Video recommendation method and device, equipment and storage medium

Abstract

The invention provides a video recommendation method and device, equipment and a storage medium. Relates to the technical field of data processing. The method comprises the steps of obtaining real-time interaction information corresponding to a video recommendation request in response to the video recommendation request, extracting at least one piece of feature information in the real-time interaction information through a user portrait module of a large language model, obtaining a user portrait based on the at least one piece of feature information, and recommending the video recommendation request to the user portrait. And performing primary screening processing on the plurality of video libraries based on the user portrait through a video recommendation module to obtain a primary recommendation result, and performing secondary screening processing on the primary recommendation result based on the user portrait and a preset recommendation rule to obtain a video recommendation data set. According to the method provided by the invention, the problem that the recommendation result of the video recommendation system in the prior art cannot be matched with the continuously changing personalized demand of the user is solved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a video recommendation method, device, equipment and storage medium. Background Art

[0002] With the development of information technology, people can obtain information resources from the Internet in more and more ways, such as reading articles, listening to audio, watching videos, etc. Among them, watching videos with rich types and rich and diverse content has become a favorite way for people to obtain information resources. How to find the required resources in the massive video resource library is a very important issue for users.

[0003] At present, the existing technology mainly uses the video recommendation system to analyze the user's historical behavior (such as the number of likes and click-through rate (CTR) of the video) and personal information, generate corresponding video recommendation results and push them to the user's terminal.

[0004] However, the video recommendation results obtained by the above video recommendation system cannot match the users' ever-changing personalized needs, and the user experience is low. Summary of the invention

[0005] The present application provides a video recommendation method, apparatus, device and storage medium, which are used to solve the problem in the prior art that the recommendation results of the video recommendation system cannot match the user's ever-changing personalized needs.

[0006] In a first aspect, the present application provides a video recommendation method, which is applied to an electronic device, wherein the electronic device includes a plurality of video libraries, and the video recommendation method includes:

[0007] In response to a video recommendation request, real-time interaction information corresponding to the video recommendation request is obtained; wherein the real-time interaction information includes the video recommendation request, at least one question information generated based on a dialogue management module in a large language model, and answer information corresponding to each question information; the large language model includes the dialogue management module, the user portrait module, and the video recommendation module; the dialogue management module, the user portrait module, and the video recommendation module are connected in sequence;

[0008] Extracting at least one feature information in the real-time interaction information through the user portrait module, and obtaining a user portrait based on the at least one feature information; wherein the user portrait includes multiple levels, each of the levels includes at least one feature attribute; each feature attribute in the user portrait corresponds to the at least one feature information in a one-to-one manner;

[0009] The video recommendation module performs a primary screening process on the multiple video libraries based on the user portrait to obtain a primary recommendation result;

[0010] The video recommendation module performs secondary screening on the primary recommendation results based on the user portrait and preset recommendation rules, and outputs a video recommendation data set.

[0011] In a possible design, extracting at least one feature information in the real-time interaction information by the user portrait module, and obtaining a user portrait based on the at least one feature information includes:

[0012] Inputting the video recommendation request, at least one question information and each answer information into the user portrait module, and extracting at least one feature information;

[0013] Determining a relationship between every two of the at least one feature information;

[0014] When the relationship indicates a sibling relationship, determining that two pieces of feature information corresponding to the sibling relationship belong to the same level;

[0015] When the relationship indicates a subordinate relationship, determining that two pieces of feature information corresponding to the subordinate relationship belong to two hierarchies having a subordinate relationship respectively;

[0016] According to the relationship between every two of the at least one feature information, the level to which each feature information belongs is determined, and a user portrait corresponding to the real-time interaction information is obtained.

[0017] In a possible design, determining the level to which each piece of feature information belongs according to the relationship between every two pieces of feature information in the at least one piece of feature information, and obtaining a user portrait corresponding to the real-time interaction information, includes:

[0018] Traversing each of the at least one feature information as target feature information;

[0019] Determine the level to which the target feature information belongs, as the target level;

[0020] The next level of the target level is determined as a target sub-level, each feature information having a subordinate relationship with the target feature information is determined as a feature attribute in the target sub-level, and the user portrait module outputs a user portrait corresponding to the real-time interaction information.

[0021] In a possible design, the video recommendation module performs a primary screening process on the multiple video libraries based on the user portrait to obtain a primary recommendation result, including:

[0022] Inputting the user portrait into the video recommendation module to determine at least one target video type based on the user portrait;

[0023] Matching a corresponding video library based on each target video type to obtain a target video library;

[0024] A third amount of video data is matched for the user portrait in each of the target video libraries to obtain a primary recommendation result; the primary recommendation result includes a second amount of video data; the third amount is less than or equal to the second amount.

[0025] In a possible design, matching a third amount of video data for the user portrait in each of the target video libraries to obtain a primary recommendation result includes:

[0026] Based on at least one feature information in the user portrait, a first matching degree between the user portrait and each video data in each target video library is calculated;

[0027] Using a preset sorting method based on a first matching degree between the user portrait and each video data in each target video library, sorting each video data in each target video library to obtain a first sequence of each target video library, and sequentially selecting a third number of video data from the first sequence of each target video library; wherein the preset sorting method includes a descending sorting method or an ascending sorting method;

[0028] A primary recommendation result is determined according to a third amount of video data corresponding to each of the target video libraries.

[0029] In a possible design, the video recommendation module performs secondary screening on the primary recommendation results based on the user portrait and preset recommendation rules to output a video recommendation data set, including:

[0030] Inputting the preset recommendation rule into the video recommendation module to filter the primary recommendation result based on the preset recommendation rule to obtain a filtered recommendation result;

[0031] Determine the weight of each level in the user portrait, and calculate the second matching degree between the user portrait and each video data in the filtered recommendation result based on the weight and the feature information in each level through the video recommendation module;

[0032] Based on the second matching degree between the user portrait and each video data in the filtered recommendation results, a first number of video data is selected from the filtered recommendation results to obtain a video recommendation data set output by the video recommendation module; wherein the first number is smaller than the second number.

