Video recommendation method, device and equipment and readable storage medium

Abstract

The invention discloses a video recommendation method, device and equipment and a readable storage medium. The method can evaluate whether a video recommendation demand needs to be subjected to demand definition improvement or not; if yes, generating a guide type statement matched with the video recommendation demand; initiating a questioning operation corresponding to the guide type statement to the target user to obtain a reply text of the target user; updating the video recommendation demand based on the reply text, and returning to the step of evaluating whether demand definition improvement needs to be carried out on the video recommendation demand or not; if not, querying a plurality of pieces of film and television information based on the video recommendation demand; evaluating the integrating degree and richness of each piece of film and television information and the video recommendation demand; when the fitting degree meets a preset fitting requirement and the richness meets a preset richness requirement, sorting the film and television information to generate a video recommendation result; therefore, the personalized video recommendation result can be provided for the target user in a targeted manner, and the use experience and the satisfaction degree of the user are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent recommendation technology, and more specifically, to a video recommendation method, device, equipment and readable storage medium. Background Art

[0002] In daily life, watching videos is a common form of leisure and entertainment. However, with the explosive growth of video content such as short videos, TV series and movies, users are troubled by information overload. Users often need to spend a lot of time screening and identifying video content that suits their interests from a large amount of video content. On this basis, how to help users quickly and efficiently obtain the video content they are interested in has become an important issue that needs to be solved urgently. Summary of the invention

[0003] In view of this, the present application provides a video recommendation method, apparatus, device and readable storage medium to help users quickly and efficiently obtain video content of their interest.

[0004] In order to achieve the above objectives, the proposed solution is as follows:

[0005] A video recommendation method, comprising:

[0006] Respond to the target user's video recommendation needs and evaluate whether the video recommendation needs need to be clarified;

[0007] If yes, the video recommendation requirement is analyzed to generate a guiding sentence matching the video recommendation requirement; a question operation corresponding to the guiding sentence is initiated to the target user to obtain a reply text from the target user; based on the reply text, the video recommendation requirement is updated, and the step of evaluating whether the video recommendation requirement needs to be improved in terms of requirement clarity is returned to be executed;

[0008] If not, based on the video recommendation needs, multiple film and television information are queried; the degree of fit between each piece of film and television information and the video recommendation needs and the recommendation richness formed by each piece of film and television information are evaluated; when the fit meets the preset fit requirements and the recommendation richness meets the preset richness requirements, the each piece of film and television information is sorted to generate a video recommendation result.

[0009] Optionally, the evaluating whether it is necessary to improve the clarity of the video recommendation requirement includes:

[0010] Segmenting the video recommendation requirement to determine whether there is a classification field for the video recommendation requirement;

[0011] If so, it is determined that there is no need to improve the clarity of the video recommendation requirement;

[0012] If not, it is determined that the video recommendation requirement needs to be improved in terms of clarity.

[0013] Optionally, analyzing the video recommendation requirement and generating a guiding statement matching the video recommendation requirement includes:

[0014] Determine user profiles for indicating the film and television preferences of target users at different time periods;

[0015] Determine a demand period corresponding to the video recommendation demand;

[0016] The guiding sentence is generated based on the demand time period, the video recommendation demand and the user portrait.

[0017] Optionally, the generating the guiding statement based on the required time period, the video recommendation requirement and the user portrait includes:

[0018] Utilizing the trained decision model, based on the video recommendation demand, taking the film and television preferences in the user portrait that match the demand period as the main key element, and taking the film and television preferences in the user portrait except the film and television preferences in the demand period as the secondary key elements, a plurality of guiding sentences with a progressive relationship are generated.

[0019] Optionally, the querying of multiple film and television information based on the video recommendation requirement includes:

[0020] Parsing the video recommendation requirement and extracting keywords of the video recommendation requirement;

[0021] Based on the keyword, determining the database type corresponding to the video recommendation requirement;

[0022] A plurality of film and television information matching the video recommendation requirement is queried from a database matching the database type.

