Creative recommendation method and device, electronic equipment and storage medium

Abstract

The application discloses a creative recommendation method and device, electronic equipment and a storage medium, wherein the method comprises: determining the user type of a user according to a user portrait; extracting platform historical hotspot information, rising trend information and user feature information of the user according to the user type and the current time; inputting the user feature information, the platform historical hotspot information and the rising trend information into a text generation model to generate a paradigm text, wherein the text generation model is determined by the user type; inputting the paradigm text into a chat generation pre-training converter to receive a reply text of the chat generation pre-training converter; performing information extraction on the paradigm text to obtain at least one recommended creative; and recommending the at least one recommended creative to the user. Using the method, a creative is automatically generated and recommended according to the information of the user, which assists the user in creative planning.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a creative recommendation method, apparatus, electronic device, and storage medium. Background Technology

[0002] The process of users creating videos / text notes mainly involves three stages: creative planning (including determining the direction and script design), execution (including shooting, copywriting, video production), and publishing and operation. Among these, the creative planning stage is the most crucial and time-consuming, and it is the cornerstone that determines whether a video is valuable and attracts attention.

[0003] Currently, the main pain points in the creative planning stage are as follows: 1. Not knowing what direction to publish content in; 2. Having a direction but running out of creative ideas; 3. Not knowing whether the video will receive promotion and attention after it is released. Therefore, creative planning consumes a lot of energy and usually requires maintaining a separate planning team to ensure a continuous output of creative ideas, resulting in huge costs in terms of both money and manpower. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, this application provides a creative recommendation method, apparatus, electronic device, and storage medium that can automatically generate and recommend creative ideas based on user information, assisting users in creative planning.

[0005] In a first aspect, embodiments of this application provide a creative recommendation method, the method comprising:

[0006] Determine the user type based on user profiles;

[0007] Based on user type and current time, extract historical hot topics, upward trend information, and user characteristic information from the platform;

[0008] User characteristic information, platform historical hot information, and upward trend information are input into the text generation model to generate paradigm text. The text generation model is determined by the user type.

[0009] Input the paradigm text into the chat-generated pre-trained converter and receive the reply text from the chat-generated pre-trained converter;

[0010] Extract information from the paradigm text to obtain at least one recommended creative;

[0011] Recommend at least one creative idea to the user.

[0012] In one possible implementation, determining the user type based on the user profile includes:

[0013] Based on user profiles, determine whether a user is an opinion leader.

[0014] If the user is an opinion leader, then the user type is determined to be type A; otherwise, the user profile is used to determine whether the user's registration time is less than n days, where n is an integer greater than or equal to 7.

[0015] If a user's registration time is less than n days, then the user type is determined to be type B. Otherwise, based on the user profile, determine whether the number of notes published by the user is greater than m and whether the average note publishing frequency is greater than q per week, where m is an integer greater than or equal to 5 and q is an integer greater than or equal to 1.

[0016] If a user publishes more than m notes and their average note publishing frequency is greater than q notes per week, then the user type is determined to be type C; otherwise, the user type is determined to be type D.

[0017] In one possible implementation, based on user type and current time, historical trend information, upward trend information, and user characteristic information of the platform are extracted, including:

[0018] Determine the extraction duration and feature information tags based on user type;

[0019] Using the current time as the endpoint, extract historical data of the same duration as the extracted time from the platform's historical database;

[0020] Analyze historical data to identify historical hot topics and upward trends on the platform;

[0021] Based on the feature information tags, extract user feature information from the user database.

[0022] In one possible implementation, when the user type is type A, the feature information tags include: opinion leader fan category preference distribution, characteristic group tags, interest point preferences in the past p days, and posting preferences, where p is an integer greater than or equal to 56;

[0023] When the user type is B, the feature information tags include: registration interests, occupation information, followed opinion leader tags, past day interest preferences, note category preferences, and preferred opinion leader tags, where w is an integer greater than or equal to 7;

[0024] When the user type is C, the feature information tags include: posting preferences, preferred opinion leader tags, recent p-day interest preferences, and followed opinion leader tags;

[0025] When the user type is D, the feature information tags include: followed opinion leader tags, preferred opinion leader tags, note category preferences, interest preferences in the past p days, and interest preferences in the past p days.

[0026] In one possible implementation, information extraction is performed on the paradigm text to obtain at least one recommended idea, including:

[0027] Information is extracted from the paradigm text to obtain at least one first idea;

[0028] For each of the at least one first creative ideas, feature extraction is performed to obtain at least one first feature, wherein the at least one first feature corresponds one-to-one with the at least one first creative idea;

[0029] Each of the at least one first feature is compared with the historical creative features in the first work library. Based on the comparison results, at least one recommended creative is determined from at least one first creative. The first work library is the collection of all the user's creative works whose play count / fan count is in the bottom 20%.