[0033] In a possible design, based on the second matching degree between the user portrait and each video data in the filtered recommendation results, selecting a first number of video data in the filtered recommendation results to obtain a video recommendation data set output by the video recommendation module includes:

[0034] Using the preset sorting method, sorting each of the video data in the filtered recommendation results based on the second matching degree between the user portrait and each of the video data in the filtered recommendation results, to obtain a second sequence;

[0035] sequentially selecting a second amount of video data in the second sequence as recommended video data;

[0036] Based on the preset sorting method and the position of each recommended video data in the second sequence, a corresponding label is added to each recommended video data to obtain a video recommendation data set output by the video recommendation module.

[0037] In a possible design, in response to a video recommendation request, obtaining real-time interaction information corresponding to the video recommendation request includes:

[0038] In response to the video recommendation request, input the video recommendation request into the dialogue management module, and obtain at least one question information corresponding to the video recommendation request output by the dialogue management module;

[0039] Receiving reply information corresponding to each of the question information;

[0040] Update each of the reply information to the video recommendation request, return to execute the step of inputting the video recommendation request into the dialogue management module, obtain at least one question information corresponding to the video recommendation request output by the dialogue management module, and subsequent steps until receiving question and answer end information; wherein the question and answer end information is used to indicate the end of the dialogue corresponding to the video recommendation request.

[0041] In a second aspect, the present application provides a video recommendation device, comprising:

[0042] An acquisition module is used to respond to a video recommendation request and acquire real-time interaction information corresponding to the video recommendation request; wherein the real-time interaction information includes the video recommendation request, at least one question information generated by a dialogue management module in a large language model, and answer information corresponding to each question information; the large language model includes the dialogue management module, the user portrait module and the video recommendation module; the dialogue management module, the user portrait module and the video recommendation module are connected in sequence;

[0043] A portrait construction module, configured to extract at least one feature information in the real-time interaction information through the user portrait module, and obtain a user portrait based on the at least one feature information; wherein the user portrait includes multiple levels, each of the levels includes at least one feature attribute; each feature attribute in the user portrait corresponds one-to-one to the at least one feature information;

[0044] A screening module, configured to perform a primary screening process on the plurality of video libraries based on the user portrait through the video recommendation module to obtain a primary recommendation result;

[0045] The video recommendation module is used to perform secondary screening processing on the primary recommendation results based on the user portrait and preset recommendation rules through the video recommendation module, and output a video recommendation data set.

[0046] In a third aspect, an embodiment of the present application provides an electronic device, comprising: at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the video recommendation method described in the first aspect and various possible designs of the first aspect.

[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer execution instructions are stored. When a processor executes the computer execution instructions, the video recommendation method described in the first aspect and various possible designs of the first aspect is implemented.

[0048] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the video recommendation method described in the first aspect and various possible designs of the first aspect.

[0049] The video recommendation method, device, equipment and storage medium provided by the present application, by responding to the video recommendation request, using the dialogue management module of the large language model to conduct question and answer, obtain real-time interaction information, using the user portrait module of the large language model to extract the user's characteristic information based on the real-time interaction information, and determine the user portrait based on the user's characteristic information, the user portrait includes multiple levels of classification portraits, each level includes at least one characteristic attribute; each characteristic attribute in the user portrait corresponds to the at least one characteristic information. The large language model of the video recommendation module used performs primary screening processing on the video library based on the user portrait to obtain the primary screening result, and performs secondary screening processing on the primary recommendation result based on the user portrait and the preset recommendation rules to obtain the video recommendation data set. In this way, a multi-layer classified user portrait can be constructed based on real-time interaction information, and the video library can be screened multiple times based on the user portrait to select a video data set that matches the user's personalized needs more, which is convenient for the user to choose, and can solve the problem that the video recommendation system existing in the prior art cannot match the user's personalized needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0051] Figure 1 An application scenario diagram of a video recommendation method in the prior art;

[0052] Figure 2 The process of the video recommendation method provided in the embodiment of the present application Figure 1 ;

[0053] Figure 3 The process of the video recommendation method provided in the embodiment of the present application Figure 2 ;

[0054] Figure 4 The process of the video recommendation method provided in the embodiment of the present application Figure 3 ;

[0055] Figure 5 A schematic diagram of a user portrait provided in an embodiment of the present application;

[0056] Figure 6 The process of the video recommendation method provided in the embodiment of the present application Figure 4 ;

[0057] Figure 7 The process of the video recommendation method provided in the embodiment of the present application Figure 5 ;

[0058] Figure 8 The process of the video recommendation method provided in the embodiment of the present application Figure 6 ;

[0059] Fig. 9 The process of the video recommendation method provided in the embodiment of the present application Figure 7 ;

[0060] Fig.10 A schematic diagram of the structure of a video recommendation device provided in an embodiment of the present application;

[0061] Fig.11 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0062] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0063] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0064] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of information such as financial data or user data involved shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.

[0065] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0066] Terminology explanation:

[0067] Large Language Model: Large Language Model, or LLM for short, is a deep learning model that can be used to process natural language tasks such as question-answering and conversation. The large language model has a powerful ability to analyze complex patterns and data relationships.

[0068] The video recommendation method refers to a method of recommending one or more specific video data to users through a video recommendation system based on user needs.

[0069] With the development of information technology, people can obtain information resources from the Internet in more and more ways, such as reading articles, listening to audio, watching videos, etc. Among them, watching videos has become a favorite way for people to obtain information resources due to the rich types and rich and diverse content of videos. However, the existing technology mainly uses video recommendation systems to recommend videos to users.

