Method and device for matching and aligning media resource programs based on vector database, computer equipment and readable storage medium

Abstract

The invention discloses a method and a device for matching and aligning media resource programs based on a vector database, computer equipment and a readable storage medium. The method comprises the following steps: firstly, acquiring related information of existing media resource programs and media resource programs to be matched in the media resource database and vectorizing; and then according to the similarity of the existing related vectors of the media assets to be matched, a matching alignment result is obtained by combining a full-text detection strategy. And if matching succeeds, determining a unique program identifier and writing the unique program identifier into a mapping file. By means of the design, the defects in the prior art are overcome, and the accuracy and efficiency of media asset program matching are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of media asset program alignment, and in particular to a method, device, computer equipment and readable storage medium for performing media asset program matching alignment based on a vector database. Background Art

[0002] In the film and television industry, due to the differences in naming standards of film and television program names from different sources, existing media resource program matching methods such as regular matching combined with approximate algorithms have many limitations. They cannot efficiently match the correct program name, cannot handle semantically consistent or reversed program names, and have low matching efficiency when the amount of data is large. Summary of the invention

[0003] The object of the present invention is to provide a method, device, computer equipment and readable storage medium for matching and aligning media resources and programs based on a vector database.

[0004] In a first aspect, an embodiment of the present invention provides a method for matching and aligning media assets and programs based on a vector database, including:

[0005] Acquire existing media asset related information of multiple existing media asset programs in the media asset library, and vectorize the existing media asset related information to obtain existing media asset related vectors corresponding to each of the multiple existing media asset programs;

[0006] Acquire the to-be-matched media asset related information of the to-be-matched media asset program, and vectorize the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program;

[0007] According to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, combined with the full-text detection strategy, a matching alignment result corresponding to the media asset related vector to be matched is obtained;

[0008] On the basis that the matching alignment result indicates a successful match, a unique identifier of the program is determined and written into a mapping file.

[0009] In a possible implementation, the step of obtaining existing media asset related information of a plurality of existing media asset programs in a media asset library, and vectorizing the existing media asset related information to obtain existing media asset related vectors corresponding to each of the plurality of existing media asset programs includes:

[0010] Acquire existing media asset related information of multiple existing media asset programs in the media asset library, wherein the existing media asset related information includes director information, actor information, program name, and program unique identifier;

[0011] Convert the program name into a word vector to generate an existing vector text;

[0012] Encapsulating the director information, actor information, program name, and program unique identifier to obtain vector metadata;

[0013] Using the program unique identifier as a vector ID;

[0014] The existing vector text, the vector metadata and the vector ID are used as the existing media asset related vector, and the existing media asset related vector is written into a vector database.

[0015] In a possible implementation manner, the obtaining of the to-be-matched media asset related information of the to-be-matched media asset program and vectorizing the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program includes:

[0016] Obtaining the relevant information of the to-be-matched media asset program, wherein the relevant information of the to-be-matched media asset program includes director information, actor information, program name, and program unique identifier;

[0017] The to-be-matched media asset related information is converted into a word vector to obtain a vector text as the to-be-matched media asset related vector of the to-be-matched media asset program.

[0018] In a possible implementation, according to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, combined with the full-text detection strategy, a matching alignment result corresponding to the media asset related vector to be matched is obtained, including:

[0019] Performing traversal similarity comparison on the vector text in the vector database to obtain the vector similarity between the vector text and each of the existing media asset related vectors;

[0020] If there is a pending existing media asset related vector whose vector similarity is higher than a preset vector similarity threshold, extracting pending director information and pending actor information corresponding to the pending existing media asset related vector;

[0021] In the case where the pending director information and the pending actor information have an intersection with the director information and the actor information included in the to-be-matched media asset related information, the pending existing media asset program corresponding to the pending existing media asset related vector is used as a successful matching alignment result of the to-be-matched media asset program;

[0022] If there is no existing media asset related vector to be determined whose vector similarity is higher than the preset vector similarity threshold, a full-text search strategy is used to obtain a matching alignment result corresponding to the media asset related vector to be matched.

[0023] In a possible implementation manner, the using of a full-text search strategy to obtain a matching alignment result corresponding to the to-be-matched media asset related vector includes:

[0024] Performing a full-text search on the vector database using the full-text search strategy to obtain pending existing media resource programs whose program names reach a preset program name text similarity threshold with the media resource program to be matched;

[0025] Extracting the pending director information and the pending actor information corresponding to the pending existing media asset related vector;

[0026] When the pending director information and the pending actor information have an intersection with the director information and the actor information included in the to-be-matched media asset related information, the to-be-determined existing media asset program corresponding to the to-be-determined existing media asset related vector is taken as a successful matching alignment result of the to-be-matched media asset program.

[0027] In a possible implementation manner, determining the program unique identifier and writing it into a mapping file based on the matching alignment result indicating successful matching includes:

[0028] On the basis that the matching alignment result indicates a successful match, obtaining at least one to-be-matched media resource program corresponding to the successful matching alignment result;

[0029] Extracting the program unique identifiers corresponding to the acquired media asset program to be matched and the at least one media asset program to be matched;

[0030] The mapping file is written according to the respective corresponding program unique identifiers.

[0031] In a possible implementation, the method further includes:

[0032] In response to a media asset program search instruction, obtaining a search program unique identifier included in the media asset program search instruction;

[0033] The searched media asset program corresponding to the searched media asset program instruction is determined from the mapping file according to the searched program unique identifier, and the searched media asset program unique identifier is associated with the searched program unique identifier and written into the mapping file.

[0034] In a second aspect, an embodiment of the present invention provides a device for matching and aligning media assets and programs based on a vector database, including:

[0035] An acquisition module is used to acquire existing media asset related information of multiple existing media asset programs in the media asset library, and vectorize the existing media asset related information to obtain existing media asset related vectors corresponding to each of the multiple existing media asset programs; acquire the to-be-matched media asset related information of the to-be-matched media asset program, and vectorize the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program;

[0036] The alignment module is used to obtain the matching alignment result corresponding to the media asset related vector to be matched according to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, combined with the full-text detection strategy; based on the matching alignment result indicating a successful match, determine the program unique identifier and write it into a mapping file.

[0037] In a third aspect, an embodiment of the present invention provides a computer device, comprising a processor and a non-volatile memory storing computer instructions, wherein when the computer instructions are executed by the processor, the computer device executes the method described in at least one possible implementation of the first aspect.

[0038] In a fourth aspect, an embodiment of the present invention provides a readable storage medium, wherein the readable storage medium includes a computer program, and when the computer program is run, the computer device where the readable storage medium is located is controlled to execute the method described in at least one possible implementation manner of the first aspect.

