Content recommendation method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a content recommendation method and device, equipment and a storage medium. The method is applied to the technical field of intelligent recommendation and comprises the steps that user information of a current user of a client side is obtained, and the user information comprises user behavior data; under the condition that the integrity of the user information is not lower than a preset value, calling a recommendation algorithm of a preset recommendation system to calculate and obtain a target recommendation list; under the condition that the integrity of the user information is lower than a preset value, calling a large language model to calculate and obtain a target recommendation list; and returning the target recommendation list to the client. In this way, the limitation of a traditional recommendation method in the aspect of cold start can be overcome.

Description

Technical Field

[0001] The present disclosure relates to the field of intelligent recommendation technology, and in particular to a content recommendation method, apparatus, device, and storage medium. Background Art

[0002] With the rapid development of information technology, online movie platforms have become one of the primary channels for people to access and watch movies. These platforms boast vast movie libraries, offering users unprecedented freedom of choice. However, with the rapid increase in the number of movies available, finding movies of interest within this vast library has become increasingly difficult. Therefore, an effective recommendation system is crucial for improving user experience. However, traditional recommendation methods, lacking sufficient historical data for new and under-informed users, struggle to accurately capture their interests and preferences, making it difficult to provide personalized recommendations.

[0003] Currently, no effective solution has been proposed to address the limitations of the above-mentioned traditional recommendation methods in cold start. Summary of the Invention

[0004] The present disclosure provides a content recommendation method, apparatus, device, and storage medium.

[0005] According to a first aspect of the present disclosure, a content recommendation method is provided. The method comprises:

[0006] Obtain user information of the current user of the client, where user information includes user behavior data;

[0007] When the completeness of the user information is not less than the preset value, the target recommendation list is calculated by calling the recommendation algorithm of the preset recommendation system; when the completeness of the user information is less than the preset value, the target recommendation list is calculated by calling the large language model;

[0008] Return the target recommendation list to the client.

[0009] According to the above aspects and any possible implementation, a further implementation is provided for obtaining user information of the current user of the client, including:

[0010] Receive a recommendation request sent by a client, wherein the recommendation request carries user information of the current user and is used to obtain a recommendation list.

[0011] In the above aspects and any possible implementation, a further implementation is provided, wherein the step of invoking a large language model to calculate and obtain a target recommendation list includes:

[0012] Fill in the movie rating template and recommendation list generation template based on user information, and input them into the large language model as prompt words to obtain two recommendation lists;

[0013] Merging the two recommendation lists to obtain a merged recommendation list;

[0014] The merged recommendation list and user information are filled into the movie rating template, which is then input into the large language model as prompt words for rating. The target recommendation list is obtained by sorting the scores.

[0015] In accordance with the above aspects and any possible implementation, a further implementation is provided, wherein a movie rating template and a recommendation list generation template are filled in according to user information, and prompt words are respectively input into a large language model to obtain two recommendation lists, including obtaining a first recommendation list in the following manner:

[0016] Use the recall algorithm to generate a list of candidate movie recommendations based on user information;

[0017] Fill the candidate movie recommendation list and user information into the movie rating template to obtain the first movie rating prompt word;

[0018] Input the first movie rating prompt word into the large language model to obtain the candidate movie ratings corresponding to the candidate movie recommendation list;

[0019] The candidate movie recommendation list is sorted according to the candidate movie scores to obtain a first recommendation list.

[0020] In accordance with the above aspects and any possible implementation, a further implementation is provided, wherein a movie rating template and a recommendation list generation template are filled in according to user information, and prompt words are respectively input into a large language model to obtain two recommendation lists, including obtaining a second recommendation list in the following manner:

[0021] Fill the user information into the recommendation list generation template to obtain the recommendation list generation prompt words;

[0022] The recommendation list generation prompt words are input into the large language model to obtain a second recommendation list.

[0023] According to the above aspects and any possible implementation, a further implementation is provided, in which two recommendation lists are merged to obtain a merged recommendation list, including:

[0024] The first recommendation list and the second recommendation list are merged, and preset filtering content is removed to obtain a merged recommendation list, wherein the preset filtering content includes duplicate content and illegal content.