[0030] In one possible implementation, each of the at least one first feature is compared with historical creative features in a first creative library, and at least one recommended creative is determined from the at least one first creative based on the comparison results, including:

[0031] Calculate the similarity between each first feature and the historical creative features in the first work library, and take the maximum value of the calculated similarity as the creative value of each first feature to obtain at least one creative value, wherein at least one creative value corresponds one-to-one with at least one first creative.

[0032] At least one recommended creative is determined from at least one first creative based on at least one creative value, wherein the creative value of each of the at least one recommended creative is greater than a preset threshold.

[0033] In one possible implementation, when the transcoding requirement is high quality, the similarity can be represented by formula ①:

[0034]

[0035] Where d(x,y) represents the similarity between each first feature x and the historical creative feature y in the first work library, xi represents the i-th element in each first feature x, yi represents the i-th element in the historical creative feature y in the first work library, h represents the total number of elements in each first feature x, and i is an integer greater than or equal to 1.

[0036] Secondly, embodiments of this application provide a creative recommendation device, comprising:

[0037] The analysis module is used to determine the user type based on the user profile;

[0038] The extraction module is used to extract historical hot topics, upward trend information, and user characteristic information of the platform based on user type and current time.

[0039] The question-and-answer module is used to input user characteristic information, platform historical hot information and upward trend information into the text generation model to generate paradigm text. The text generation model is determined by the user type. The paradigm text is then input into the chat generation pre-trained converter and the module receives the reply text from the chat generation pre-trained converter.

[0040] The extraction module is also used to extract information from the paradigm text to obtain at least one recommended idea;

[0041] The recommendation module is used to recommend at least one creative idea to the user.

[0042] Thirdly, embodiments of this application provide an electronic device, including: a processor connected to a memory for storing a computer program, and the processor for executing the computer program stored in the memory to cause the electronic device to perform the method as described in the first aspect.

[0043] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that causes a computer to perform the method as described in the first aspect.

[0044] Fifthly, embodiments of this application provide a computer program product, the computer program product including a non-transitory computer-readable storage medium storing a computer program, and a computer operable to perform the method as described in the first aspect.

[0045] Implementing the embodiments of this application has the following beneficial effects:

[0046] In this embodiment, different text generation models with different styles are designed for different types of users. Then, based on user characteristics, platform historical trend information, and upward trend information, different styles of creative text are generated. This creative text is then input into a chat generation pre-trained converter to obtain creative response text through dialogue. Finally, information is extracted from the response text to obtain creative content for user recommendation. This achieves automatic creative generation and recommendation, saving significant manpower and operating costs. Furthermore, the generated creative content, based on user characteristics, platform historical trend information, and upward trend information, resonates with users while ensuring a certain level of promotion and attention. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0048] Figure 1 A schematic diagram of the hardware structure of a creative recommendation device provided for an embodiment of this application;

[0049] Figure 2 A flowchart illustrating a creative recommendation method provided for an embodiment of this application;

[0050] Figure 3 A functional module block diagram of a creative recommendation device provided for embodiments of this application;

[0051] Figure 4 This is a schematic diagram of the structure of an electronic device provided for an embodiment of this application. Detailed Implementation

[0052] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0053] The terms "first," "second," "third," and "fourth," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0054] In this document, the term "implementation" means that a specific feature, result, or characteristic described in connection with an implementation may be included in at least one implementation of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same implementation, nor is it a separate or alternative implementation mutually exclusive with other implementations. It will be explicitly and implicitly understood by those skilled in the art that the implementations described herein can be combined with other implementations.

[0055] See Figure 1, Figure 1 This is a schematic diagram of the hardware structure of a creative recommendation device provided in an embodiment of this application. The creative recommendation device 100 includes at least one processor 101, a communication line 102, a memory 103, and at least one communication interface 104.

[0056] In this embodiment, the processor 101 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application.

[0057] Communication line 102 may include a path for transmitting information between the aforementioned components.

[0058] The communication interface 104 can be any transceiver-like device (such as an antenna) used to communicate with other devices or communication networks, such as Ethernet, RAN, wireless local area networks (WLAN), etc.

[0059] The memory 103 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0060] In this embodiment, the memory 103 can exist independently and be connected to the processor 101 via the communication line 102. Alternatively, the memory 103 can be integrated with the processor 101. The memory 103 provided in this embodiment is typically non-volatile. The memory 103 stores computer execution instructions for implementing the scheme of this application, and its execution is controlled by the processor 101. The processor 101 executes the computer execution instructions stored in the memory 103 to implement the method provided in the following embodiments of this application.

[0061] In an optional implementation, the computer execution instructions may also be referred to as application code, and this application does not specifically limit this terminology.

[0062] In an optional implementation, processor 101 may include one or more CPUs, for example... Figure 1 CPU0 and CPU1 in the CPU.