[0023] Optionally, also include:

[0024] When the fit degree meets the preset fit requirement and the recommendation richness does not meet the richness requirement, the network is queried for film and television information that matches the video recommendation requirement, and the step of evaluating the recommendation richness formed by each film and television information is returned to be executed.

[0025] Optionally, the step of sorting the film and television information to generate video recommendation results includes:

[0026] Perform entity extraction for video recommendation needs;

[0027] Based on the extracted entities and the popularity of each film and television information, calculate the ranking score of each film and television information;

[0028] Based on the ranking score of each piece of film and television information, the various film and television information are sorted to obtain the video recommendation results.

[0029] A video recommendation device, comprising:

[0030] An evaluation module is used to respond to the target user's video recommendation needs and evaluate whether the video recommendation needs need to be clarified; if so, the questioning module is called; if not, the generation module is called;

[0031] The questioning module is used to analyze the video recommendation demand and generate a guiding sentence matching the video recommendation demand; initiate a question operation corresponding to the guiding sentence to the target user and obtain a reply text from the target user; based on the reply text, update the video recommendation demand and return to the execution evaluation module;

[0032] A generation module is used to query multiple film and television information based on video recommendation needs; evaluate the degree of fit between each piece of film and television information and the video recommendation needs and the recommendation richness formed by each piece of film and television information; when the fit meets the preset fit requirements and the recommendation richness meets the preset richness requirements, sort the various film and television information to generate video recommendation results.

[0033] A video recommendation device, comprising a memory and a processor;

[0034] The memory is used to store programs;

[0035] The processor is used to execute the program to implement each step of the above-mentioned video recommendation method.

[0036] A readable storage medium stores a computer program thereon, wherein when the computer program is executed by a processor, each step of the above-mentioned video recommendation method is implemented.

[0037] It can be seen from the above technical solutions that the video recommendation method provided by the present application can respond to the video recommendation needs of the target users and evaluate whether it is necessary to improve the clarity of the video recommendation needs; if so, analyze the video recommendation needs and generate a guiding sentence matching the video recommendation needs; initiate a question operation corresponding to the guiding sentence to the target user to obtain a reply text from the target user; based on the reply text, update the video recommendation needs and return to execute the step of evaluating whether it is necessary to improve the clarity of the video recommendation needs; if not, query multiple film and television information based on the video recommendation needs; query multiple film and television information based on the video recommendation needs; evaluate the degree of fit between each piece of film and television information and the video recommendation needs and the recommendation richness formed by each piece of film and television information; when the fit meets the preset fit requirement and the recommendation richness meets the preset richness requirement, sort each piece of film and television information to generate a video recommendation. Recommendation results; on this basis, this application can generate guiding questions by conducting in-depth analysis of the video recommendation needs when the target user's video recommendation needs are not clear enough, such as problems such as polysemy or ambiguity; through guiding questions, guide the target user to provide more demand details and key elements to update the video recommendation needs, make the video recommendation needs more targeted, and improve the personalization and pertinence of subsequent video recommendations; when clarification is needed, this application does not need to conduct guided questions and directly enter the subsequent steps, saving unnecessary repeated communication and improving recommendation efficiency; then, the content relevance and recommendation quantity of the multiple film and television information queried can be measured from the fit and recommendation richness to measure whether they meet the needs, so as to provide rich recommendation content closely related to the needs of the target user, and at the same time, further sort the various film and television contents that meet the fit and recommendation richness, highlight the key film and television information in the video sorting results, and further facilitate the target user to select personalized videos. It can be seen that this application can provide personalized video recommendation results for target users in a targeted manner, improving the user experience and satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0039] Figure 1 A flow chart of a video recommendation method disclosed in an embodiment of the present application;

[0040] Figure 2 This is a structural block diagram of a video recommendation device disclosed in an embodiment of the present application;

[0041] Figure 3 This is a hardware structure block diagram of a video recommendation device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0043] The embodiment of the present application provides a video recommendation method, which can be applied to various video platforms or TV set-top boxes, and can also be applied to various computer terminals or smart terminals. The execution subject can be a processor or server of the computer terminal or smart terminal. The method flow chart of the video recommendation method is as follows: Figure 1 As shown, specifically including:

[0044] Next, combine Figure 1 The video recommendation method of this application is introduced in detail, including the following steps:

[0045] Step S1, responding to the target user's video recommendation demand, evaluating whether it is necessary to improve the demand clarity of the video recommendation demand; if so, executing step S2; if not, executing step S3.