[0070] Figure 1 This is an application scenario diagram of the video recommendation method in the prior art.

[0071] like Figure 1 As shown, the video recommendation method in the prior art mainly reads the user's historical behavior data and personal information through the recommendation system, and determines the video recommendation result based on the user's historical behavior data.

[0072] However, the number of video resources is very large, users have different needs for watching videos at different times, and the update speed of video resources is also very fast. The video recommendation results obtained by the above method cannot match the user's ever-changing personalized needs, resulting in a low user experience.

[0073] In view of the above technical problems, the present application proposes a video recommendation method, which is applied to electronic devices. The method includes: responding to a video recommendation request, using a dialogue management module of a large language model to conduct question and answer, obtaining real-time interaction information, using a user portrait module of a large language model to extract user feature information based on real-time interaction information, and determining a user portrait based on the user feature information, the user portrait includes multiple levels of classification portraits, each level includes at least one feature attribute; each feature attribute in the user portrait corresponds to the at least one feature information. The large language model of the video recommendation module used performs primary screening processing on the video library based on the user portrait to obtain a primary screening result, and performs secondary screening processing on the primary recommendation result based on the user portrait and preset recommendation rules to obtain a video recommendation data set. In this way, a multi-layer classified user portrait can be constructed based on real-time interaction information, and the video library can be screened multiple times based on the user portrait to select a video data set that matches the user's personalized needs more, which is convenient for users to choose, and can solve the problem that the video recommendation system in the prior art cannot match the user's personalized needs.

[0074] The above-mentioned electronic devices may include smart phones, tablet computers, laptop computers, personal computers (PCs), etc. The embodiments of the present application do not impose any special restrictions on the specific types of the electronic devices.

[0075] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0076] Figure 2 Flow chart of the video recommendation method provided in the embodiment of the present application. Figure 2 As shown, the video recommendation method includes:

[0077] S101. In response to a video recommendation request, obtain real-time interaction information corresponding to the video recommendation request; wherein the real-time interaction information includes the video recommendation request, at least one question information generated based on a dialogue management module in a large language model, and answer information corresponding to each question information; the large language model includes the dialogue management module, the user portrait module and the video recommendation module; the dialogue management module, the user portrait module and the video recommendation module are connected in sequence.

[0078] Specifically, in response to a video recommendation request, a dialogue management module based on a large language model generates at least one question information, and receives reply information corresponding to each question information, thereby obtaining real-time interaction information. The large language model includes a dialogue management module, a user portrait module, and a video recommendation module; the dialogue management module, the user portrait module, and the video recommendation module are connected in sequence.

[0079] S102. Extract at least one feature information in the real-time interaction information through the user portrait module, and obtain a user portrait based on the at least one feature information; wherein the user portrait includes multiple levels, each of the levels includes at least one feature attribute; each feature attribute in the user portrait corresponds one-to-one to the at least one feature information.

[0080] Specifically, the real-time interaction information is input into the user portrait module, and at least one feature information of the user in the real-time interaction information is extracted through the user portrait module. According to the subordinate relationship between every two feature information in the at least one feature information, the level of each feature information is determined, and a user portrait including multiple levels is constructed. The user portrait includes multiple levels, each level includes at least one feature attribute, and each feature attribute in the user portrait corresponds to the at least one feature information one by one. For each feature attribute in each level, there is at least one sub-feature attribute with a subordinate relationship in the next level.

[0081] S103: Performing a primary screening process on the multiple video libraries based on the user portrait by the video recommendation module to obtain a primary recommendation result.

[0082] Specifically, the user portrait is input into the video recommendation module, and the video recommendation module performs a primary screening process on the video data in the plurality of video libraries to obtain a primary recommendation result, which includes a second amount of video data.

[0083] S104: The video recommendation module performs secondary screening on the primary recommendation results based on the user portrait and preset recommendation rules, and outputs a video recommendation data set.

[0084] Specifically, the video recommendation module performs secondary screening on each video data in the primary recommendation result based on the user portrait and the preset recommendation rule to obtain a video recommendation data set output by the video recommendation module, which includes a first number of video data. The first number is smaller than the second number.

[0085] The first quantity and the second quantity can be set according to actual conditions. For example, the first quantity is set to 10 and the second quantity is set to 200. Alternatively, the first quantity is set to 50 and the second quantity is set to 500.

[0086] Specifically, the preset recommendation rules can be specifically set according to the actual situation. Exemplarily, the preset recommendation rules include: filtering video data containing sensitive words. Or the preset recommendation rules include: retaining video data containing specified information. The specified information may include but is not limited to product promotion information and warning information. Exemplarily, the warning information may include: do not imitate, shoot in closed roads, etc.

[0087] The above-mentioned dialogue management module, user portrait module and video recommendation module can be connected in sequence to form an overall large language model, so as to complete real-time dialogue operations through a large language model, and output the corresponding video recommendation data set based on real-time interactive information, so as to effectively identify the personalized needs of users at the current moment, break through the problem of information cocoon, and recommend video data sets that match personalized needs to users, so as to facilitate user selection. At the same time, it can improve the exposure conversion rate of video data and improve the efficiency of content resource data transmission.

[0088] Figure 3 Flow chart of step S101 of the video recommendation method provided in the embodiment of the present application. Figure 3 As shown, step S101 of the video recommendation method includes:

[0089] S1011. In response to the video recommendation request, input the video recommendation request into the dialogue management module, and obtain at least one question information corresponding to the video recommendation request output by the dialogue management module;

[0090] S1012: Receive reply information corresponding to each of the question information.

