc.

[0109] For example, attribution videos and / or related videos can be recommended to users who generated target search data on the information push page to help them form a target search awareness regarding the search engine. Subsequently, users, having formed a target search awareness, can be more likely to initiate searches using the search engine.

[0110] For example, based on each user's target search data, the attribution video for each user can be determined. Then, based on the target search cognition corresponding to each attribution video, the target users under each target search cognition can be identified. Since a target search cognition corresponds to a search vertical—for example, the cognition of searching for food corresponds to the food vertical, and searching for video A corresponds to the video A vertical—videos associated with that vertical can be distributed to the target users corresponding to that vertical. This helps target users form a deeper target search cognition. For example, in a video app, food videos can be distributed to target users corresponding to the food vertical, helping them form the search cognition that they can find various food videos in the video app. Subsequently, a larger number of target users can actively search for food videos using the video app's search engine after opening the app, thereby increasing search traffic.

[0111] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0112] Based on the same inventive concept, this disclosure also provides a search attribution device corresponding to the search attribution method. Since the principle of the device in this disclosure for solving the problem is similar to the search attribution method described above in this disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0113] like Figure 6 The diagram shown is a schematic representation of a search attribution device provided in an embodiment of this disclosure, comprising:

[0114] The acquisition module 601 is configured to acquire target search data.

[0115] The first determination module 602 is configured to determine a target demand scenario corresponding to the target search data based on target category information corresponding to the target search data in a preset multi-level category.

[0116] The second determination module 603 is configured to determine an attribution video corresponding to the target search data in an attribution manner matched with the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition.

[0117] In a possible implementation, in a case where the target demand scenario is a generalized demand scenario, the attribution manner includes determining the attribution video based on the multi-level category.

[0118] The second determination module 603, when determining the attribution video corresponding to the target search data in the attribution manner matched with the target demand scenario, is configured to:

[0119] determine a target leaf category in the target category information corresponding to the target search data.

[0120] determine a historical browsing video matching the target leaf category in historical browsing videos in a preset time period before the target search data is generated as the attribution video.

[0121] In a possible implementation, in a case where the target demand scenario is a non-generalized demand scenario, the attribution manner includes determining the attribution video based on text similarity.

[0122] The second determination module 603, when determining the attribution video corresponding to the target search data in the attribution manner matched with the target demand scenario, is configured to:

[0123] determine a historical browsing video with text similarity greater than a set threshold between each historical browsing video in a preset time period before the target search data is generated and the target search data as the attribution video.

[0124] In a possible implementation, the second determination module 603 is configured to determine the text similarity between the historical browsing video and the target search data according to the following steps:

[0125] acquire video text data corresponding to the historical browsing video; the video text data at least includes a search entity word feature corresponding to the historical browsing video;

[0126] The text similarity between the historical browsing video and the target search data is obtained by calculating the similarity between the video text data and the target search data.

[0127] In a possible implementation, the apparatus further includes:

[0128] The first clustering module 604 is configured to pre-cluster each target leaf category according to the following steps:

[0129] The video text data of each first browsing video corresponding to the generalized demand scenario in the historical search result is pre-acquired;

[0130] Based on the video text data of each first browsing video, a pre-trained first clustering model is used to cluster each first browsing video, and each interest point category obtained by clustering is taken as a target leaf category.

[0131] In a possible implementation, the apparatus further includes:

[0132] The second clustering module 605 is configured to pre-cluster each search entity word feature according to the following steps:

[0133] The video text data of each second browsing video corresponding to the non-generalized demand scenario in the historical search result is pre-acquired;

[0134] Based on the video text data of each second browsing video, a pre-trained second clustering model is used to cluster each second browsing video, and a word sense clustering result is obtained, wherein one category in the word sense clustering result corresponds to one search entity word feature.

[0135] In a possible implementation, the apparatus further includes:

[0136] The distribution module 606 is configured to, after determining the attribution video corresponding to the target search data, distribute an associated video related to the attribution video on an information push page.

[0137] The description of the processing procedure of each module in the apparatus and the interaction procedure between the modules can refer to the related description in the above method embodiments, and will not be repeated here.

[0138] Based on the same technical concept, the embodiments of the present application further provide a computer device. Referring to FIG. 6, Figure 7 The structure schematic diagram of the computer device provided by the embodiments of the present application includes:

[0139] The processor 71, the memory 72 and the bus 73. Among them, the memory 72 stores machine readable instructions executable by the processor 71, and the processor 71 is used to execute the machine readable instructions stored in the memory 72, and the processor 71 executes the following steps when the machine readable instructions are executed by the processor 71: S201: obtaining target search data; S202: determining the target demand scene corresponding to the target search data based on the target category information corresponding to the target search data under the preset multi-level category; and S203: determining the attribution video corresponding to the target search data in accordance with the attribution mode matched with the target demand scene; the attribution video is a video that brings target search cognition, and the target search data is search behavior data initiated based on the target search cognition.

[0140] The memory 72 described above includes the memory 721 and the external memory 722; here, the memory 721 is also called the internal memory, and is used to temporarily store the operation data in the processor 71 and the data exchanged with the external memory 722 such as a hard disk, and the processor 71 exchanges data with the external memory 722 through the memory 721, and when the computer device is running, the processor 71 and the memory 72 communicate through the bus 73, so that the processor 71 executes the instructions mentioned in the above method embodiments.

[0141] The embodiment of the present disclosure further provides a computer readable storage medium, and the computer readable storage medium stores a computer program. When the computer program is run by a processor, the steps of the search attribution method described in the above method embodiments are executed. Among them, the storage medium can be a volatile or non-volatile computer readable storage medium.