[0046] Specifically, the target user may input the video recommendation demand in a variety of ways; for example, the target user may respond to the video recommendation demand input by voice; or the target user may respond to the video recommendation demand input by characters.

[0047] The target user can be a user who has made a video recommendation request, a robot or an ordinary user.

[0048] Video recommendation requirements can be presented in text form.

[0049] When receiving the video recommendation demand of the target user, it can evaluate whether the video recommendation demand is clear and whether matching film and television content can be found through the database;

[0050] If the clarity of the video recommendation requirement is not high, it is necessary to improve the clarity of the video recommendation requirement so as to provide customized film and television information to the target user. At this time, step S2 can be executed.

[0051] If the clarity of the video recommendation requirement meets the basic requirement, you can try to query the film and television information. At this time, you can execute step S3.

[0052] Step S2, analyzing the video recommendation demand and generating a guiding statement matching the video recommendation demand; initiating a question operation corresponding to the guiding statement to the target user and obtaining a reply text from the target user; based on the reply text, updating the video recommendation demand and returning to execute step S1.

[0053] Specifically, semantic understanding and intent recognition can be performed on the video recommendation demand, and the video recommendation demand can be reconstructed and expanded to form three or more leading question sentences with a progressive relationship.

[0054] A question operation corresponding to the guiding question sentence can be initiated to the target user, and a reply text of each guiding question sentence can be obtained;

[0055] Keywords can be extracted from each reply text, and the video recommendation requirements can be adjusted and updated based on each reply text to form new video recommendation requirements;

[0056] After forming a new video recommendation demand, the process may return to step S1.

[0057] Step S3: based on the video recommendation requirement, query multiple film and television information; evaluate the degree of fit between each film and television information and the video recommendation requirement and the recommendation richness formed by each film and television information; when the fit meets the preset fit requirement and the recommendation richness meets the preset richness requirement, sort the film and television information to generate a video recommendation result.

[0058] Specifically, based on the video recommendation requirement, matching film and television information can be queried from the database.

[0059] Among them, each film and television information may include one or more data related to the film and television content, such as the corresponding film and television name, starring identification, film and television type, director identification, story background, story theme, release location, release time, plot introduction, popular lines and popular scenes.

[0060] Among them, the degree of fit between each film and television information and the video recommendation requirement can be calculated in a variety of ways, and whether the degree of fit meets the preset fit requirement can be determined;

[0061] For example, various film and television information and video recommendation requirements can be input into the decision model. Through the decision model, the degree of fit between each film and television information and the video recommendation requirement is calculated, and each degree of fit is evaluated to see whether it meets the preset fit requirements.

[0062] Furthermore, the decision model can be used to calculate the completeness score, timeliness score, relevance score and accuracy score between each piece of film and television information and the video recommendation requirements; based on the completeness score, timeliness score, relevance score and accuracy score of each piece of film and television information, the degree of fit of the corresponding film and television information is calculated.

[0063] The Euclidean distance between the keywords, background and / or theme of each film and television information and the video recommendation requirement may also be calculated, and each Euclidean distance may be used as the degree of fit between the corresponding film and television information and the video recommendation requirement, and then it may be determined whether the degree of fit meets the fit requirement.

[0064] The film and television information whose degree of conformity does not meet the conformity requirements is deleted. When all the film and television information does not meet the conformity requirements, the process returns to step S1.

[0065] The recommendation richness after combining the various film and television information can be calculated in a variety of ways, and it can be determined whether the recommendation richness meets the preset richness requirements;

[0066] For example, various film and television information and video recommendation requirements can be input into the decision model. Through the decision model, the recommendation richness achieved after combining various film and television information is calculated, and it is evaluated whether the recommendation richness meets the preset richness requirements.

[0067] The quantity of each piece of film and television information may also be calculated, and based on the ratio between the quantity of film and television information and a preset quantity standard, the ratio may be used as a recommended richness, and then it may be determined whether the recommended richness meets the richness requirement.