[0091] Specifically, in response to a video recommendation request, a dialogue management module is used to analyze the video recommendation request, extract key information carried in the video recommendation request, and generate at least one question corresponding to the target video type according to the extracted key information. For example, the key information includes but is not limited to the title, attributes and / or content information of the video.

[0092] Specifically, the method of receiving the reply corresponding to each question information includes one or more. When the user interacts directly through the audio and video hardware device of the current electronic device, the voice information sent by the user is used as a video recommendation request, each question information can be displayed through the display screen of the current electronic device, and the reply information fed back by the user can be received. When the user sends a video recommendation request to the current electronic device through the user terminal, the user terminal device corresponding to the video recommendation request is determined, at least one question corresponding to the video recommendation request is sent to the user terminal, and the reply information corresponding to each question information is received, so as to obtain real-time interaction information with the user.

[0093] S1013. Update each of the reply information to the video recommendation request, return to execute the step of inputting the video recommendation request into the dialogue management module, obtaining at least one question information corresponding to the video recommendation request output by the dialogue management module, and subsequent steps until receiving question and answer end information; wherein the question and answer end information is used to indicate the end of the dialogue corresponding to the video recommendation request.

[0094] Specifically, each time a reply message is received, the reply message is updated to a new video recommendation request, and the process returns to input the video recommendation request into the dialogue management module, obtains the video recommendation request output by the dialogue management module, generates at least one corresponding question message, and subsequent steps, so that the dialogue management module can generate new question information based on the new video recommendation request, and receive corresponding reply information until the question and answer end information is received.

[0095] The question-and-answer end information is used to indicate the end of the conversation corresponding to the video recommendation request. The question-and-answer end information may include an instruction sent by a user terminal device of the user to indicate the end of the question-and-answer session, or information automatically generated to indicate the end of the question-and-answer session when the number of questions and answers in the real-time interactive information with the user reaches a first question-and-answer number threshold.

[0096] Specifically, the first question-and-answer number threshold can be set according to actual needs. For example, the first question-and-answer number threshold is set to 10 times.

[0097] By continuously updating the real-time interactive information and communicating with the user, the user portrait can be continuously updated, and the user's personalized needs can be analyzed and determined in a timely manner, thereby dynamically adjusting the video recommendation data set to match the user's personalized needs. Moreover, based on the above user portrait, the above user portrait can be further updated through the next real-time interactive information, so that the user portrait information is continuously enriched and updated, providing data support for subsequent video recommendation operations.

[0098] Figure 4 This is a flow chart of step S102 of the video recommendation method provided in the embodiment of the present application. Figure 4 As shown, step S102 of the video recommendation method includes:

[0099] S1021. Input the video recommendation request, at least one question information, and each answer information into the user portrait module to extract at least one feature information.

[0100] Specifically, the video recommendation request, at least one question information, and the answer information corresponding to each question information in the real-time interaction information are input into the user portrait module, and the user portrait module extracts at least one feature information from the video recommendation request, at least one question information, and the answer information corresponding to each question information. The feature information may include information associated with the title, content, introduction, video type, duration, and / or attributes of the video in the real-time interaction information. Exemplarily, if the answer information includes "make a Hunan dish", the information related to the video content can be extracted as "Hunan cuisine".

[0101] S1022. Determine a relationship between every two pieces of the at least one piece of feature information;

[0102] S1023: when the relationship indicates a same-level relationship, determining that two pieces of feature information corresponding to the same-level relationship belong to the same level;

[0103] S1024: When the relationship indicates a subordinate relationship, determine that two pieces of feature information corresponding to the subordinate relationship belong to two hierarchies having a subordinate relationship, respectively.

[0104] Specifically, the relationship between each two feature information in the at least one feature information is identified through the user portrait module. Relationships include but are not limited to peer relationships and subordinate relationships. When the relationship between two feature information indicates a peer relationship, it is determined that the two feature information corresponding to the peer relationship belong to the same level. When the relationship between two feature information indicates a subordinate relationship, it is determined that the two feature information corresponding to the subordinate relationship belong to two different levels respectively, and there is a subordinate relationship between the two different levels.

[0105] Exemplarily, the feature information extracted by the user portrait module includes: Hunan cuisine, spicy pork, Sichuan cuisine, mapo tofu, and spicy chicken. It can be determined that mapo tofu and spicy chicken are in the same level, mapo tofu and Sichuan cuisine are subordinate to each other, and spicy chicken and Sichuan cuisine are subordinate to each other. Sichuan cuisine and Hunan cuisine are in the same level, and spicy pork and Hunan cuisine are subordinate to each other. Based on this, it can be determined that Sichuan cuisine and Hunan cuisine are at the same level, mapo tofu, spicy chicken, and spicy pork are at the same level, and the level where mapo tofu, spicy chicken, and spicy pork are located is subordinate to the level where Sichuan cuisine and Hunan cuisine are located.

[0106] S1025. Determine the level to which each piece of feature information belongs based on the relationship between every two pieces of feature information in the at least one piece of feature information, and obtain a user portrait corresponding to the real-time interaction information.

[0107] Specifically, the user portrait module outputs a user portrait corresponding to the real-time interaction information, including multiple levels, and each level including at least one characteristic attribute, based on the level to which each characteristic information in the at least one characteristic information belongs. Among them, at least one characteristic attribute in the user portrait corresponds one-to-one to at least one characteristic information.

[0108] In some embodiments, step S1025 includes:

[0109] Traversing each of the at least one feature information as target feature information;

[0110] Determine the level to which the target feature information belongs, as the target level;

[0111] The next level of the target level is determined as a target sub-level, each feature information having a subordinate relationship with the target feature information is determined as a feature attribute in the target sub-level, and the user portrait module outputs a user portrait corresponding to the real-time interaction information.