[0142] The computer program product of the search attribution method provided by the embodiment of the present disclosure includes a computer readable storage medium storing program codes, and the instructions included in the program codes can be used to execute the steps of the search attribution method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be described here.

[0143] The computer program product can be specifically realized by hardware, software or a combination thereof. In one optional embodiment, the computer program product is specifically embodied as a computer storage medium, and in another optional embodiment, the computer program product is specifically embodied as a software product, such as a software development kit (Software Development Kit, SDK) and the like.

[0144] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the apparatus described above can refer to the corresponding process in the foregoing method embodiment, and will not be repeated here. In several embodiments provided in the present disclosure, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are only schematic. For example, the division of the units is only a logical function division, and another division can be made in actual implementation. For example, a plurality of units or components can be combined, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some communication interfaces, and can be electrical, mechanical or other forms.

[0145] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. can be located in one place or distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0146] In addition, the functional units in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0147] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a non-volatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present disclosure essentially or the part of the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present disclosure. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0148] If the technical solution of the present application involves personal information, the product applying the technical solution of the present application has been explicitly informed of the personal information processing rules before processing the personal information, and has obtained the personal independent consent. If the technical solution of the present application involves sensitive personal information, the product applying the technical solution of the present application has obtained the personal independent consent before processing the sensitive personal information, and at the same time meets the requirement of "explicit consent". For example, at the personal information collection device such as camera, a clear and prominent mark is set to inform that it has entered the personal information collection range and will collect personal information. If the individual voluntarily enters the collection range, it is considered to agree to collect personal information. Or on the device for processing personal information, through the pop-up information or by uploading personal information by the individual, the individual's authorization is obtained under the condition of using obvious mark / information to inform the individual of the personal information processing rules. The personal information processing rules can include personal information processor, personal information processing purpose, processing method, personal information type, etc.

[0149] Finally, it should be noted that: the above-described embodiments are only specific embodiments of the present disclosure, used to illustrate the technical solutions of the present disclosure, and are not limitations thereof. The protection scope of the present disclosure is not limited thereto. Although the present disclosure has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that any skilled person familiar with the technical field can modify or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features thereof within the technical range disclosed by the present disclosure. The modification, change or replacement does not make the corresponding technical solution deviate from the spirit and scope of the technical solution of the embodiments of the present disclosure, and should be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A search attribution method, characterized by, The method comprises the following steps: acquiring target search data; determining a target demand scenario corresponding to the target search data based on target category information corresponding to the target search data under a preset multi-level category; determining an attribution video corresponding to the target search data in an attribution manner matching the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition; in a case where the target demand scenario is a non-generalized demand scenario, the attribution manner comprises determining the attribution video based on text similarity; the step of determining the attribution video corresponding to the target search data in the attribution manner matching the target demand scenario comprises: determining, as the attribution video, a historical browsing video having a text similarity greater than a set threshold between the historical browsing video and the target search data within a preset time period before the target search data is generated.

2. The method of claim 1, wherein, in a case where the target demand scenario is a generalized demand scenario, the attribution manner comprises determining the attribution video based on the multi-level category; the step of determining the attribution video corresponding to the target search data in the attribution manner matching the target demand scenario comprises: determining a target leaf category in the target category information corresponding to the target search data; determining, as the attribution video, a historical browsing video matching the target leaf category within a preset time period before the target search data is generated.

3. The method of claim 1, wherein, The text similarity between the historical browsing video and the target search data is determined according to the following steps: acquiring video text data corresponding to the historical browsing video; the video text data at least comprises a search entity word feature corresponding to the historical browsing video; obtaining the text similarity between the historical browsing video and the target search data by calculating the similarity between the video text data and the target search data.

4. The method of claim 2, wherein, The method further comprises pre-clustering to obtain each target leaf category according to the following steps: pre-acquiring video text data of each first browsing video corresponding to the generalized demand scenario in historical search results; based on the video text data of each first browsing video, clustering each first browsing video by using a pre-trained first clustering model to obtain each interest point category, and taking each interest point category obtained by clustering as a target leaf category.

5. The method of claim 3, wherein, The method further comprises pre-clustering to obtain each search entity word feature according to the following steps: pre-acquiring video text data of each second browsing video corresponding to the non-generalized demand scenario in historical search results; based on the video text data of each second browsing video, clustering each second browsing video by using a pre-trained second clustering model to obtain a word meaning clustering result, and one category in the word meaning clustering result corresponds to one search entity word feature.

6. The method of claim 1, wherein, After determining the attribution video corresponding to the target search data, the method further comprises: distributing, on an information pushing page, an associated video related to the attribution video.

7. A search attribution apparatus characterized by comprising: The method comprises the following steps: An acquisition module is configured to acquire target search data; A first determination module is configured to determine a target demand scenario corresponding to the target search data based on target category information corresponding to the target search data under a preset multi-level category; A second determination module is configured to determine an attribution video corresponding to the target search data according to an attribution manner matching the target demand scenario; the attribution video is a video bringing target search cognition, and the target search data is search behavior data initiated based on the target search cognition; In a case where the target demand scenario is a non-generalized demand scenario, the attribution manner includes determining the attribution video based on text similarity; The second determination module, when determining the attribution video corresponding to the target search data according to the attribution manner matching the target demand scenario, is configured to: determine, as the attribution video, a historical browsing video having a text similarity greater than a set threshold to the target search data among historical browsing videos in a preset time period before the target search data is generated.

8. A computer device, comprising: comprise: a processor and a memory, the memory storing machine readable instructions executable by the processor, the processor being configured to execute the machine readable instructions stored in the memory, and the machine readable instructions, when executed by the processor, causing the processor to perform the steps of the search attribution method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program, when executed by a computer device, causes the computer device to perform the steps of the search attribution method according to any one of claims 1 to 6.

Images