[0039] Compared with the prior art, the beneficial effects provided by the present invention include: using a method, device, computer equipment and readable storage medium disclosed by the present invention to match and align media programs based on a vector database, by obtaining and vectorizing the relevant information of existing media programs and to-be-matched media programs in the media library. Then, based on the similarity of the relevant vectors of the existing and to-be-matched media, the matching alignment result is obtained in combination with the full-text detection strategy. If the match is successful, the unique identifier of the program is determined and written into the mapping file. Such a design solves the shortcomings of the prior art and improves the accuracy and efficiency of media program matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can also be obtained based on these drawings without creative work.

[0041] Figure 1 A schematic diagram of the steps of a method for matching and aligning media resources and programs based on a vector database provided in an embodiment of the present invention;

[0042] Figure 2 A schematic block diagram of the structure of an apparatus for media resource program matching and alignment based on a vector database provided in an embodiment of the present invention;

[0043] Figure 3 A schematic block diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0045] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings.

[0046] In order to solve the technical problems in the aforementioned background technology, Figure 1 This is a flow chart of a method for matching and aligning media assets and programs based on a vector database provided in an embodiment of the present disclosure. The method for matching and aligning media assets and programs based on a vector database is introduced in detail below.

[0047] Step S201, obtaining existing media asset related information of multiple existing media asset programs in a media asset library, and vectorizing the existing media asset related information to obtain existing media asset related vectors corresponding to each of the multiple existing media asset programs;

[0048] Step S202, obtaining the to-be-matched media asset related information of the to-be-matched media asset program, and vectorizing the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program;

[0049] Step S203, obtaining a matching alignment result corresponding to the media asset related vector to be matched according to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, in combination with a full-text detection strategy;

[0050] Step S204, based on the matching alignment result indicating successful matching, determine the program unique identifier and write it into a mapping file.

[0051] In the embodiment of the present invention, for example, it is assumed that a large amount of film and television program information is stored in the media resource library to which the server is connected. First, the server starts to obtain relevant information of multiple existing media resource programs in the media resource library.

[0052] For example, the server obtains the following existing media resources:

[0053] 1. Program A: The director is Zhang San, the actors are Li Si and Wang Wu, the program name is "Magical Adventure Journey", and the program unique identifier is 001.

[0054] 2. Program B: The director is Zhao Liu, the actors are Sun Qi and Zhou Ba, the program name is "Dreamlike Interstellar Travel", and the program's unique identifier is 002.

[0055] 3. Program C: The director is Qian Jiu, the actors are Wu Shi and Zheng Shiyi, the program name is "Legend of Heroes XX", and the program's unique identifier is 003.

[0056] The server converts the program names into word vectors. Taking the name of program A, "Amazing Adventures", as an example, the word vector conversion technology converts it into a set of vector representations of specific dimensions. This vector can capture the semantic and text features in the program name.

[0057] The server encapsulates the director information, actor information, program name and program unique identifier to obtain vector metadata. For example, for program A, the encapsulated vector metadata may contain the relevant features of director Zhang San, the relevant features of actors Li Si and Wang Wu, and the vector features converted from the program name, and the program unique identifier 001 is used as the vector ID.

[0058] Finally, the server takes these existing vector texts, vector metadata and vector IDs as existing media asset related vectors and writes them into the vector database. In this way, the vector database stores vector representations of multiple existing media asset programs, preparing for subsequent matching work.

[0059] Assume that there is a new media resource program to be matched and needs to be matched and aligned. The server obtains the relevant information of the media resource program to be matched.

[0060] For example, program D to be matched: the director is Liu Twelve, the actors are Chen Shisan and Ma Shisi, the program name is "Fantasy Adventure Journey", and the program unique identifier is 004.

[0061] The server performs word vector conversion on the program name of the to-be-matched program D, and converts it into a vector that can represent its semantic and text features. This vector is the relevant vector of the to-be-matched media asset.

[0062] The server compares the similarity between the to-be-matched media asset related vector of the to-be-matched program D and the existing media asset related vectors stored in the vector database.

[0063] Assume that, through calculation, the similarity between the existing media asset related vector of program A and the vector of program D to be matched is high, which is higher than the preset vector similarity threshold. The server extracts the director information (Zhang San) and actor information (Li Si, Wang Wu) corresponding to program A.

[0064] Then, the server compares the director information (Liu Twelve) and the actor information (Chen Thirteen, Ma Fourteen) of the program to be matched D. It is found that there is no intersection between the two, and it cannot be directly determined that the match is successful.

[0065] Since no fully matching existing media resource program is found in the first round of vector similarity comparison, the server initiates the full-text detection strategy.

[0066] The server performs a full-text search on the program names in the media library. Assume that in the full-text search, a program E is found, whose name is "Amazing Adventure Journey", directed by Liu Twelve, and actors include Chen Shisan and Ma Shisi. The server extracts the director and actor information of program E and finds that the director and actor information of the program D to be matched are exactly the same. Therefore, program E is regarded as the successful matching alignment result of program D to be matched.

[0067] Since the to-be-matched program D successfully matches program E, the server obtains the unique program identifier of program E. Assume that the unique identifier of program E is 005.

[0068] The server writes the program unique identifier 004 of the to-be-matched program D and the program unique identifier 005 of the program E into the mapping file. In this way, in subsequent searches and uses, the matching media asset program can be quickly found through the mapping file.

[0069] Assume another program F to be matched: the director is Wang Shiwu, the actors are Lin Shiliu and Xu Shiqi, the program name is "The Eternal Love Legend", and the program unique identifier is 006.

[0070] The server follows the above steps:

[0071] 1. Compare the similarity with the existing media-related vectors in the vector database, and find no existing media programs with similarity higher than the threshold and overlapping director and actor information.

[0072] 2. The full-text detection strategy was initiated, but no matching programs were found.

[0073] In this case, there is no successful match result for the to-be-matched program F, and the server will not write its program unique identifier into the mapping file.

[0074] For example, there is a program G to be matched: the director is Zhao Shiba, the actors are Sun Shijiu and Zhou Ershi, the program name is "Baahubali: The Beginning", and the program unique identifier is 007.

[0075] 1. In the vector similarity comparison, program H is found. Its program name is "Baahubali: The Beginning", and its director and actor information overlap with the program G to be matched.

[0076] 2. It is determined that the match is successful, and the program unique identifier of program H is obtained, assuming it is 008, and 007 and 008 are written into the mapping file.

[0077] Assuming that there are a large number of media asset programs to be matched, the server executes the above matching process for each media asset program to be matched in turn, and continuously improves and updates the mapping file to ensure that the media asset program matching and alignment service can be accurately provided to the user.