[0025] In accordance with the above aspects and any possible implementation, a further implementation is provided, wherein the merged recommendation list and user information are added to a movie rating template, and the movie rating template is input into a large language model as prompt words for rating. The target recommendation list is obtained by sorting the scores, including:

[0026] Fill the merged recommendation list and user information into the movie rating template to obtain a second movie rating prompt word;

[0027] Input the second movie rating prompt word into the large language model to obtain the movie rating corresponding to the merged recommendation list;

[0028] Sort the merged recommendation list according to the movie ratings corresponding to the merged recommendation list to obtain the target recommendation list.

[0029] According to a second aspect of the present disclosure, a content recommendation device is provided. The device includes:

[0030] An acquisition module is used to obtain user information of the current user of the client, wherein the user information includes user behavior data;

[0031] A calling module is used to call the recommendation algorithm of the preset recommendation system to calculate and obtain the target recommendation list when the completeness of the user information is not less than the preset value; when the completeness of the user information is less than the preset value, call the large language model to calculate and obtain the target recommendation list;

[0032] The return module is used to return the target recommendation list to the client.

[0033] According to a third aspect of the present disclosure, an electronic device is provided, comprising: a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the program.

[0034] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method according to the first aspect of the present disclosure is implemented.

[0035] The present disclosure utilizes the deep semantic understanding capability of the large language model to recommend content when user information is insufficiently complete, thereby improving the relevance and personalization of recommendations and overcoming the limitations of traditional recommendation methods in cold start.

[0036] It should be understood that the contents described in the Summary of the Invention section are not intended to limit the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are provided for a better understanding of the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, among which:

[0038] Figure 1 A flowchart of a content recommendation method according to an embodiment of the present disclosure is shown;

[0039] Figure 2 shows a topology diagram of a content recommendation system according to an embodiment of the present disclosure;

[0040] Figure 3 A block diagram of a content recommendation device according to an embodiment of the present disclosure is shown;

[0041] Figure 4 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0042] To make the purpose, technical solutions, and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present disclosure.

[0043] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0044] The content recommended in the present disclosure includes but is not limited to the following types: text content (news articles, blogs, books, magazines, academic papers, etc.), video content (movies, TV series, short films, video blogs, online courses, live broadcasts, etc.), audio content (music, podcasts, audio books, radio programs, etc.), pictures and images (photographs, illustrations, designs, user-generated pictures, etc.), goods and services (goods on e-commerce platforms such as clothing, electronic products, daily necessities, and food; mobile applications, computer software, games; travel services, catering services, financial services, health services, etc.).

[0045] The present disclosure utilizes the deep semantic understanding capability of the large language model to recommend content when user information is insufficiently complete, thereby improving the relevance and personalization of recommendations and overcoming the limitations of traditional recommendation methods in cold start.

[0046] Figure 1 1 is a flowchart of a content recommendation method 100 according to an embodiment of the present disclosure. The method 100 includes the following steps:

[0047] Step S110, obtaining user information of the current user of the client, wherein the user information includes user behavior data;

[0048] Step S120: If the completeness of the user information is not less than a preset value, the target recommendation list is calculated by calling the recommendation algorithm of the preset recommendation system; if the completeness of the user information is less than the preset value, the target recommendation list is calculated by calling the large language model;

[0049] Step S130: Return the target recommendation list to the client.

[0050] In some embodiments, step S110, obtaining user information of the current user of the client, includes:

[0051] Receive a recommendation request sent by a client, wherein the recommendation request carries user information of the current user and is used to obtain a recommendation list.

[0052] Optionally, the user information may include the user's behavioral data (including viewing history, search history, likes, comments, etc.) and the user's current context information (such as search terms, currently paused or completed viewing content information, geographic location, device type, etc.).

[0053] According to the embodiments of the present disclosure, by receiving a recommendation request sent by a client, it is possible to identify and confirm the user of the current request, thereby ensuring the personalization and pertinence of the recommended content.

[0054] In some embodiments, the user information completeness in step S120 is evaluated according to a preset completeness assessment method. For example, the input parameters of the recommendation algorithm of a preset recommendation system include basic user information, user behavior data, user context information, etc. In the user information completeness assessment, the recommendation system can set a threshold. Only when the completeness of the user information reaches this threshold will the recommendation algorithm of the preset recommendation system be triggered. This is done to ensure the accuracy and relevance of the recommendation results and avoid inaccurate inferences when insufficient information is available. In this way, personalized recommendation services can be provided to users more effectively.