[0063] In an alternative implementation, the creative recommendation device 100 may include multiple processors, such as... Figure 1 Processors 101 and 107 are shown in the diagram. Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor here may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0064] In optional implementations, if the creative recommendation device 100 is a server, for example, it can be a standalone server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The creative recommendation device 100 may further include an output device 105 and an input device 106. The output device 105 communicates with the processor 101 and can display information in various ways. For example, the output device 105 can be a liquid crystal display (LCD), a light emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector. The input device 106 communicates with the processor 101 and can receive user input in various ways. For example, the input device 106 can be a mouse, keyboard, touchscreen device, or sensor device.

[0065] The aforementioned creative recommendation device 100 can be a general-purpose device or a special-purpose device. The embodiments of this application do not limit the type of creative recommendation device 100.

[0066] The following will provide a detailed description of one of the creative recommendation methods disclosed in this application.

[0067] See Figure 2 , Figure 2 A flowchart illustrating an idea recommendation method provided for an embodiment of this application. The idea recommendation method includes the following steps:

[0068] 201: Determine the user type based on the user profile.

[0069] In this implementation, user types can include four categories: A, B, C, and D. Category A corresponds to Key Opinion Leaders (KOLs), who are experts or authorities in a particular field, or possess more accurate information, are accepted or trusted by relevant groups, and have a significant influence on the behavior of those groups. Category B corresponds to new users, such as users who registered less than n days ago. In this implementation, n is an integer greater than or equal to 7, typically 7. Category C corresponds to active users, i.e., among users not in categories A or B, those who have published more than m notes and whose average note posting frequency is greater than q notes per week. Here, m is an integer greater than or equal to 5, typically 5, and q is an integer greater than or equal to 1, typically 1. The remaining users are classified as category D.

[0070] Based on this, when determining user type, first check the user tags in the user profile to determine if the user is a KOL (Key Opinion Leader) user. If so, classify the user as Type A. Otherwise, check the user registration time in the user profile to determine if the user registered less than 7 days ago. If so, classify the user as Type B. Otherwise, check the user history in the user profile to determine if the user has published more than 5 notes and the average posting frequency is more than 1 note per week. If so, classify the user as Type C; otherwise, classify the user as Type D.

[0071] 202: Extract historical hot topics, upward trend information, and user characteristic information from the platform based on user type and current time.

[0072] In this implementation, the extraction duration and feature information tags are first determined based on user type. Specifically, for Type A users, who are the platform's core users, the extraction duration can be extended as much as possible, for example, extracting 2 / 3 of their registration time. For Type B users, who are new users, their entire registration time can be used as the extraction duration. For Type C users, who are active users, the extraction duration can be the past 3 months. For Type D users, who are inactive users, the extraction duration can be the past 1 month. Then, using the current time as the endpoint, historical data of the same duration as the extraction duration is extracted from the platform's historical database. This historical data is then analyzed to determine historical hotspot information and upward trend information for the platform.

[0073] In this implementation, when the user type is A, the feature information tags include: opinion leader follower category preference distribution, distinctive user tags, recent p-day interest preferences, and posting preferences, where p is an integer greater than or equal to 56; when the user type is B, the feature information tags include: registration interests, professional information, followed opinion leader tags, recent w-day interest preferences, note category preferences, and preferred opinion leader tags, where w is an integer greater than or equal to 7; when the user type is C, the feature information tags include: posting preferences, preferred opinion leader tags, recent p-day interest preferences, and followed opinion leader tags; when the user type is D, the feature information tags include: followed opinion leader tags, preferred opinion leader tags, note category preferences, recent p-day interest preferences, and recent p-day interest preferences. After determining the feature information tags, the corresponding user feature information can be extracted from the user database based on the feature information tags.

[0074] 203: Input user characteristic information, platform historical hot information, and upward trend information into the text generation model to generate paradigm text.

[0075] In this embodiment, the text generation model is determined by the user type. For example, the text generation model can be a piece of tagged text, where the tags in the text correspond one-to-one with tags in user characteristic information, platform historical hot topic information, and upward trend information. Therefore, by filling in one of the user characteristic information, platform historical hot topic information, or upward trend information into the corresponding tag location in the text, the corresponding paradigm text can be generated.

[0076] Specifically, the text generation model for type A users is as follows:

[0077] I'm a content creator on the internet platform [Platform Name], and I'd like your help in brainstorming content creation ideas. I'm a content creation KOL, and my followers primarily prefer content related to [the top three content preferences of my followers]. My interests over the past 7 days have mainly focused on [the top three interests of the past 7 days]. My usual content posting preferences are [the top three content preferences]. Today's trending topics on Xiaohongshu are [the top three trending tags matching my posting preferences on relevant platforms]. Based on this information, what are some good content ideas?