[0078] For example, in one day, the server received 1,000 media resource programs to be matched. Through efficient calculation and matching processes, it successfully found matching existing media resource programs for 800 programs and wrote the corresponding program unique identifiers into the mapping file, providing convenient and accurate support for users' media resource retrieval and use.

[0079] In actual applications, the server may also need to handle various abnormal situations, such as data updates in the media library, incomplete information acquisition due to network delays, etc. However, through optimized algorithms and stable system architecture, the server can effectively cope with these challenges and ensure the smooth progress of media asset program matching and alignment.

[0080] Assume that the system where the server is located needs to process media matching requests from multiple users at the same time. Each user may submit multiple media programs to be matched at the same time. The server processes these requests in parallel through multi-threading or distributed processing, which greatly improves processing efficiency and reduces user waiting time.

[0081] For example, user A submits 5 media programs to be matched, and user B submits 3 media programs to be matched. The server matches these 8 programs at the same time, performs vector similarity calculation and full-text detection independently without interfering with each other, and quickly provides matching results to users.

[0082] Assume that the network environment where the server is located has certain fluctuations, resulting in a short delay or interruption when obtaining media-related information or performing vector calculations. The server has the functions of automatic reconnection and breakpoint resumption, and can continue unfinished work after the network returns to normal, ensuring that the matching work of each media program to be matched can be completed smoothly without data loss or matching errors.

[0083] Assuming that over time, the program information in the media library is constantly updated, new programs are added, and old program information is changed. The server regularly synchronizes and updates the vector database to ensure that the existing media-related vectors can accurately reflect the latest status of the media library, thereby ensuring the accuracy and effectiveness of the matching.

[0084] For example, a comprehensive scan of the media resource library is performed once a week to update the relevant vectors in the vector database. If there is an urgent important program update within a week, a local vector update operation can also be triggered immediately to meet the real-time requirements.

[0085] Assuming that the media programs to be matched received by the server come from different regions and languages, the program names may be expressed in multiple languages. The server can accurately convert and match program names in different languages ​​through the built-in multi-language processing module, which broadens the coverage and applicable scenarios of the service.

[0086] For example, the name of the program K to be matched is "L'AventureMagique" in French. The server can accurately match it with the corresponding existing French media program, providing high-quality services for users who use different languages.

[0087] Assume that the server needs to interact and share data with other systems or platforms to obtain more comprehensive media information or provide matching results for other systems. The server has a complete interface and data transmission protocol, which can communicate with external systems safely and efficiently to achieve seamless data connection and collaborative work.

[0088] For example, we can cooperate with third-party film and television data providers to obtain their exclusive media information, and share the matching results with cooperative video playback platforms to improve the service quality and efficiency of the entire industry chain.

[0089] Assume that the server needs to adjust the matching strategy and parameters according to different user needs and usage scenarios. For example, for some users who have extremely high requirements for matching accuracy, the server can increase the vector similarity threshold and the strictness of full-text detection; for users who pay more attention to matching speed, the server can appropriately relax some conditions to improve the processing speed while ensuring a certain degree of accuracy.

[0090] Through the various detailed scenario examples above, we can see that the application and processing methods of the media asset program matching and alignment method based on the vector database in different situations can effectively improve the accuracy and efficiency of media asset program matching and provide users with a better service experience.

[0091] In the embodiment of the present invention, the aforementioned step S201 can be implemented through the following example execution.

[0092] Acquire existing media asset related information of multiple existing media asset programs in the media asset library, wherein the existing media asset related information includes director information, actor information, program name, and program unique identifier;

[0093] Convert the program name into a word vector to generate an existing vector text;

[0094] Encapsulating the director information, actor information, program name, and program unique identifier to obtain vector metadata;

[0095] Using the program unique identifier as a vector ID;

[0096] The existing vector text, the vector metadata and the vector ID are used as the existing media asset related vector, and the existing media asset related vector is written into a vector database.

[0097] In the embodiment of the present invention, illustratively, the server is connected to a huge media resource library and is ready to obtain relevant information of a plurality of existing media resource programs. First, the server obtains a series of program information from the media resource library.

[0098] For example, the following existing media resources programs were obtained:

[0099] Program A: The director is the well-known director Zhang XX, the actors include Gong X, Chen XX, etc., the program name is "Red XX", and the program's unique identifier is 001.

[0100] Program B: The director is Feng XX, the actors are Ge X and Xu X, the program name is "A XX", and the program unique identifier is 002.

[0101] Program C: The director is Li X, the actors are Liang XX and Tang X, the program name is "X Ring", and the program unique identifier is 003.

[0102] For each program, the server starts to convert the program name into a word vector. Taking "Red XX" of program A as an example, through advanced natural language processing technology and word vector model, the name "Red XX" is converted into a vector representation of a specific dimension. This vector can capture the semantic and text features of the word "Red XX" and generate existing vector text.

[0103] Next, the server encapsulates the director information, actor information, program name, and program unique identifier to form vector metadata. For program A, the vector metadata contains the relevant feature information of Zhang XX, Gong X, Chen XX, and the features of the name "Red XX", forming a rich descriptive data set.

[0104] Then, the server uses the program unique identifier as the vector ID. For program A, its vector ID is 001.

[0105] Finally, the server integrates the generated existing vector text, the encapsulated vector metadata, and the determined vector ID as the existing media asset related vector. For example, the existing media asset related vector of program A includes the existing vector text of "Red XX", the vector metadata containing information such as the director and actors, and the vector ID 001.

[0106] The server writes these existing media asset related vectors into the vector database one by one for storage. In this way, the vector database has detailed vector representations of multiple programs such as program A, B, and C.

[0107] For another example, some new existing media programs are obtained:

[0108] Program D: The director is Chen XX, the actors are Zhang XX and Zhang XXX, the program name is "Overbearing XX", and the unique identifier of the program is 004.

[0109] Program E: The director is Wu XX, the actors are Zhou XX and Di XX, the program name is "Hero XX", and the unique identifier of the program is 005.

[0110] The server still processes these new programs according to the same process. Perform word vector transformation on "Overbearing XX" of Program D and "Hero XX" of Program E to generate the corresponding existing vector texts. Then encapsulate the director, actors, program name, and unique identifier as vector metadata, use 004 and 005 as vector IDs respectively, and finally form existing media-related vectors and write them into the vector database.

[0111] As time goes by, the server continuously obtains new existing media programs and repeats the above processing process, making the data in the vector database more and more rich and perfect.

[0112] Suppose during a busy period of film and television data processing, the server needs to process a large amount of existing media program information in a short time. For example, within one hour, 100 newly imported existing media programs need to be processed. With its powerful computing power and efficient algorithms, the server methodically performs operations such as name word vector transformation, data encapsulation, determining vector IDs, and writing into the vector database for each program.