[0068] The recommendation richness may also be determined to be high quality, low quality, or uncertain based on the quantity of each film and television information. When the recommendation richness is high quality, it is determined that the recommendation richness meets the richness requirement.

[0069] It should be noted that the matching requirements and the enrichment requirements can be determined based on the target users' feedback on the historical video recommendation results.

[0070] It can be seen from the above technical solutions that the video recommendation method provided by the present application can respond to the video recommendation needs of the target users and evaluate whether it is necessary to improve the clarity of the video recommendation needs; if so, analyze the video recommendation needs and generate a guiding sentence that matches the video recommendation needs; initiate a question operation corresponding to the guiding sentence to the target user to obtain a reply text from the target user; based on the reply text, update the video recommendation needs and return to execute the step of evaluating whether it is necessary to improve the clarity of the video recommendation needs; if not, query multiple film and television information based on the video recommendation needs; evaluate the degree of fit between each film and television information and the video recommendation needs and the recommendation richness formed by each film and television information; when the fit meets the preset fit requirement and the recommendation richness meets the preset richness requirement, sort each film and television information to generate a video recommendation result; on this basis On this basis, this application can generate guiding questions by conducting an in-depth analysis of the video recommendation needs when the target user's video recommendation needs are not clear enough, such as problems such as ambiguity or polysemy; through guiding questions, guide the target user to provide more demand details and key elements to update the video recommendation needs, make the video recommendation needs more targeted, and improve the personalization and pertinence of subsequent video recommendations; when clarification is needed, this application does not need to conduct guided questions and directly enters the subsequent steps, saving unnecessary repeated communication and improving recommendation efficiency; then, the content relevance of the multiple film and television information queried can be measured from the fit and recommendation richness to measure whether it meets the needs, so as to provide rich recommendation content closely related to the needs of the target user, and at the same time, further sort the various film and television contents that meet the fit and recommendation richness, highlight the key film and television information in the video sorting results, and further facilitate the target user to select personalized videos. It can be seen that this application can provide personalized video recommendation results for target users in a targeted manner, improving the user's experience and satisfaction.

[0071] In some embodiments of the present application, it is considered that there may be a situation where the degree of fit meets the fit requirement but the recommendation richness does not meet the richness requirement. In this case, the video recommendation requirement does not need to improve the clarity, but only needs to enrich the recommendation results to avoid affecting the user experience of the target user. Next, the process of enriching the recommendation results will be described in detail, and the steps are as follows:

[0072] S4. When the fit degree meets the preset fit requirement and the recommendation richness does not meet the richness requirement, query the network for film and television information that matches the video recommendation requirement, and return to execute the recommendation richness formed by evaluating each film and television information in step S3.

[0073] Specifically, when the degree of fit meets the preset fit requirement and the richness does not meet the richness requirement, film and television information matching the video recommendation requirement can be queried through the Internet.

[0074] It can be seen from the above technical solution that compared with the previous embodiment, this embodiment newly adds a process of enriching the recommendation results. Through the above method, when the amount of film and television information found in the database is small, the application can further search for more film and television information online, thereby improving the richness of the subsequently generated video recommendation results and improving the user experience of the target users.

[0075] In some embodiments of the present application, the process of step S1, responding to the target user's video recommendation demand and evaluating whether the video recommendation demand needs to be improved in terms of demand clarity, is described in detail, and the steps are as follows:

[0076] S10, segmenting the video recommendation demand to determine whether there is a classification field for the video recommendation demand; if there is, executing step S11; if not, executing step S12.

[0077] Specifically, the video recommendation demand may be segmented and POS tagged to determine whether there is a classification field in the video recommendation demand.

[0078] If the classification field exists, execute step S11; if the classification field does not exist, execute step S12.

[0079] The classification field is used to represent the video type required by the corresponding target user.

[0080] For example, the classification fields can be fields that clearly define the query direction, such as actor ID, director ID, film and television name, comedy, happy variety show, movies for the elderly, and family movies.

[0081] S11. Determine that there is no need to improve the clarity of the video recommendation requirement.