[0055] In some embodiments, step S120, invoking the large language model to calculate and obtain the target recommendation list includes:

[0056] Fill in the movie rating template and recommendation list generation template based on user information, and input them into the large language model as prompt words to obtain two recommendation lists;

[0057] Merging the two recommendation lists to obtain a merged recommendation list;

[0058] The merged recommendation list and user information are filled into the movie rating template, which is then input into the large language model as prompt words for rating. The target recommendation list is obtained by sorting the scores.

[0059] Movie rating templates and recommendation list generation templates can be set according to actual needs, for example:

[0060] [Movie Rating Template] Please generate a rating for the following {candidate movie recommendation list} based on {user information}. The rating range is 1 to 5, with 5 being the highest.

[0061] [Recommendation list generation template] Based on {user information}, recommend 10 movies to the user. The recommended movies should be highly relevant.

[0062] According to an embodiment of the present disclosure, the generation capability of a large language model is utilized to preliminarily generate two recommendation lists based on user information and templates based on different recommendation logics or algorithms, and then the two recommendation lists are merged into one to increase the diversity and coverage of the recommended content; through multi-step screening and scoring, user preferences can be more accurately predicted, thereby providing movie recommendations that better meet user expectations.

[0063] In some embodiments, a movie rating template and a recommendation list generation template are populated based on user information, and prompt words are respectively input into a large language model to obtain two recommendation lists, including obtaining a first recommendation list in the following manner:

[0064] Use the recall algorithm to generate a list of candidate movie recommendations based on user information;

[0065] Fill the candidate movie recommendation list and user information into the movie rating template to obtain the first movie rating prompt word;

[0066] Input the first movie rating prompt word into the large language model to obtain the candidate movie ratings corresponding to the candidate movie recommendation list;

[0067] The candidate movie recommendation list is sorted according to the candidate movie scores to obtain a first recommendation list.

[0068] Optionally, the recall algorithm includes but is not limited to the following types: (1) content-based recall: recommending similar content based on the characteristics of content that the user liked in the past; (2) collaborative filtering recall: leveraging the preferences of user groups to recommend content that is popular or liked by similar users; (3) model-based recall: using machine learning models to predict content that the user may be interested in; (4) hybrid recall: combining the above methods to improve the coverage and accuracy of recommendations.

[0069] In some embodiments, a model-based recall algorithm may be selected to generate a candidate movie recommendation list based on user information.

[0070] Specifically, the recall problem can be modeled as a proximity search problem in vector space through a unified modeling architecture using vectorized recall. Suppose the recall problem contains two entities Q and T. Q is a user and T is a movie. Finding movies that the user likes directly is user-to-item (U2I) recall. Q and T are both movies. Finding other movies similar to the one the user likes is item-to-item (I2I) recall. Q and T are both users. First, find users similar to the current user and then recommend movies that similar users like to the current user. This is user-to-user-item (U2U2I) recall. The basic steps of vectorized recall include: training a model M, mapping each instance q in Q and each instance t in T to the same vector; mapping all instances in T into vectors, storing them in a vector database, and indexing them; when serving online, for an instance q in Q, map it to a vector Emb through M. q Then, in the vector database, we use the approximate nearest neighbor ANN search algorithm to find the vectors that match Emb. q The K nearest neighbor vectors of category T are used as the recall result. Vectorized recall algorithms include Item2Vec, Aibib recall, FM recall, and the Twin Towers model. The goal of a recall algorithm is to ensure that as many relevant recommendations as possible are obtained. It is typically used in the initial screening phase to generate a large candidate set, which is then further filtered and ranked using other algorithms.

[0071] For example:

[0072] The {candidate movie recommendation list} obtained by the recall algorithm is as follows:

[0073] 1. Movie Name: Movie A

[0074] 2. Movie Name: Movie B

[0075] 3. Movie Name: Movie C

[0076] [Movie Rating Template] Please generate a rating for the following {candidate movie recommendation list} based on {user information}. The rating range is 1 to 5, with 5 being the highest.