[0113] During the processing, some special situations may be encountered. For example, there are rare words or specific industry terms in some program names. The server can accurately perform word vector transformation through its pre-trained rich word library and intelligent language model.

[0114] Suppose some programs have a large number of actor information. The server can also comprehensively and accurately incorporate this actor information into the vector metadata to ensure the integrity and accuracy of the data.

[0115] After continuous processing, the vector database gradually becomes an important resource library containing a large amount of accurate existing media-related vectors, providing a solid data foundation for subsequent media program matching and alignment work.

[0116] Suppose the system where the server is located has an upgrade, and a more advanced word vector model and data encapsulation algorithm are newly introduced. When processing new existing media programs, the server can utilize these new technologies to generate more accurate and expressive existing vector texts and vector metadata, further improving the quality and usability of the vector database.

[0117] For example, for a program F with a complex name and many actors: the director is Jia XX, the actors include Zhao X, Wang XX and many others, the program name is "Mountains and Rivers XX", and the program unique identifier is 006. The new technology can more accurately convert the poetic and connotative name "Mountains and Rivers XX" into existing vector text, and more effectively encapsulate the complex information of many actors into vector metadata, generate high-quality existing media asset-related vectors for program F and write them into the vector database.

[0118] Through the above detailed scenario examples, we can clearly see how the server can efficiently and accurately complete various tasks in the process of acquiring, processing and storing existing media asset-related vectors, laying a solid foundation for media asset program matching and alignment.

[0119] In the embodiment of the present invention, the aforementioned step S202 may be implemented through the following example execution.

[0120] Obtaining the relevant information of the to-be-matched media asset program, wherein the relevant information of the to-be-matched media asset program includes director information, actor information, program name, and program unique identifier;

[0121] The to-be-matched media asset related information is converted into a word vector to obtain a vector text as the to-be-matched media asset related vector of the to-be-matched media asset program.

[0122] In the embodiment of the present invention, the server is illustratively ready to receive relevant information of the to-be-matched media asset program at any time. For example, at a certain moment, a to-be-matched media asset program is received.

[0123] The media resource program to be matched contains the following information: the director is A, the actors are B and C, the program name is "Xinhua XX", and the program unique identifier is 007.

[0124] The server first obtains the detailed information about the media assets to be matched, including the description of director A, the feature information of actors B and C, and the program name "Xinhua XX" and the unique identifier 007.

[0125] Next, the server begins to convert the information related to the media assets to be matched into word vectors. For the program name "Xinhua XX", the server uses advanced natural language processing technology and word vector models to convert it into a vector of a specific dimension. This vector can capture the semantic and textual characteristics of the word "Xinhua XX".

[0126] During the conversion process, the server considers the context, semantic relationships, and common language patterns of the words. For example, the relaxed and joyful emotions contained in the word "Xinhua XX" and its possible associations and differences with other similar movie titles are all incorporated into the generated vector.

[0127] After a series of complex calculations and processes, the server successfully converts "Heart Flower XX" into a vector text that can accurately reflect its semantics and text features.

[0128] This vector text becomes the vector related to the media asset to be matched for the media asset program to be matched.

[0129] For another example, another media asset program to be matched is received, and its relevant information is as follows: the director is D, the actors are E and F, the program name is "Fantasy XX Treasure Box", and the unique program identifier is 008.

[0130] The server also follows the above steps to obtain all the relevant information of this media asset program to be matched. Then, it performs word vector conversion for the program name "Fantasy XX".

[0131] The server will carefully analyze this relatively long name with specific cultural connotations. It not only needs to consider the overall concept of "Fantasy XX", but also pay attention to the role and significance of the specific element "XX Treasure Box" in the whole name.

[0132] Through in-depth semantic understanding and text analysis, the server generates a vector text that can accurately represent "Fantasy XX" as the vector related to the media asset to be matched for this media asset program to be matched.

[0133] Suppose during a period when popular影视作品emerge, the server receives a large number of media asset programs to be matched in a short period of time.

[0134] For example, 50 media asset programs to be matched are received in one day. These programs cover various types and styles, and the names are also diverse.

[0135] Some program names are simple and clear, such as "War X"; some are more complex and creative, such as "Super Space-Time XX".

[0136] For each media asset program to be matched, the server quickly and accurately obtains its relevant information and performs word vector conversion.

[0137] When processing a simple and straightforward program name like "War X", the server can quickly generate a vector text with distinct features.

[0138] For a complex name like "Super Space-Time XX", the server will fully explore the semantic levels and text structures therein to ensure that the generated vector text can comprehensively and accurately reflect its unique connotations.

[0139] Even when faced with program names that contain special vocabulary or industry terms, the server can successfully complete the word vector conversion and generate high-quality media-related vectors to be matched with its powerful vocabulary and intelligent learning capabilities.

[0140] Assume that the system where the server is located needs to process media resource programs to be matched from different channels at the same time.

[0141] For example, it receives both new programs from online video platforms and upcoming programs from TV stations.

[0142] For the media programs to be matched from online video platforms, the relevant information may focus more on popular Internet elements and the preferences of young audiences.

[0143] The media programs to be matched from TV stations may more likely reflect the traditional film and television production style and the preferences of mainstream audiences.

[0144] When processing these to-be-matched media asset programs from different sources, the server can flexibly adapt to various characteristics and requirements, and generate accurate and targeted to-be-matched media asset related vectors for each program.

[0145] Assume that the server receives a media resource program with international influence to be matched, for example, the program name is "Inception", the director is AA, the actors are BB, and so on.

[0146] When processing the English program name, the server can also use its multilingual processing capabilities to perform accurate word vector conversion.

[0147] It takes into account the multiple meanings and uses of the word "Inception" in English contexts, as well as semantic associations with the film's themes and plot.

[0148] Finally, a media asset-related vector to be matched is generated that can effectively reflect the characteristics of the program, preparing for subsequent matching and alignment work.

[0149] The above detailed scenario examples illustrate how the server can efficiently and accurately obtain relevant information about the media asset program to be matched and convert it into relevant vectors of the media asset to be matched, providing a key pre-processing step for achieving media asset program matching alignment.

[0150] In the embodiment of the present invention, the aforementioned step S203 can be implemented through the following example execution.

[0151] Performing traversal similarity comparison on the vector text in the vector database to obtain the vector similarity between the vector text and each of the existing media asset related vectors;

[0152] If there is a pending existing media asset related vector whose vector similarity is higher than a preset vector similarity threshold, extracting pending director information and pending actor information corresponding to the pending existing media asset related vector;

[0153] In the case where the pending director information and the pending actor information have an intersection with the director information and the actor information included in the to-be-matched media asset related information, the pending existing media asset program corresponding to the pending existing media asset related vector is used as a successful matching alignment result of the to-be-matched media asset program;

[0154] If there is no existing media asset related vector to be determined whose vector similarity is higher than the preset vector similarity threshold, a full-text search strategy is used to obtain a matching alignment result corresponding to the media asset related vector to be matched.