[0082] Specifically, when a classification field exists in a video recommendation demand, it may be determined that there is no need to improve the demand clarity of the video recommendation demand.

[0083] S12: Determine that the video recommendation requirement needs to be improved in terms of clarity.

[0084] Specifically, when there is no classification field in the video recommendation requirement, it may be determined that the requirement clarity of the video recommendation requirement needs to be improved.

[0085] It can be seen from the above technical solution that this embodiment provides an optional method for responding to the target user's video recommendation demand and evaluating whether the video recommendation demand needs to be clarified. The above method can analyze each character in the video recommendation demand to identify whether there is a classification field, thereby evaluating whether the video recommendation demand needs to be clarified.

[0086] In some embodiments of the present application, the process of analyzing the video recommendation requirement in step S2 and generating a guiding sentence matching the video recommendation requirement is described in detail, and the steps are as follows:

[0087] S20. Determine a user portrait for indicating the film and television preferences of the target user at different time periods.

[0088] Specifically, a user profile that characterizes the target user's film and television preferences, film and television interests and hobbies, and film and television viewing habits at different time periods can be constructed based on the target user's film and television clicks at different time periods.

[0089] S21. Determine a demand period corresponding to the video recommendation demand.

[0090] Specifically, the time period corresponding to the video recommendation demand provided by the target user may be determined as the demand time period.

[0091] S22. Generate the guiding sentence based on the required time period, the video recommendation requirement and the user portrait.

[0092] Specifically, natural language processing technology can be used to analyze video recommendation needs, and multiple guiding sentences with progressive relationships can be constructed with reference to the demand period, classification fields of video recommendation needs and user portraits.

[0093] It can be seen from the above technical solution that this embodiment provides an optional method for analyzing video recommendation needs and generating guiding sentences that match the video recommendation needs. Through the above method, guiding sentences can be generated in combination with users' film and television preferences and video recommendation needs in different time periods, and guiding sentences can be customized for target users in different time periods, further improving the matching of guiding sentences with target users, and then ensuring the fit between new video recommendation needs and the actual needs of target users.

[0094] In some embodiments of the present application, the process of generating the guided sentence based on the required time period, the video recommendation requirement and the user portrait in step S22 is described in detail, and the steps are as follows:

[0095] S220. Utilizing a trained decision model, based on the video recommendation demand, using the film and television preferences in the user portrait that match the demand period as the main key element, and using the film and television preferences in the user portrait other than the film and television preferences in the demand period as the secondary key elements, a plurality of guiding sentences with a progressive relationship are generated.

[0096] Specifically, the decision-making model can be combined with the classification field of video recommendation needs as the basis, the film and television preferences in the user portrait that match the demand period as the main key element, and other film and television preferences in the user portrait as secondary key elements to generate multiple logical and coherent guiding sentences.

[0097] It should be noted that the introductory sentences are not repeated.

[0098] For example, after multiple guiding sentences containing only major key elements are rejected by target users, the major key elements may be removed and attempts may be made to generate multiple guiding sentences containing only some minor key elements.

[0099] It can be seen from the above technical developments that this embodiment provides an optional way to customize the generation of guided sentences. Through the above method, the present application can generate guided sentences based on the classification fields and extend them in sequence through the film and television preferences of the user portrait, thereby speeding up the rate of finding the actual preferences of the target user.

[0100] In some embodiments of the present application, the process of querying multiple film and television information based on the video recommendation demand in step S3 is described in detail, and the steps are as follows:

[0101] S30: parsing the video recommendation requirement and extracting keywords of the video recommendation requirement.

[0102] Specifically, natural language technology can be used to parse the video recommendation demand and extract all keywords in the video recommendation demand.

[0103] S31. Based on the keyword, determine the database type corresponding to the video recommendation requirement.

[0104] Specifically, classification fields may be extracted from various keywords, and the database type corresponding to the video recommendation requirement may be determined through the classification fields.

[0105] Different classification fields may correspond to different database types.

[0106] For example, classification fields containing media characters such as actor ID and director ID may correspond to traditional databases.