[0112] Specifically, each feature information in at least one feature information is traversed as a target attribute, the level where the target attribute is located is determined to be the target level, each feature information having a same-level relationship with the target attribute is determined to be a feature attribute in the target level, the next level of the target level is determined to be a target sub-level of the target level, each feature information having a subordinate relationship with the target attribute is determined to be a feature attribute of the target sub-level, and the feature attribute corresponding to each feature information is determined accordingly, so as to obtain a user portrait corresponding to the real-time interaction information output by the user portrait module.

[0113] It is understandable that the feature attributes can be used to indicate the location of the feature information in the user portrait and the relationship between the feature information. Exemplary, the feature information includes Hunan cuisine, stir-fried pork with peppers, Sichuan cuisine, mapo tofu, and spicy chicken. Figure 5 The user portrait of user A is shown.

[0114] Figure 5 In the example, the target video type of user A's portrait is a food video type. The first-level feature attributes include Sichuan cuisine and coriander, and the second-level feature attributes include fried pork with peppers, mapo tofu, and spicy chicken.

[0115] according to Figure 5 It can be seen that the two feature information of Sichuan cuisine and Hunan cuisine have a peer relationship, and the three feature information of fried pork with pepper, mapo tofu and spicy chicken have a peer relationship, while the two feature information of Hunan cuisine and fried pork with pepper have a subordinate relationship, and the two feature information of mapo tofu and spicy chicken both have a subordinate relationship with the feature information of Sichuan cuisine.

[0116] Figure 6 This is a flow chart of step S103 of the video recommendation method provided in the embodiment of the present application. Figure 6 As shown, step S103 of the video recommendation method includes:

[0117] S1031, inputting the user portrait into the video recommendation module to determine at least one target video type based on the user portrait;

[0118] S1032, matching a corresponding video library based on each target video type to obtain a target video library;

[0119] S1033. Match a third amount of video data for the user portrait in each of the target video libraries to obtain a primary recommendation result; the primary recommendation result includes a second amount of video data; the third amount is less than or equal to the second amount.

[0120] Specifically, the electronic device is connected to multiple video libraries, each of which includes multiple video data, and the video types of every two video libraries are different; the user portrait is input into the video recommendation module, and the video recommendation module is used to determine at least one target video type based on at least one characteristic attribute in the user portrait. A video library corresponding to each target video type is matched from the multiple video libraries as the target video library. In each target video library, a corresponding third amount of video data is matched based on at least one characteristic information in the user portrait to obtain a primary recommendation result.

[0121] For example, the characteristic information includes Sichuan cuisine, mapo tofu, Ronaldo, and football. Based on Sichuan cuisine and mapo tofu, a target video type can be determined as a food video, and based on star A and football, a target video type can be determined as a football video. A third number of video data corresponding to the user portrait and including Sichuan cuisine and mapo tofu content is matched from the food video, and a third number of video data corresponding to the user portrait and including star A and football-related content is matched from the football video.

[0122] It is understandable that when there is only one target video type, the third number may be equal to the second number. When there is more than one target video type, the sum of the third numbers corresponding to each target video type is equal to the second number.

[0123] In actual applications, the video data in the video library is constantly updated. Video data that matches the user portrait at the current moment is screened out from the constantly updated video library, so that the video recommendation results are real-time and match the user's real-time personalized needs.

[0124] Figure 7 This is a flow chart of step S1033 of the video recommendation method provided in the embodiment of the present application. Figure 7 As shown, step S1033 of the video recommendation method includes:

[0125] S10331. Based on at least one feature information in the user portrait, calculate a first matching degree between the user portrait and each video data in each target video library;

[0126] S10332, using a preset sorting method to sort each video data in each target video library based on a first matching degree between the user portrait and each video data in each target video library, to obtain a first sequence of each target video library, and sequentially selecting a third number of video data from the first sequence of each target video library; wherein the preset sorting method includes a descending sorting method or an ascending sorting method;

[0127] S10333. Determine a primary recommendation result according to the third amount of video data corresponding to each of the target video libraries.

[0128] Specifically, based on at least one feature information in the user portrait and key information of each video data in each target video library, a first matching degree between the user portrait and each video data in each target video library is calculated, wherein the key information includes at least one of the title, content, type, etc. of the video data.

[0129] Specifically, a preset sorting method is used to sort each video data in each target video library based on the first matching degree between the user portrait and each video data in each target video library to obtain a first sequence corresponding to each target video library. Based on the first matching sequence corresponding to each target video library, the third number of first matching degrees corresponding to each target video library and the third number of first matching degrees corresponding to each target video library are selected in sequence to determine the third number of video data corresponding to each target video library, and the third number of video data corresponding to each target video library is determined as the primary recommendation result. The preset sorting method can be specifically set according to needs.

[0130] Exemplarily, the preset sorting method includes descending order. In this way, the first matching degree sequence obtained based on the preset sorting method is a sequence from large to small, and the first matching degrees of the third number selected in sequence include the first matching degrees ranked in the front third number from the largest value to the largest value. For example, if the third number is 30, 30 first matching degrees ranked 30th from the largest value to the largest value can be selected.

[0131] Figure 8 This is a flow chart of step S104 of the video recommendation method provided in the embodiment of the present application. Figure 8 As shown, step S104 of the video recommendation method includes:

[0132] S1041, inputting the preset recommendation rule into the video recommendation module to filter the primary recommendation result based on the preset recommendation rule to obtain a filtered recommendation result;

[0133] S1042, determining the weight of each level in the user portrait, and calculating, by the video recommendation module, a second matching degree between the user portrait and each video data in the filtered recommendation result based on the weight and feature information in each level;

[0134] S1043. Based on the second matching degree between the user portrait and each video data in the filtered recommendation results, select a first number of video data from the filtered recommendation results to obtain a video recommendation data set output by the video recommendation module; wherein the first number is smaller than the second number.