[0155] In the embodiment of the present invention, the server obtains a vector text of a media resource program to be matched, and prepares to perform a similarity comparison in an existing vector database. Assume that the vector text of the media resource program to be matched is about a film and television work called "Xingchen XX", directed by Director Wang, and actors Li and Zhang.

[0156] The server starts to traverse the vector text in the vector database for similarity comparison. The vector database stores a large number of existing media asset related vectors, such as:

[0157] Existing media program 1: The director is Director Chen, the actors are Actor Zhao and Actor Liu, and the vector similarity is 0.3.

[0158] Existing media program 2: The director is Director Lin, the actors are actor Sun and actor Zhou, and the vector similarity is 0.2.

[0159] Existing media program 3: The director is Director Wang, the actors are Actor Li and Actor Chen, and the vector similarity is 0.7.

[0160] After comparison and calculation, the server obtains the vector similarity between the vector text and each existing media asset related vector. In this example, the vector similarity of the existing media asset program 3 is 0.7, which is higher than the preset vector similarity threshold of 0.6.

[0161] Since there is a pending existing media asset related vector (ie, existing media asset program 3) whose vector similarity is higher than the preset threshold, the server extracts the corresponding pending director information (Director Wang) and pending actor information (actor Li and actor Chen).

[0162] Then, the server compares the pending director information and the pending actor information with the director information (director Wang) and actor information (actor Li and actor Zhang) of the media resource program to be matched, and finds that the director information is the same and the actor information has an intersection (actor Li is included in both).

[0163] In this case, the server uses the existing media asset program 3 as the successful matching alignment result of the media asset program to be matched.

[0164] Assume that there is another media resource program to be matched, whose vector text is about a work called "Dream Adventure", directed by Director Zhao, and starring actors Wu and Zheng.

[0165] After traversing and comparing the similarities in the vector database, the server finds that the similarities of all existing media asset-related vectors are lower than a preset vector similarity threshold.

[0166] At this point, the server starts the full-text search strategy, which performs a comprehensive text search on the program name, director information, actor information, etc. in the entire media resource library.

[0167] During the full-text search, the server found a program called "Fantasy Adventure", directed by Zhao, and starring Wu and Ma.

[0168] The server extracts the director information and actor information of the pending existing media resource program and compares them with the director information and actor information of the media resource program to be matched. It is found that the director information is the same and the actor information overlaps (both have actor Wu).

[0169] Therefore, the server uses the pending existing media asset program as a successful matching alignment result of the media asset program to be matched.

[0170] Assume that in a period when film and television resources are abundant, the server has to process a large number of matching requests every day. For example, 100 media resource programs to be matched are received in one day.

[0171] For the first media resource program to be matched, "Mysterious XX", the director is Director Liu, and the actors are Actor Wang and Actor Chen. The server traverses and compares the vector database and finds that there are 5 existing media resource related vectors with high similarity.

[0172] Among them, the vector similarity of an existing media resource program 4 reaches 0.8. The server extracts its director information and actor information, and after comparison, finds that there is an intersection with the media resource program to be matched, thus determining it as a successful matching alignment result.

[0173] For the second media resource program to be matched, "Legend XX", the director is Director Yang, and the actors are Actor Zhou and Actor Zheng. The server does not find any existing media resource related vectors with a similarity higher than the threshold in the vector database.

[0174] Therefore, the server starts the full-text search strategy and searches in a wider range of media resources. Finally, a similar program "Ancient XX" is found, whose director and actor information overlaps with the media resource program to be matched, and it is determined to be a successful match alignment result.

[0175] Over time, the server continuously optimizes its matching algorithms and full-text retrieval strategies to improve matching accuracy and efficiency.

[0176] For example, for some programs with similar names but different contents, the server can analyze the director and actor information more carefully to avoid mismatching.

[0177] For example, when the amount of data in the media resource library continues to increase, the server can quickly adjust the retrieval strategy to ensure that the matching work is completed within a reasonable time.

[0178] Assume that the server receives a foreign media program to be matched, the name of which is "Star Adventure", the director is a foreign director John, and the actors are Tom and Jerry.

[0179] The server first compares its vector text with the existing media-related vectors in the vector database. Due to language and cultural differences, it may not find vectors with similarity higher than the threshold at first.

[0180] However, through the full-text search strategy, the server can take into account the translation of the name and the expression in different languages, and finally find the existing media resources program that matches it.

[0181] Through the above detailed scenario examples, the server can accurately and efficiently find suitable matching alignment results for the media resource programs to be matched based on vector similarity and full-text detection strategies.

[0182] In the embodiment of the present invention, the above-mentioned use of the full-text search strategy to obtain the matching alignment result corresponding to the to-be-matched media asset related vector can be implemented through the following examples.

[0183] Performing a full-text search on the vector database using the full-text search strategy to obtain pending existing media resource programs whose program names reach a preset program name text similarity threshold with the media resource program to be matched;

[0184] Extracting the pending director information and the pending actor information corresponding to the pending existing media asset related vector;

[0185] When the pending director information and the pending actor information have an intersection with the director information and the actor information included in the to-be-matched media asset related information, the to-be-determined existing media asset program corresponding to the to-be-determined existing media asset related vector is taken as a successful matching alignment result of the to-be-matched media asset program.

[0186] In the embodiment of the present invention, illustratively, the server receives relevant information of a to-be-matched media resource program, for example, the to-be-matched media resource program is named "Legend of the Magic Journey", the director is Director Zhang, and the actors are actor Liu and actor Chen.

[0187] The server first uses the full-text search strategy to perform a full-text search on the vector database. In this process, the server will perform a detailed analysis and comparison of all the program names in the database.

[0188] Assume that the following program names are stored in the vector database:

[0189] "The Secret of the Magical Journey"; "The Legend of the Magical Journey"; "The Adventure of the Magical Journey".

[0190] After calculation and comparison, the server finds that "The Secret of the Magic Journey" and the program name of the to-be-matched media resource program "The Legend of the Magic Journey" have reached the preset program name text similarity threshold.

[0191] Then, the server extracts the pending director information and pending actor information corresponding to the pending existing media resource program "The Secret of the Magical Journey". Assume that its director is Director Li, and the actors are Actor Wang and Actor Zhao.

[0192] Next, the server compares the pending director information (Director Li) and the pending actor information (Actor Wang and Actor Zhao) with the director information (Director Zhang) and actor information (Actor Liu and Actor Chen) of the to-be-matched media resource program.

[0193] After careful verification, it was found that the director information of the two was different and the actor information had no overlap.