[0107] Category fields such as Family Movies and Comedy that do not contain media characters may correspond to vector databases.

[0108] The video recommendation requirements can be input into the decision model to obtain the database type predicted by the decision model.

[0109] S32: Query a plurality of film and television information matching the video recommendation requirement from a database matching the database type.

[0110] Specifically, all film and television information matching the video recommendation requirements can be searched from a traditional database or a vector database.

[0111] It can be seen from the above technical solution that this embodiment provides an optional method for querying multiple film and television information based on video recommendation needs. Through the above method, film and television information can be searched from different databases for different classification fields, further improving the reliability and accuracy of film and television information search.

[0112] In some embodiments of the present application, the process of sorting the film and television information and generating the video recommendation result in step S3 is described in detail, and the steps are as follows:

[0113] S33. Extract entities for video recommendation needs.

[0114] Specifically, natural language technology can be used to extract entities for video recommendation requirements.

[0115] S34. Calculate the ranking score of each piece of film and television information based on the extracted entities and the popularity of each piece of film and television information.

[0116] Specifically, the field score can be calculated based on the matching degree between the film and television information and the entity;

[0117] If the film and television information is obtained from a vector database, the vector database score of each film and television information is determined.

[0118] Calculate the popularity score based on the popularity of each film and television information;

[0119] All scores of each film and television information are weighted and fused to form a ranking score for each film and television information.

[0120] For example, each score can be standardized by the Z-score standardization formula, and the weighted sum of the standardized scores is calculated in the ratio of 0.4:0.3:0.3. Finally, the sigmoid activation function is applied to calculate the ranking score of each film and television information.

[0121] The Z-score normalization formula can be shown as follows:

[0122]

[0123] In the formula, μ is the mean of the scores and σ is the standard deviation of the scores.

[0124] The sigmoid activation function can be shown as follows:

[0125]

[0126] Where σ(x) is the output value for input x, and e is a natural constant (approximately 2.718). The sigmoid function maps any real number to the interval (0,1).

[0127] S35. Sort each piece of film and television information based on the ranking score to obtain a video recommendation result.

[0128] Specifically, the film and television information may be sorted in descending order, and the film and television recommendation results may be obtained after sorting.

[0129] It can be seen from the above technical solutions that this embodiment provides an optional method for sorting various film and television information and generating video recommendation results. Through the above method, the recommendation order of the queried film and television information can be adjusted based on the video recommendation needs and the popularity of the film and television information itself, so that the film and television content that better meets user needs, has higher popularity, and is more popular is displayed in the front, further improving the user experience of the target users.

[0130] The application can be executed using a trained multi-level decision model to achieve full automation of video recommendation.

[0131] Next, we will combine Figure 2 The video recommendation device provided in the present application is introduced in detail. The video recommendation device provided below can be compared with the video recommendation method provided above.

[0132] See also Figure 2 It can be found that the video recommendation device may include:

[0133] The evaluation module 10 is used to respond to the target user's video recommendation demand and evaluate whether the video recommendation demand needs to be improved in terms of demand clarity; if so, call the questioning module; if not, call the generation module;

[0134] The questioning module 20 is used to analyze the video recommendation demand and generate a guiding sentence matching the video recommendation demand; initiate a question operation corresponding to the guiding sentence to the target user and obtain a reply text from the target user; based on the reply text, update the video recommendation demand and return to the execution evaluation module;

[0135] The generation module 30 is used to query multiple film and television information based on the video recommendation demand; evaluate the fit of each film and television information with the video recommendation demand and the recommendation richness formed by each film and television information; when the fit meets the preset fit requirement and the recommendation richness meets the preset richness requirement, sort the film and television information to generate a video recommendation result.

[0136] Further, the assessment module may include:

[0137] The classification field determination unit is used to segment the video recommendation demand and determine whether there is a classification field for the video recommendation demand; if so, it is determined that there is no need to improve the clarity of the video recommendation demand; if not, it is determined that there is a need to improve the clarity of the video recommendation demand.