[0135] Specifically, the preset recommendation rules are input into the video recommendation module, and the primary recommendation results are filtered by the video recommendation module based on the preset recommendation rules, and the video data that is not recommended in the primary recommendation results is screened out to obtain the filtered recommendation results. In this way, the number of video data in the filtered recommendation results is less than or equal to the second number. The weight of each level in the user portrait is determined, and the video recommendation module is used to calculate the second matching degree between the user portrait and each video data in the filtered recommendation results based on the weight of each level and at least one feature information in each level. Based on the second matching degree between the user portrait and each video data in the filtered recommendation results, a first number of video data that meets the first condition is selected in the filtered recommendation results to obtain a video recommendation set, and the first number is less than the second number.

[0136] The first condition may be specifically set according to actual conditions. For example, the first condition includes video data whose numerical ranking is in the first numerical position in the filtered recommendation results.

[0137] Fig. 9 Flow chart of step S101 of the video recommendation method provided in the embodiment of the present application. Fig. 9 As shown, step S1043 of the video recommendation method includes:

[0138] S10431, sorting each video data in the filtered recommendation results based on the second matching degree between the user portrait and each video data in the filtered recommendation results using the preset sorting method to obtain a second sequence;

[0139] S10432, sequentially selecting a second amount of video data in the second sequence as recommended video data;

[0140] Specifically, a preset sorting method is used to sort each corresponding video data in the filtered recommendation results based on the second matching degree between the user portrait and each video data in the filtered recommendation results, so as to obtain a second sequence of all video data in the filtered recommendation results. A second number of video data are sequentially selected in the second sequence of the filtered recommendation results as the recommended video data.

[0141] It is understandable that, in order to improve the stability of the recommendation results, the sorting method of the first matching degree is the same as the sorting method of the second matching degree.

[0142] As an example but not limitation, by sorting the second matching degrees in descending order and then selecting a second number of second matching degrees in sequence, the recommended video data determined based on the second number of second matching degrees selected above can be: the second number of video data with a higher second matching degree with the user portrait.

[0143] S10433. Based on the preset sorting method and the position of each recommended video data in the second sequence, add a corresponding label to each recommended video data to obtain a video recommendation data set output by the video recommendation module.

[0144] Specifically, for each recommended video data, based on a preset sorting method and a second matching degree corresponding to each video, a label is added to each corresponding recommended video data to obtain a video recommendation data set output by the video recommendation module.

[0145] Exemplarily, the label of the first data in the video recommendation data set is: the video data with the greatest matching degree with the user's video recommendation request.

[0146] By adding labels to each recommended video data in the video recommendation dataset, users can quickly understand the reasons for recommending the video data, which is more in line with the user's personalized needs.

[0147] Fig.10 This is a schematic diagram of the structure of the video recommendation device provided in the embodiment of the present application. Fig.10 As shown, the video recommendation device includes:

[0148] The acquisition module 1001 is used to respond to a video recommendation request and acquire real-time interaction information corresponding to the video recommendation request; wherein the real-time interaction information includes the video recommendation request, at least one question information generated by a dialogue management module in a large language model, and answer information corresponding to each question information; the large language model includes the dialogue management module, the user portrait module and the video recommendation module; the dialogue management module, the user portrait module and the video recommendation module are connected in sequence;

[0149] A portrait construction module 1002 is used to extract at least one feature information in the real-time interaction information through the user portrait module, and obtain a user portrait based on the at least one feature information; wherein the user portrait includes multiple levels, each of the levels includes at least one feature attribute; each feature attribute in the user portrait corresponds to the at least one feature information in a one-to-one manner;

[0150] A screening module 1003 is used to perform a primary screening process on the multiple video libraries based on the user portrait through the video recommendation module to obtain a primary recommendation result;

[0151] The video recommendation module 1004 is used to perform secondary screening processing on the primary recommendation results based on the user portrait and preset recommendation rules through the video recommendation module, and output a video recommendation data set.

[0152] In some embodiments, the portrait building module includes:

[0153] A feature information extraction unit, configured to input the video recommendation request, at least one question information, and each answer information into the user portrait module, and extract at least one feature information;

[0154] a relationship determining unit, configured to determine a relationship between every two pieces of the at least one piece of feature information;

[0155] A first level determination unit, configured to determine, when the relationship indicates a same-level relationship, that two pieces of feature information corresponding to the same-level relationship belong to the same level;

[0156] A second level determination unit, configured to, when the relationship indicates a subordinate relationship, determine that two pieces of feature information corresponding to the subordinate relationship respectively belong to two levels having a subordinate relationship;

[0157] The portrait construction unit is used to determine the level to which each piece of feature information belongs according to the relationship between every two pieces of feature information in the at least one feature information, and obtain a user portrait corresponding to the real-time interaction information.

[0158] In some embodiments, the portrait construction unit includes:

[0159] A traversal subunit, used for traversing each of the at least one feature information as target feature information;

[0160] A level determination subunit, used to determine the level to which the target feature information belongs, which is the target level;

[0161] The portrait construction subunit is used to determine the next level of the target level as the target sublevel, determine each feature information having a subordinate relationship with the target feature information as the feature attribute in the target sublevel, and the user portrait module outputs the user portrait corresponding to the real-time interaction information.

[0162] In some embodiments, the screening module comprises:

[0163] a video type determination unit, configured to input the user portrait into the video recommendation module to determine at least one target video type based on the user portrait;

[0164] A video library matching unit, used for matching a corresponding video library based on each target video type to obtain a target video library;

[0165] A primary recommendation unit is used to match a third amount of video data for the user portrait in each of the target video libraries to obtain a primary recommendation result; the primary recommendation result includes a second amount of video data; the third amount is less than or equal to the second amount.