[0194] Then, the server continues to search for other possible matching programs in the results of the full-text search.

[0195] For another example, the server finds that another program name "The Miracle of Magical Journey" also reaches the preset program name text similarity threshold.

[0196] The server extracts the director information corresponding to this program, which is Director Zhang, and the actors are actors Liu and Ma.

[0197] Comparing it with the director information and actor information of the media resource program to be matched, it is found that the director information is the same and the actor information has an intersection (both have actor Liu).

[0198] In this case, the server uses the pending existing media resource program "The Miracle of Magical Journey" as the successful matching alignment result of the media resource program to be matched.

[0199] Assume that during a busy media resource processing period, the server needs to process a large number of full-text retrieval tasks in a short period of time.

[0200] For example, 50 requests for media resource programs to be matched are received at the same time.

[0201] For the first media resource program to be matched, "The Journey of Interstellar Adventure", the server found three existing media resource programs with similar program names through full-text search.

[0202] After comparing the director and actor information, I found that none of them matched, so I continued searching.

[0203] Finally, a successful matching result was found in the fifth program with a similar name, which was called "Exploration of Interstellar Adventure". The director and actor information overlapped with the media program to be matched.

[0204] For the second media resource program to be matched, "Fantasy of the Fairy Tale World", the server quickly performed a full-text search and quickly found "Dream of the Fairy Tale World" with a similar name.

[0205] By extracting the director and actor information for comparison, we found that the directors were the same and the actors partially overlapped, which was determined to be a successful match alignment result.

[0206] With the continuous updating and enrichment of media resources, the tasks faced by servers are becoming more and more complex.

[0207] For example, a batch of foreign media programs have been newly added to the database, and the names and language habits are different from those in China.

[0208] When receiving a foreign media program to be matched, such as "The Adventure of Wonderland", the server can still accurately find domestic media programs with similar names, such as "The Adventure of Wonderland", through a powerful full-text search strategy.

[0209] Then, according to the established process, the director and actor information is extracted for comparison to complete the matching and alignment work.

[0210] In the process of processing large amounts of data, the server continuously optimizes its full-text retrieval algorithm to improve retrieval accuracy and efficiency.

[0211] For example, for some ambiguous words or expressions, the server can accurately determine the similarity of program names through more intelligent semantic understanding and context analysis.

[0212] Assume that the server receives a media resource program to be matched with a special theme, such as "The Epic of Ancient Wars".

[0213] Through full-text search, we found several programs with similar names, such as "Legends of Ancient Wars" and "Storms of Ancient Wars".

[0214] After comparing and screening the director and actor information one by one, "The Winds and Clouds of Ancient Wars" was finally determined to be a successfully matched program.

[0215] Through the above detailed scenario examples, the server can effectively use the full-text search strategy to accurately find suitable matching alignment results for the media asset programs to be matched, thereby improving the efficiency and quality of media asset management and utilization.

[0216] In the embodiment of the present invention, the aforementioned step S204 may be implemented through the following example execution.

[0217] On the basis that the matching alignment result indicates a successful match, obtaining at least one to-be-matched media resource program corresponding to the successful matching alignment result;

[0218] Extracting the program unique identifiers corresponding to the acquired media asset program to be matched and the at least one media asset program to be matched;

[0219] The mapping file is written according to the respective corresponding program unique identifiers.

[0220] In the embodiment of the present invention, the server determines, after a series of matching calculations and comparisons, that a matching alignment result of a to-be-matched media resource program is characterized as a successful match. For example, the to-be-matched media resource program is a film and television work titled "Song of Youth", directed by Director Li, and starring actors Wang and Zhao.

[0221] On the basis of successful matching, the server obtains at least one to-be-matched media resource program corresponding to the successful matching alignment result. Assume that two existing media resource programs matching with it are found, namely "Melody of Youth" and "Flying Youth".

[0222] Then, the server starts to extract the program unique identifiers corresponding to these programs. For "Song of Youth", its program unique identifier is 001; the program unique identifier of "Melody of Youth" is 002; and the program unique identifier of "Flying Youth" is 003.

[0223] Next, the server writes the corresponding unique identifiers of the extracted programs into a mapping file. In the mapping file, the corresponding relationship between these identifiers is clearly recorded, such as: 001-002, 001-003.

[0224] For example, another media resource program to be matched is "Dream Journey", directed by Director Zhang, and starring actors Liu and Chen. After the server determines that the match is successful, it obtains the corresponding three existing media resource programs, namely "Dream Journey", "Dream Journey" and "Dream Journey".

[0225] The unique program identifiers of the three existing media resource programs are 004, 005 and 006. The server writes these identifiers into the mapping file according to the corresponding relationship, such as: 007-004, 007-005, 007-006.

[0226] During a busy film and television data processing period, the server continuously completes the matching and alignment work and writes the mapping file.

[0227] For example, multiple successfully matched media resource programs were processed in a short period of time. One of them was "Echoes of History", which was matched with "Memories of the Past" and "Traces of Time", whose program unique identifiers were 008 and 009 respectively. The server quickly wrote 010-008 and 010-009 into the mapping file.

[0228] Then, the program "Poems of Love" was processed and successfully matched to "Romantic Love" and "Sweet Love", with the unique program identifiers 011 and 012. The server promptly recorded 013-011 and 013-012 in the mapping file.

[0229] As time goes by, the number of media programs processed by the server increases.

[0230] Assume that hundreds of successfully matched media resource programs are processed in one day, a large number of mapping relationships are generated and written into the mapping file accurately.

[0231] Even when faced with complex program information and numerous matching results, the server always remains efficient and accurate.

[0232] For example, for a media resource program to be matched named "Cultural Heritage", multiple related programs were successfully matched, such as "Continuation of Tradition", "Context of Civilization" and "Source of Culture", and the unique program identifiers are 014, 015 and 016 respectively.

[0233] The server quickly extracts the identifier and accurately writes 017-014, 017-015, and 017-016 to the mapping file.

[0234] During the processing, the server may encounter some special situations, for example, some to-be-matched media asset programs may be highly similar to multiple existing media asset programs, or some existing media asset programs may be successfully matched with the to-be-matched media asset programs in different categories or versions.

[0235] However, the server can accurately extract the unique identifier of the program and write it into the mapping file according to the rules to ensure the integrity and accuracy of the data.

[0236] Assume that the server receives a series of media resource programs to be matched, such as "The Legend of Martial Arts: Sword and Shadow Jianghu", "The Legend of Martial Arts: The Storm Rises Again", and "The Legend of Martial Arts: The Decisive Battle at the Peak".

[0237] After matching, they were successfully matched with existing media resource programs such as "The Legend of Swordsman", "The Record of Martial Arts" and "The Peak of Martial Arts". The unique program identifiers were 018, 019 and 020 respectively, and the corresponding unique identifiers of the media resource programs to be matched were 021, 022 and 023.