[0138] Furthermore, the questioning module may include:

[0139] A user portrait determination unit, used to determine a user portrait indicating the film and television preferences of the corresponding target user at different time periods;

[0140] A demand period determination unit, used to determine the demand period corresponding to the video recommendation demand;

[0141] The guiding sentence generating unit is used to generate the guiding sentence based on the required time period, the video recommendation requirement and the user portrait.

[0142] Further, the guided sentence generating unit may include:

[0143] The user portrait utilization component is used to utilize the trained decision model, based on the video recommendation demand, with the film and television preferences in the user portrait that match the demand period as the main key element, and the film and television preferences in the user portrait except the film and television preferences in the demand period as the secondary key elements, to generate multiple guiding sentences with a progressive relationship.

[0144] Furthermore, the generation module may include:

[0145] A keyword extraction unit, used to parse the video recommendation requirement and extract keywords of the video recommendation requirement;

[0146] A database type determination unit, used to determine the database type corresponding to the video recommendation requirement based on the keyword;

[0147] The film and television information query unit is used to query a plurality of film and television information matching the video recommendation requirement from a database matching the database type.

[0148] Furthermore, the video recommendation device may further include:

[0149] The online query unit is used to query the film and television information matching the video recommendation requirement online when the fit degree meets the preset fit requirement and the recommendation richness does not meet the richness requirement, and return to call the generation module.

[0150] Furthermore, the generation module may also include:

[0151] An entity extraction unit, used to extract entities for video recommendation requirements;

[0152] A ranking score calculation unit, used to calculate the ranking score of each piece of film and television information based on the extracted entities and the popularity of each piece of film and television information;

[0153] The film and television information sorting unit is used to sort each film and television information based on the sorting score of each film and television information to obtain a video recommendation result.

[0154] The video recommendation device provided in the embodiment of the present application can be applied to video recommendation devices, such as PC terminals, cloud platforms, servers and server clusters. Figure 3 The hardware structure diagram of the video recommendation device is shown in FIG. Figure 3 ,The hardware structure of the video recommendation device may include: at least one processor 1, at least one communication interface 2, at least one memory 3, and at least one communication bus 4;

[0155] In the embodiment of the present application, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 communicate with each other through the communication bus 4;

[0156] The processor 1 may be a central processing unit CPU, or an application-specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention, etc.;

[0157] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), etc., such as at least one disk memory;

[0158] The memory stores a program, and the processor can call the program stored in the memory, wherein the program is used to:

[0159] Respond to the target user's video recommendation needs and evaluate whether the video recommendation needs need to be clarified;

[0160] If yes, the video recommendation requirement is analyzed to generate a guiding sentence matching the video recommendation requirement; a question operation corresponding to the guiding sentence is initiated to the target user to obtain a reply text from the target user; based on the reply text, the video recommendation requirement is updated, and the step of evaluating whether the video recommendation requirement needs to be improved in terms of requirement clarity is returned to be executed;

[0161] If not, based on the video recommendation needs, multiple film and television information are queried; the degree of fit between each piece of film and television information and the video recommendation needs and the recommendation richness formed by each piece of film and television information are evaluated; when the fit meets the preset fit requirements and the recommendation richness meets the preset richness requirements, the each piece of film and television information is sorted to generate a video recommendation result.

[0162] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0163] The embodiment of the present application further provides a readable storage medium, which may store a program suitable for execution by a processor, wherein the program is used to:

[0164] Respond to the target user's video recommendation needs and evaluate whether the video recommendation needs need to be clarified;

[0165] If yes, the video recommendation requirement is analyzed to generate a guiding sentence matching the video recommendation requirement; a question operation corresponding to the guiding sentence is initiated to the target user to obtain a reply text from the target user; based on the reply text, the video recommendation requirement is updated, and the step of evaluating whether the video recommendation requirement needs to be improved in terms of requirement clarity is returned to be executed;

[0166] If not, based on the video recommendation needs, multiple film and television information are queried; the degree of fit between each piece of film and television information and the video recommendation needs and the recommendation richness formed by each piece of film and television information are evaluated; when the fit meets the preset fit requirements and the recommendation richness meets the preset richness requirements, the each piece of film and television information is sorted to generate a video recommendation result.