[0166] In some embodiments, the primary recommendation unit includes:

[0167] A first matching degree determination subunit, configured to calculate a first matching degree between the user portrait and each video data in each target video library based on at least one feature information in the user portrait;

[0168] The video data determination subunit is used to sort the video data in each of the target video libraries based on the first matching degree between the user portrait and each video data in each of the target video libraries in a preset sorting manner to obtain a first sequence of each of the target video libraries, and sequentially select a third number of video data from the first sequence of each of the target video libraries; wherein the preset sorting manner includes a descending sorting method or an ascending sorting method;

[0169] The primary recommendation subunit is used to determine the primary recommendation result according to the third amount of video data corresponding to each of the target video libraries.

[0170] In some embodiments, the video recommendation module includes:

[0171] A filtering unit, configured to input the preset recommendation rule into the video recommendation module, so as to filter the primary recommendation result based on the preset recommendation rule to obtain a filtered recommendation result;

[0172] A second matching degree determination unit is used to determine the weight of each level in the user portrait, and calculate the second matching degree between the user portrait and each video data in the filtered recommendation result based on the weight and the feature information in each level through the video recommendation module;

[0173] A video recommendation unit is used to select a first number of video data from the filtered recommendation results based on a second matching degree between the user portrait and each video data in the filtered recommendation results, to obtain a video recommendation data set output by the video recommendation module; wherein the first number is smaller than the second number.

[0174] In some embodiments, the video recommendation unit includes:

[0175] A second sequence determination subunit is used to sort each of the video data in the filtered recommendation results based on the second matching degree between the user portrait and each of the video data in the filtered recommendation results by using the preset sorting method to obtain a second sequence;

[0176] a video recommendation subunit, configured to sequentially select a second amount of video data in a second sequence as recommended video data;

[0177] The label unit is used to add a corresponding label to each of the recommended video data based on the preset sorting method and the position of each of the recommended video data in the second sequence to obtain the video recommendation data set output by the video recommendation module.

[0178] In some embodiments, the acquisition module includes:

[0179] a question information generating unit, configured to, in response to the video recommendation request, input the video recommendation request into the dialogue management module, and obtain at least one question information corresponding to the video recommendation request output by the dialogue management module;

[0180] An information sending unit, used for receiving reply information corresponding to each of the question information;

[0181] A loop unit is used to update each of the reply information to the video recommendation request, return to execute the steps of inputting the video recommendation request into the dialogue management module, obtaining at least one question information corresponding to the video recommendation request output by the dialogue management module, and subsequent steps until receiving the question and answer end information; wherein the question and answer end information is used to indicate the end of the dialogue corresponding to the video recommendation request.

[0182] The video recommendation device provided in the embodiment of the present application can be used to implement the technical solution of the video recommendation method in the above embodiment. Its implementation principle and technical effect are similar and will not be repeated here.

[0183] It should be noted that it should be understood that the division of the various modules of the above device is only a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. For example, the portrait construction module 1002 can be a separately established processing element, or it can be integrated in a chip of the above device. In addition, it can also be stored in the memory of the above device in the form of program code, and called and executed by a processing element of the above device. The function of the above portrait construction module 1002. The implementation of other modules is similar. In addition, all or part of these modules can be integrated together, or they can be implemented independently. The processing element here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each module above can be completed by an integrated logic circuit of hardware in the processor element or instructions in the form of software.

[0184] Fig.11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig.11As shown, the electronic device may include: a transceiver 1101 , a processor 1102 , and a memory 1103 .

[0185] The processor 1102 executes the computer execution instructions stored in the memory, so that the processor 1102 executes the scheme in the above embodiment. The processor 1102 can be a general-purpose processor, including a central processing unit CPU, a network processor (network processor, NP), etc.; it can also be a digital signal processor DSP, an application-specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0186] The memory 1103 is connected to the processor 1102 via a system bus and completes communication between them. The memory 1103 is used to store computer program instructions.

[0187] The transceiver 1101 may be used to obtain tasks to be executed and configuration information of the tasks to be executed.

[0188] The system bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The transceiver is used to realize the communication between the database access device and other computers (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM) and may also include non-volatile memory.

[0189] An embodiment of the present application also provides a chip for running instructions, which is used to execute the technical solution of the video recommendation method in the above embodiment.

[0190] The embodiment of the present application further provides a computer-readable storage medium, in which computer instructions are stored. When the computer instructions are executed on a computer, the computer executes the technical solution of the video recommendation method of the above embodiment.

[0191] An embodiment of the present application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When at least one processor executes the computer program, the technical solution of the video recommendation method in the above embodiment can be implemented.

[0192] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0193] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to implement the solution of this embodiment.

[0194] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The above-mentioned module-composed unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0195] The above-mentioned integrated module implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the methods of various embodiments of the present application.

[0196] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor.

[0197] The memory may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a magnetic disk or an optical disk, etc.

[0198] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.

[0199] The above storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0200] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the storage medium can also exist as discrete components in an electronic control unit or a main control device.

[0201] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0202] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A video recommendation method, characterized in that: Applied in an electronic device, the electronic device includes a plurality of video libraries, and the video recommendation method includes: In response to a video recommendation request, real-time interaction information corresponding to the video recommendation request is obtained; wherein the real-time interaction information includes the video recommendation request, at least one question information generated based on a dialogue management module in a large language model, and answer information corresponding to each question information; the large language model includes the dialogue management module, the user portrait module, and the video recommendation module; the dialogue management module, the user portrait module, and the video recommendation module are connected in sequence; Extracting at least one feature information in the real-time interaction information through the user portrait module, and obtaining a user portrait based on the at least one feature information; wherein the user portrait includes multiple levels, each of the levels includes at least one feature attribute; each feature attribute in the user portrait corresponds to the at least one feature information in a one-to-one manner; The video recommendation module performs a primary screening process on the multiple video libraries based on the user portrait to obtain a primary recommendation result; The video recommendation module performs secondary screening on the primary recommendation results based on the user portrait and preset recommendation rules, and outputs a video recommendation data set.