[0238] The server methodically writes 021-018, 022-019, and 023-020 into the mapping file, providing a clear and accurate reference for subsequent media asset management and use.

[0239] Through the above detailed scenario examples, the server can accurately obtain and process the program unique identifier based on successful matching, and accurately write it into the mapping file, providing a solid foundation for the effective management and utilization of media assets.

[0240] In the embodiments of the present invention, the following implementation modes are also provided.

[0241] In response to a media asset program search instruction, obtaining a search program unique identifier included in the media asset program search instruction;

[0242] The searched media asset program corresponding to the searched media asset program instruction is determined from the mapping file according to the searched program unique identifier, and the searched media asset program unique identifier is associated with the searched program unique identifier and written into the mapping file.

[0243] In the embodiment of the present invention, the following is an example of a detailed scenario of the steps in the above technical content, with the server as the execution subject:

[0244] In daily operation, the server is always ready to respond to various instructions. At a certain moment, the server receives an instruction to search for media resources.

[0245] This instruction includes a specific unique identifier of the program to be searched, such as "050". After receiving this instruction, the server immediately obtains the unique identifier of the program to be searched "050".

[0246] Then, the server starts to search from the mapping file according to the search program unique identifier. Assume that a large number of program unique identifiers and corresponding association information are stored in the mapping file.

[0247] The server quickly searches and compares the mapping file. After searching, the server finds the information associated with "050" in the mapping file.

[0248] Assuming that the media resource program to be searched is "Fantasy Adventure Journey", its corresponding program unique identifier is "050", and the unique identifiers of other related media resource programs that match it are also recorded, such as "051" and "052".

[0249] The server determines that "Fantasy Adventure Journey" is the media asset program to be searched corresponding to the media asset program search instruction.

[0250] For another example, the server receives another instruction to search for a media resource program, in which the unique identifier of the searched program is "100".

[0251] The server also quickly obtains this identifier and looks it up in the mapping file.

[0252] After a careful search, it was found that the media program corresponding to "100" is "Romantic Love Story", and it is also associated with the unique identifiers of two related media programs "101" and "102".

[0253] During a busy period, the server may receive a large number of media resource program search instructions.

[0254] For example, within one hour, 50 different search instructions were received.

[0255] The unique identifier of the search program in the first instruction is "080", and the server quickly finds the corresponding search media program "Historical Legend Chapter" and the associated "081" and "082".

[0256] Next, the second instruction's unique identifier for the searched program is "120", and the server determines that the corresponding searched media resource program is "Mysteries of Scientific Exploration", and obtains the associated "121" and "122".

[0257] In this way, the server methodically processes one search instruction after another, quickly and accurately determining the corresponding search media resource program from the mapping file.

[0258] Assume that a user wants to search for a series of media programs on a specific theme, such as the theme of "martial arts".

[0259] The server receives multiple search instructions, each of which carries a specific program unique identifier.

[0260] By searching in the mapping file, the server not only finds the media resource programs directly specified by the user, such as "Swordsman Chronicles" (unique identifier "150"), but also provides other related martial arts-themed media resource programs, such as "The Swordsman" (unique identifier "151") and "The Legend of Wulin Heroes" (unique identifier "152").

[0261] While processing the search instructions, the server may encounter some complex situations.

[0262] For example, some associations in the mapping file may change due to data updates.

[0263] However, the server will ensure that media resource programs can be accurately found based on the latest mapping relationships through real-time data synchronization and update mechanisms.

[0264] Assume that the server receives an older unique identifier "020" for searching a program. Due to multiple updates and arrangements of the media resource library, this identifier may no longer exist in the original mapping file or may have changed.

[0265] The server will start the data backtracking and update verification mechanism, reconfirm the currently valid mapping relationship, and finally find the media resource program "Classic Drama Collection" corresponding to "020", as well as the related "021" and "022".

[0266] Through the above detailed scenario examples, the server can efficiently and accurately respond to the media asset program search instruction, determine the media asset program search required by the user from the mapping file, and provide the user with fast and accurate media asset services.

[0267] In order to more clearly describe the solution implemented in this application, a relatively complete implementation scheme is provided below.

[0268] 1. Extract the directors, actors, program names (including program aliases), and program unique identifiers of all media assets in the matching media asset library, convert the program names into word vectors to generate vector text, encapsulate the extracted media asset directors, actors, program names, program unique identifiers and other data into vector metadata, use the program unique identifier as the vector ID, and finally write the vector text, vector metadata, and vector ID into the vector database together.

[0269] 2. Extract the director, actor, program name (including program alias), and program unique identifier of the media resource program to be retrieved, and convert the program name into a word vector.

[0270] 3. Perform similarity matching on the vector text corresponding to the program name generated in step 2 in the vector library generated in step 1 to obtain the vector metadata (i.e., director, actor, program name, and program unique identifier) ​​corresponding to the data with the highest similarity.

[0271] 4. For the program names that were not successfully matched in step 3, use full-text search to match the original media resource library to obtain relevant information of similar programs (director, actor, program name, program unique identifier).

[0272] 5. The media resource programs matched in step 3 and step 4 are then tested for the overlap of director and actor information with the initial media resource program. If the same director and actor information exists, it is determined to be a successful match. The unique identifier of the successfully matched media resource is written into the mapping file for subsequent use.

[0273] Please refer to Figure 2 , a device 110 for performing media resource program matching alignment based on a vector database, comprising:

[0274] The acquisition module 1101 is used to acquire existing media asset related information of multiple existing media asset programs in the media asset library, and vectorize the existing media asset related information to obtain existing media asset related vectors corresponding to each of the multiple existing media asset programs; acquire the to-be-matched media asset related information of the to-be-matched media asset program, and vectorize the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program;

[0275] The alignment module 1102 is used to obtain the matching alignment result corresponding to the media asset related vector to be matched according to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, combined with the full-text detection strategy; based on the matching alignment result indicating a successful match, determine the program unique identifier and write it into the mapping file.

[0276] It should be noted that the implementation principle of the aforementioned device 110 for matching and aligning media programs based on a vector database can refer to the implementation principle of the aforementioned method for matching and aligning media programs based on a vector database, and will not be repeated here. 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 a physical entity, or physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be fully 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 device 110 for matching and aligning media programs based on a vector database 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 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 described 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 a processor element or an instruction in the form of software.