[0167] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0168] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0169] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0170] The above description of the disclosed embodiments enables professionals and technicians in the field to implement or use the present application. Various modifications to these embodiments will be apparent to professionals and technicians in the field, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. The various embodiments of the present application can be combined with each other. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A video recommendation method, characterized in that: include: Respond to the target user's video recommendation needs and evaluate whether the video recommendation needs need to be clarified; If yes, the video recommendation requirement is analyzed to generate a guiding sentence matching the video recommendation requirement; a question operation corresponding to the guiding sentence is initiated to the target user to obtain a reply text of the target user; Based on the reply text, the video recommendation requirement is updated, and the step of returning to execute the step of evaluating whether the requirement clarity of the video recommendation requirement needs to be improved; If not, based on the video recommendation requirement, query multiple film and television information; evaluate the degree of fit between each film and television information and the video recommendation requirement and the recommendation richness formed by each film and television information; When the fit degree meets the preset fit requirements and the recommendation richness meets the preset richness requirements, the various film and television information are sorted to generate video recommendation results.

2. The video recommendation method according to claim 1, characterized in that: The evaluating whether it is necessary to improve the clarity of the video recommendation requirement includes: Segmenting the video recommendation requirement to determine whether there is a classification field for the video recommendation requirement; If so, it is determined that there is no need to improve the clarity of the video recommendation requirement; If not, it is determined that the video recommendation requirement needs to be improved in terms of clarity.

3. The video recommendation method according to claim 1, characterized in that: The step of analyzing the video recommendation requirement and generating a guiding statement matching the video recommendation requirement includes: Determine user profiles for indicating the film and television preferences of target users at different time periods; Determine a demand period corresponding to the video recommendation demand; The guiding sentence is generated based on the demand time period, the video recommendation demand and the user portrait.

4. The video recommendation method according to claim 3, characterized in that: The generating the guiding statement based on the required time period, the video recommendation requirement and the user portrait includes: Utilizing the trained decision model, based on the video recommendation demand, taking the film and television preferences in the user portrait that match the demand period as the main key element, and taking the film and television preferences in the user portrait except the film and television preferences in the demand period as the secondary key elements, a plurality of guiding sentences with a progressive relationship are generated.

5. The video recommendation method according to claim 1, characterized in that: The querying of multiple film and television information based on the video recommendation requirement includes: Parsing the video recommendation requirement and extracting keywords of the video recommendation requirement; Based on the keyword, determining the database type corresponding to the video recommendation requirement; A plurality of film and television information matching the video recommendation requirement is queried from a database matching the database type.

6. The video recommendation method according to claim 5, characterized in that: Also includes: When the fit degree meets the preset fit requirement and the recommendation richness does not meet the richness requirement, the network is queried for film and television information that matches the video recommendation requirement, and the step of evaluating the recommendation richness formed by each film and television information is returned to be executed.

7. The video recommendation method according to claim 1, characterized in that: The step of sorting the film and television information to generate video recommendation results includes: Perform entity extraction for video recommendation needs; Based on the extracted entities and the popularity of each film and television information, calculate the ranking score of each film and television information; Based on the ranking score of each piece of film and television information, the film and television information is sorted to obtain the video recommendation result.

8. A video recommendation device, characterized in that: include: An evaluation module, used to respond to the target user's video recommendation demand and evaluate whether the video recommendation demand needs to be improved in terms of demand clarity; If so, the question module is called; If not, the generation module is called; A questioning module is used to analyze the video recommendation demand and generate a guiding sentence matching the video recommendation demand; initiate a question operation corresponding to the guiding sentence to the target user and obtain a reply text from the target user; Based on the reply text, update the video recommendation requirement and return to the execution evaluation module; A generation module is used to query multiple film and television information based on video recommendation requirements; evaluate the degree of fit between each film and television information and the video recommendation requirements and the recommendation richness formed by each film and television information; When the fit degree meets the preset fit requirements and the recommendation richness meets the preset richness requirements, the various film and television information are sorted to generate video recommendation results.

9. A video recommendation device, characterized in that: including memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the video recommendation method according to any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the video recommendation method according to any one of claims 1 to 7 is implemented.

Images