[0077] The candidate movie recommendation list and user information are filled into the movie rating template to obtain the first movie rating prompt word; the first movie rating prompt word is input into the large language model to obtain the candidate movie ratings corresponding to the candidate movie recommendation list as follows:

[0078] 1. Movie name: "Movie A", Rating: 3

[0079] 2. Movie name: "Movie B", Rating: 2

[0080] 3. Movie name: "Movie C", Rating: 5

[0081] Sort the candidate movie recommendation list according to the candidate movie scores to get the first recommendation list:

[0082] 1. Movie name: "Movie C", Rating: 5

[0083] 2. Movie name: "Movie A", Rating: 3

[0084] 3. Movie name: "Movie B", Rating: 2

[0085] According to the embodiments of the present disclosure, a recall algorithm is used to form a larger candidate set, ensuring that content recommendations can cover various types of movies that users may be interested in. The generation and understanding capabilities of the large language model are used to generate a score for each movie in the candidate movie list. This score reflects the model's predicted user preference for each movie. Sorting is performed based on the score results of the large language model, ensuring that movies with high scores are ranked first. The first recommendation list generated in this way will prioritize the movies that users may be most interested in. The breadth of coverage of the recall algorithm and the deep personalized scoring of the large language model are combined to provide both extensive and accurate personalized movie recommendations. In this way, the recommendation system can better meet the personalized needs of users, improve user satisfaction and the relevance of recommendations.

[0086] In some embodiments, a movie rating template and a recommendation list generation template are populated based on user information, and prompt words are respectively input into the large language model to obtain two recommendation lists, including obtaining the second recommendation list in the following manner:

[0087] Fill the user information into the recommendation list generation template to obtain the recommendation list generation prompt words;

[0088] The recommendation list generation prompt words are input into the large language model to obtain a second recommendation list.

[0089] The number of contents in the recommendation list can be determined based on actual needs. The recommendation list generation template can also include movie quality requirements, for example:

[0090] [Recommendation list generation template] Based on {user information}, recommend {number of recommendations} movies to the user. The recommended movies should be highly relevant and have a public rating on the platform of no less than {minimum rating}.

[0091] According to an embodiment of the present disclosure, the advanced text processing and generation capabilities of a large language model are utilized in combination with user information to generate a personalized movie recommendation list to improve user satisfaction and the overall effect of the recommendation system.

[0092] In some embodiments, merging two recommendation lists to obtain a merged recommendation list includes:

[0093] The first recommendation list and the second recommendation list are merged, and preset filtering content is removed to obtain a merged recommendation list, wherein the preset filtering content includes duplicate content and illegal content.

[0094] For example, the preset filtering content includes content that needs to be filtered out, such as duplicate content, content involving pornography and violence, and content that does not have copyright.

[0095] According to the embodiments of the present disclosure, removing duplicate content can ensure that each item in the recommendation list is unique, preventing users from seeing the same recommendations and improving user experience; removing illegal content is to ensure that the recommendation system complies with laws, regulations and platform policies, and avoid recommending content that may cause controversy or be inappropriate.

[0096] In some embodiments, the merged recommendation list and user information are added to a movie rating template, which is then used as a prompt word to input into a large language model for rating. The target recommendation list is obtained by sorting the scores, including:

[0097] Fill the merged recommendation list and user information into the movie rating template to obtain a second movie rating prompt word;

[0098] Input the second movie rating prompt word into the large language model to obtain the movie rating corresponding to the merged recommendation list;

[0099] Sort the merged recommendation list according to the movie ratings corresponding to the merged recommendation list to obtain the target recommendation list.

[0100] According to the embodiments of the present disclosure, through scoring and sorting, information overload can be effectively reduced, making it easier for users to find content they are interested in; by utilizing the advanced processing capabilities of the large language model, combined with user information and recommendation lists, a target recommendation list with personalized scoring and optimized sorting is generated to provide more accurate, relevant and high-quality movie recommendations.

[0101] In some embodiments, after returning the target recommendation list to the client, the client is used to record user behavior data and update user information based on the user behavior data, where the user behavior data includes user operations on the content in the target recommendation list, such as clicks, plays, comments, etc.