2. The method according to claim 1, characterized in that The extracting at least one feature information in the real-time interaction information by the user portrait module, and obtaining a user portrait based on the at least one feature information, includes: Inputting the video recommendation request, at least one question information and each answer information into the user portrait module, and extracting at least one feature information; Determining a relationship between every two of the at least one feature information; When the relationship indicates a sibling relationship, determining that two pieces of feature information corresponding to the sibling relationship belong to the same level; When the relationship indicates a subordinate relationship, determining that two pieces of feature information corresponding to the subordinate relationship belong to two hierarchies having a subordinate relationship respectively; According to the relationship between every two of the at least one feature information, the level to which each feature information belongs is determined, and a user portrait corresponding to the real-time interaction information is obtained.

3. The method according to claim 2, characterized in that The determining, according to the relationship between every two of the at least one feature information, the level to which each of the feature information belongs, and obtaining a user portrait corresponding to the real-time interaction information, includes: Traversing each of the at least one feature information as target feature information; Determine the level to which the target feature information belongs, as the target level; The next level of the target level is determined as a target sub-level, each feature information having a subordinate relationship with the target feature information is determined as a feature attribute in the target sub-level, and the user portrait module outputs a user portrait corresponding to the real-time interaction information.

4. The method according to claim 1 or 2, characterized in that: The video recommendation module performs a primary screening process on the multiple video libraries based on the user portrait to obtain a primary recommendation result, including: Inputting the user portrait into the video recommendation module to determine at least one target video type based on the user portrait; Matching a corresponding video library based on each target video type to obtain a target video library; A third amount of video data is matched for the user portrait in each of the target video libraries to obtain a primary recommendation result; the primary recommendation result includes a second amount of video data; the third amount is less than or equal to the second amount.

5. The method according to claim 4, characterized in that The matching of the third amount of video data for the user portrait in each of the target video libraries to obtain a primary recommendation result includes: Based on at least one feature information in the user portrait, a first matching degree between the user portrait and each video data in each target video library is calculated; Using a preset sorting method based on a first matching degree between the user portrait and each video data in each target video library, sorting each video data in each target video library to obtain a first sequence of each target video library, and sequentially selecting a third number of video data from the first sequence of each target video library; wherein the preset sorting method includes a descending sorting method or an ascending sorting method; A primary recommendation result is determined according to a third amount of video data corresponding to each of the target video libraries.

6. The method according to any one of claims 1 to 5, characterized in that The video recommendation module performs secondary screening on the primary recommendation results based on the user portrait and preset recommendation rules, and outputs a video recommendation data set, including: Inputting the preset recommendation rule into the video recommendation module to filter the primary recommendation result based on the preset recommendation rule to obtain a filtered recommendation result; Determine the weight of each level in the user portrait, and calculate the second matching degree between the user portrait and each video data in the filtered recommendation result based on the weight and the feature information in each level through the video recommendation module; Based on the second matching degree between the user portrait and each video data in the filtered recommendation results, a first number of video data is selected from the filtered recommendation results to obtain a video recommendation data set output by the video recommendation module; wherein the first number is smaller than the second number.

7. The method according to claim 6, characterized in that The selecting a first amount of video data from the filtered recommendation results based on the second matching degree between the user portrait and each video data in the filtered recommendation results to obtain a video recommendation data set output by the video recommendation module includes: Using the preset sorting method, sorting each of the video data in the filtered recommendation results based on the second matching degree between the user portrait and each of the video data in the filtered recommendation results, to obtain a second sequence; sequentially selecting a second amount of video data in the second sequence as recommended video data; Based on the preset sorting method and the position of each recommended video data in the second sequence, a corresponding label is added to each recommended video data to obtain a video recommendation data set output by the video recommendation module.

8. The method according to any one of claims 1 to 7, characterized in that The step of obtaining, in response to the video recommendation request, real-time interaction information corresponding to the video recommendation request includes: In response to the video recommendation request, input the video recommendation request into the dialogue management module, and obtain at least one question information corresponding to the video recommendation request output by the dialogue management module; Receiving reply information corresponding to each of the question information; Update each of the reply information to the video recommendation request, return to execute the step of inputting the video recommendation request into the dialogue management module, obtain at least one question information corresponding to the video recommendation request output by the dialogue management module, and subsequent steps until receiving question and answer end information; wherein the question and answer end information is used to indicate the end of the dialogue corresponding to the video recommendation request.

9. A video recommendation device, characterized in that: include: An acquisition module is used to respond to a video recommendation request and acquire real-time interaction information corresponding to the video recommendation request; wherein the real-time interaction information includes the video recommendation request, at least one question information generated by a dialogue management module in a large language model, and answer information corresponding to each question information; the large language model includes the dialogue management module, the user portrait module and the video recommendation module; the dialogue management module, the user portrait module and the video recommendation module are connected in sequence; A portrait construction module, configured to extract at least one feature information in the real-time interaction information through the user portrait module, and obtain a user portrait based on the at least one feature information; wherein the user portrait includes multiple levels, each of the levels includes at least one feature attribute; each feature attribute in the user portrait corresponds one-to-one to the at least one feature information; A screening module, configured to perform a primary screening process on the plurality of video libraries based on the user portrait through the video recommendation module to obtain a primary recommendation result; The video recommendation module is used to perform secondary screening processing on the primary recommendation results based on the user portrait and preset recommendation rules through the video recommendation module, and output a video recommendation data set.

10. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 8 when executed by a processor.

12. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 8 when being executed by a processor.

Images