[0277] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASIC), or one or more microprocessors (digital signal processors, DSP), or one or more field programmable gate arrays (FPGA), etc. For another example, when a module above is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0278] The embodiment of the present invention provides a computer device 100, which includes a processor and a non-volatile memory storing computer instructions. When the computer instructions are executed by the processor, the computer device 100 executes the aforementioned apparatus 110 for matching and aligning media resources and programs based on a vector database. Figure 3 As shown, Figure 3 The computer device 100 is a block diagram of a structure of a computer device 100 provided in an embodiment of the present invention. The computer device 100 comprises a device 110 for performing media asset program matching alignment based on a vector database, a memory 111 , a processor 112 and a communication unit 113 .

[0279] In order to realize data transmission or interaction, the memory 111, the processor 112 and the communication unit 113 are electrically connected to each other directly or indirectly. For example, the electrical connection between these components can be realized through one or more communication buses or signal lines. The device 110 for matching and aligning media programs based on a vector database includes at least one software function module that can be stored in the memory 111 in the form of software or firmware or fixed in the operating system (OS) of the computer device 100. The processor 112 is used to execute the device 110 for matching and aligning media programs based on a vector database stored in the memory 111, such as the software function modules and computer programs included in the device 110 for matching and aligning media programs based on a vector database.

[0280] An embodiment of the present invention provides a readable storage medium, which includes a computer program. When the computer program is running, it controls the computer device where the readable storage medium is located to execute the aforementioned device 110 for media asset program matching and alignment based on a vector database.

[0281] For illustrative purposes, the foregoing description is made with reference to specific embodiments. However, the above illustrative discussion is not intended to be exhaustive or to limit the present disclosure to the precise form disclosed. Numerous modifications and variations are possible in accordance with the above teachings. These embodiments are selected and described in order to best illustrate the principles of the present disclosure and its practical application, so that those skilled in the art can best utilize the present disclosure and utilize various embodiments with different modifications to suit the intended specific application.

Claims

1. A method for media resource program matching and alignment based on a vector database, characterized in that: include: Acquire existing media asset related information of multiple existing media asset programs in the media asset library, and vectorize the existing media asset related information to obtain existing media asset related vectors corresponding to each of the multiple existing media asset programs; Acquire the to-be-matched media asset related information of the to-be-matched media asset program, and vectorize the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program; According to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, combined with the full-text detection strategy, a matching alignment result corresponding to the media asset related vector to be matched is obtained; On the basis that the matching alignment result indicates a successful match, a unique identifier of the program is determined and written into a mapping file.

2. The method according to claim 1, characterized in that The step of obtaining existing media asset related information of a plurality of existing media asset programs in the media asset library, and vectorizing the existing media asset related information to obtain existing media asset related vectors corresponding to each of the plurality of existing media asset programs includes: Acquire existing media asset related information of multiple existing media asset programs in the media asset library, wherein the existing media asset related information includes director information, actor information, program name, and program unique identifier; Convert the program name into a word vector to generate an existing vector text; Encapsulating the director information, actor information, program name, and program unique identifier to obtain vector metadata; Using the program unique identifier as a vector ID; The existing vector text, the vector metadata and the vector ID are used as the existing media asset related vector, and the existing media asset related vector is written into a vector database.

3. The method according to claim 2, characterized in that The step of obtaining the to-be-matched media asset related information of the to-be-matched media asset program and vectorizing the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program includes: Obtaining the relevant information of the to-be-matched media asset program, wherein the relevant information of the to-be-matched media asset program includes director information, actor information, program name, and program unique identifier; The to-be-matched media asset related information is converted into a word vector to obtain a vector text as the to-be-matched media asset related vector of the to-be-matched media asset program.

4. The method according to claim 3, characterized in that The obtaining, based on the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, a matching alignment result corresponding to the media asset related vector to be matched in combination with a full-text detection strategy, includes: Performing traversal similarity comparison on the vector text in the vector database to obtain the vector similarity between the vector text and each of the existing media asset related vectors; If there is a pending existing media asset related vector whose vector similarity is higher than a preset vector similarity threshold, extracting pending director information and pending actor information corresponding to the pending existing media asset related vector; In the case where the pending director information and the pending actor information have an intersection with the director information and the actor information included in the to-be-matched media asset related information, the pending existing media asset program corresponding to the pending existing media asset related vector is used as a successful matching alignment result of the to-be-matched media asset program; If there is no existing media asset related vector to be determined whose vector similarity is higher than the preset vector similarity threshold, a full-text search strategy is used to obtain a matching alignment result corresponding to the media asset related vector to be matched.

5. The method according to claim 4, characterized in that The method of using a full-text search strategy to obtain a matching alignment result corresponding to the media asset related vector to be matched includes: Performing a full-text search on the vector database using the full-text search strategy to obtain pending existing media resource programs whose program names reach a preset program name text similarity threshold with the media resource program to be matched; Extracting the pending director information and the pending actor information corresponding to the pending existing media asset related vector; When the pending director information and the pending actor information have an intersection with the director information and the actor information included in the to-be-matched media asset related information, the to-be-determined existing media asset program corresponding to the to-be-determined existing media asset related vector is taken as a successful matching alignment result of the to-be-matched media asset program.

6. The method according to claim 5, characterized in that The determining of the program unique identifier and writing it into the mapping file based on the matching alignment result indicating successful matching includes: On the basis that the matching alignment result indicates a successful match, obtaining at least one to-be-matched media resource program corresponding to the successful matching alignment result; Extracting the program unique identifiers corresponding to the acquired media asset program to be matched and the at least one media asset program to be matched; The mapping file is written according to the respective corresponding program unique identifiers.

7. The method according to claim 1, characterized in that The method further comprises: In response to a media asset program search instruction, obtaining a search program unique identifier included in the media asset program search instruction; The searched media asset program corresponding to the searched media asset program instruction is determined from the mapping file according to the searched program unique identifier, and the searched media asset program unique identifier is associated with the searched program unique identifier and written into the mapping file.

8. A device for matching and aligning media resources and programs based on a vector database, characterized in that: include: An acquisition module is used to acquire existing media asset related information of multiple existing media asset programs in the media asset library, and vectorize the existing media asset related information to obtain existing media asset related vectors corresponding to each of the multiple existing media asset programs; acquire the to-be-matched media asset related information of the to-be-matched media asset program, and vectorize the to-be-matched media asset related information to obtain the to-be-matched media asset related vector of the to-be-matched media asset program; An alignment module, configured to obtain a matching alignment result corresponding to the media asset related vector to be matched according to the vector similarity between each of the existing media asset related vectors and the media asset related vector to be matched, in combination with a full-text detection strategy; On the basis that the matching alignment result indicates a successful match, a unique identifier of the program is determined and written into a mapping file.

9. A computer device, characterized in that: The computer device comprises a processor and a non-volatile memory storing computer instructions. When the computer instructions are executed by the processor, the computer device executes the method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium includes a computer program, and when the computer program is executed, the computer device where the readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

Images