[0102] According to the embodiments of the present disclosure, by collecting user behavior data and updating user information, the continuously updated user information is used for content recommendation, which can improve the relevance and accuracy of the recommendation and the overall user experience.

[0103] The method 100 of the embodiment of the present disclosure is described below with reference to a specific implementation case:

[0104] The embodiments of the present disclosure can recommend movies to new users based on a large language model, such as Figure 2 FIG. 1 shows a topology diagram of a content recommendation system according to an embodiment of the present disclosure. The embodiment of the present disclosure specifically includes the following steps:

[0105] Step 1: When the client searches, enters the movie playback page, pauses playback, or completes playback, etc., and a recommendation list is needed, the client initiates a request to the recommendation service gateway to obtain the movie recommendation list, and carries the user's context information, including search terms, the currently paused or completed movie, the user's current geographic location, and other information.

[0106] Step 2: After receiving the user request, the recommendation service gateway calculates the movie recommendation list. This is achieved in the following way:

[0107] (1) The recommendation service gateway obtains user behavior data (including viewing history, search history, likes, comments, etc.), context information (including search terms, currently paused or completed movies, user’s current geographic location, etc.) and personalized information.

[0108] (2) Analyze whether the user's information is suitable for the recommendation algorithm of the traditional recommendation system. If there is enough information, call the recommendation algorithm of the traditional recommendation system to calculate the recommendation list. If the information is insufficient, call the interface of the large language model recommendation server to calculate the recommendation list.

[0109] (3) After receiving the request, the large model recommendation server fills in the candidate movie ratings and recommendation lists to generate two prompt templates, and inputs the two prompts into the large language model to obtain two recommendation lists: ① The large model recommendation server calls the recall algorithm of the traditional recommendation system to obtain the candidate movie list, and fills the candidate movie rating prompt template with the candidate movie list and user information, inputs the large language model, and obtains recommendation list 1 according to the rating sorting; ② Fills the recommendation list with user information to generate a prompt template, inputs the large language model, and obtains recommendation list 2.

[0110] (4) Merge the recommendation list 1 and the recommendation list 2 to remove duplicate content, pornographic and violent content, and content that does not have copyright, etc. that needs to be filtered out.

[0111] (5) The merged recommendation list and user information are used to fill the candidate movie rating prompt template, input into the large language model for scoring, and the final recommendation list is obtained by sorting according to the score.

[0112] (6) The large model recommendation server returns the final recommendation list to the recommendation service gateway.

[0113] Step 3: The recommendation service gateway returns the final recommendation list to the client.

[0114] Step 4: The client displays the recommendation list to the user and records the user's behavior, such as clicks, plays, comments, etc., and transmits the user's behavior data to the recommendation service gateway.

[0115] Step 5: After receiving the behavioral data transmitted by the client, the recommendation service gateway updates the user information for use in the next recommendation.

[0116] According to the embodiments of the present disclosure, the following technical effects are achieved: providing personalized movie recommendations by analyzing user behavior data (including viewing history, search history, likes, comments, etc.), contextual information (including search terms, currently paused or finished movies, user's current geographic location, etc.) and personalized information. The system can handle new users and insufficient information, automatically filter inappropriate content, and dynamically optimize recommendation results based on user feedback to adapt to changes in user preferences. This technology utilizes the deep semantic understanding capabilities of large language models to improve the relevance and personalization of recommendations, and solves the limitations of traditional recommendation methods in cold starts for new users and users with insufficient information.

[0117] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0118] The above is an introduction to the method embodiment. The following is a further explanation of the solution disclosed in the present disclosure through an apparatus embodiment.

[0119] Figure 3 FIG. 3 shows a block diagram of a content recommendation device 300 according to an embodiment of the present disclosure. Figure 3 As shown, the apparatus 300 includes:

[0120] An acquisition module 310 is configured to acquire user information of the current user of the client, wherein the user information includes user behavior data;

[0121] The calling module 320 is configured to, if the completeness of the user information is not less than a preset value, call the recommendation algorithm of the preset recommendation system to calculate and obtain a target recommendation list; if the completeness of the user information is less than a preset value, call the large language model to calculate and obtain the target recommendation list;

[0122] The return module 330 is configured to return the target recommendation list to the client.

[0123] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0124] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0125] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0126] Figure 4 A block diagram of an exemplary electronic device 400 capable of implementing embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0127] The electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a ROM 402 or a computer program loaded from a storage unit 408 into a RAM 403. The RAM 403 may also store various programs and data required for the operation of the electronic device 400. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An I / O interface 405 is also connected to the bus 404.

[0128] Multiple components in the electronic device 400 are connected to the I / O interface 405, including an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a magnetic disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0129] The computing unit 401 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as method 100. For example, in some embodiments, the method 100 can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the method 100 described above can be performed. Alternatively, in other embodiments, the computing unit 401 can be configured to perform the method 100 in any other appropriate manner (e.g., by means of firmware).

[0130] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0131] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0132] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0133] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0134] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0135] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0136] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0137] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A content recommendation method, characterized in that: include: Obtaining user information of the current user of the client, wherein the user information includes user behavior data; When the completeness of the user information is not less than a preset value, the target recommendation list is calculated by calling the recommendation algorithm of the preset recommendation system; when the completeness of the user information is less than a preset value, the target recommendation list is calculated by calling the large language model; The target recommendation list is returned to the client.

2. The method according to claim 1, characterized in that The obtaining of user information of the current user of the client includes: A recommendation request sent by a client is received, wherein the recommendation request carries user information of a current user, and the recommendation request is used to obtain a recommendation list.

3. The method according to claim 1, characterized in that: The calling of the large language model to calculate and obtain the target recommendation list includes: Fill in a movie rating template and a recommendation list generation template according to the user information, and input them into a large language model as prompt words to obtain two recommendation lists; Merging the two recommendation lists to obtain a merged recommendation list; The merged recommendation list and the user information are filled into the movie rating template, and are input into a large language model as prompt words for rating, and the target recommendation list is obtained according to the rating sorting.

4. The method according to claim 3, characterized in that The step of filling the movie rating template and the recommendation list generation template according to the user information and inputting the prompt words into the large language model to obtain two recommendation lists includes obtaining the first recommendation list in the following manner: Generate a candidate movie recommendation list based on the user information using a recall algorithm; Filling the candidate movie recommendation list and the user information into a movie rating template to obtain a first movie rating prompt word; Inputting the first movie rating prompt word into the large language model to obtain the candidate movie ratings corresponding to the candidate movie recommendation list; The candidate movie recommendation list is sorted according to the candidate movie scores to obtain a first recommendation list.

5. The method according to claim 3, characterized in that: The filling of the movie rating template and the recommendation list generation template according to the user information, and inputting the prompt words into the large language model respectively to obtain two recommendation lists, includes obtaining the second recommendation list in the following manner: Fill the user information into a recommendation list generation template to obtain a recommendation list generation prompt word; The recommendation list generation prompt words are input into the large language model to obtain a second recommendation list.

6. The method according to claim 3, characterized in that The merging of the two recommendation lists to obtain a merged recommendation list includes: The first recommendation list and the second recommendation list are merged, and preset filtering content is removed to obtain a merged recommendation list, wherein the preset filtering content includes duplicate content and illegal content.

7. The method according to claim 3, characterized in that The step of filling the merged recommendation list and the user information into the movie rating template, inputting the merged recommendation list and the user information into a large language model as prompt words for rating, and obtaining the target recommendation list according to the rating sorting includes: Filling the merged recommendation list and the user information into a movie rating template to obtain a second movie rating prompt word; Inputting the second movie rating prompt word into the large language model to obtain the movie rating corresponding to the merged recommendation list; The merged recommendation list is sorted according to the movie scores corresponding to the merged recommendation list to obtain a target recommendation list.

8. A content recommendation device, characterized in that: include: An acquisition module, used to acquire user information of the current user of the client, wherein the user information includes user behavior data; A calling module, configured to, when the completeness of the user information is not less than a preset value, call a recommendation algorithm of a preset recommendation system to calculate and obtain a target recommendation list; when the completeness of the user information is less than a preset value, call a large language model to calculate and obtain a target recommendation list; A returning module is used to return the target recommendation list to the client.

9. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

Images