[0078] The text generation model for Type C users is as follows:

[0079] I'm a content creator on the internet platform [Platform Name], and I'd like your help in brainstorming content creation ideas. My profession is [Industry Information], and my interests when registering on the platform were [Top Three Interests Selected During Platform Registration]. I primarily follow content creators whose content aligns with [Top Three Follower Preferences]. Over the past 7 days, my interests have mainly focused on [Recent 7 Days' Interests]. I mainly view articles and videos related to [Top Three Consumer Content Preferences]. Today's trending topics on Xiaohongshu are [Top Three Trending Topics on the Platform]. Based on this information, what are some good content creation ideas?

[0080] The part in brackets 【】 represents the tags in the text.

[0081] 204: Input the paradigm text into the Chat Generative Pre-trained Transformer (ChatGPT) and receive the reply text from the Chat Generative Pre-trained Transformer.

[0082] 205: Extract information from the paradigm text to obtain at least one recommended creative.

[0083] In this embodiment, information can be extracted from the paradigm text to obtain at least one first creative idea. Then, feature extraction is performed on each of the at least one first creative ideas to obtain at least one first feature corresponding to each of the at least one first creative idea. Finally, each of the at least one first feature is compared with the historical creative features in the first works library, and at least one recommended creative idea is determined from the at least one first creative idea based on the comparison results. The first works library is a set of works whose play count / follower count is in the bottom 20% of all the user's creative works.

[0084] Specifically, the view count of user-generated content based on intelligent creative features can be monitored. Works with view count / follower count (reflecting content popularity) in the bottom 20% can be collected to create a primary work library for that user. Simultaneously, historical creative features are generated by recording the creative generation process of works in the primary work library, including user information, semantic paradigm input parameters, and keywords used to generate the creative, and then generating feature vectors.

[0085] In this embodiment, by calculating the similarity between each first feature and historical creative features in the first works library, the maximum value of the calculated similarity is taken as the creative value of each first feature, thus obtaining at least one creative value corresponding to at least one first creative. That is, the degree of similarity between each first creative and historical creatives in the first works library is determined, and the maximum similarity value is taken as the creative value of the first creative corresponding to each first feature.

[0086] Specifically, similarity can be expressed by formula ②:

[0087]

[0088] Where d(x,y) represents the similarity between each first feature x and the historical creative feature y in the first work library, xi represents the i-th element in each first feature x, yi represents the i-th element in the historical creative feature y in the first work library, h represents the total number of elements in each first feature x, and i is an integer greater than or equal to 1.

[0089] Then, in this embodiment, at least one recommended creative can be determined from at least one first creative based on at least one creative value. Specifically, the creative value of each of the at least one recommended creative is greater than a preset threshold.

[0090] In an optional implementation, for creatives with a similarity less than or equal to a preset threshold, ChatGPT can be proactively prompted to change the creative, thus avoiding different users from obtaining similar creatives with poor content quality.

[0091] Specifically, users can also proactively request changes to the creative concept, as follows:

[0092] (1) Satisfied with the idea but want to see more ideas: The user enters a command to change the idea. For example, when the method is running in the software, the user can click the "Change" button provided in the software to add a "Change Idea" related prompt to ChatGPT. For example, the user enters prompt in ChatGPT: Please give me N more ideas to get a reply containing other creative solutions.

[0093] (2) Want more creative ideas to be more divergent and fun: Users can input more creative instructions. For example, when the method is running in the software, they can click the "Dive More into Creativity" button provided in the software to add "Enhance Creativity" related prompts and submit them to ChatGPT. For example, inputting prompt in ChatGPT: Your answer is not creative enough. Please give me N more interesting and imaginative ideas to get a reply containing more creative solutions.

[0094] (3) To make creative ideas more relevant to personal information: Users input creative instructions related to themselves. For example, when this method is running in the software, users can click the "More Relevant to Me" button provided in the software to add "Relevant to Me" prompts to ChatGPT. For example, inputting a prompt to ChatGPT: Please pay attention to my personal information and give me N more relevant content ideas. At the same time, the key feature information is re-filtered, and the content of the "Platform Hot Information" related tags in the text generation model is deleted, only reflecting personal and personal related information. At the same time, the user's secondary interest tags are additionally obtained and updated to the semantic paradigm variables, including but not limited to secondary tags of note category preference, secondary tags of posting preference, etc. The content of secondary tags is more detailed than that of primary tags. For example, "Sports" is a primary tag, "Sports-Football" is a secondary tag, and "Sports-Football-Manchester United" is a tertiary tag. This step can make the input closer to the user's actual vertical direction. Then, the regenerated text is input into ChatGPT to obtain more creative ideas related to oneself.

[0095] 206: Recommend at least one creative idea to the user.

[0096] In this implementation, at least one recommended creative can be segmented into Chinese words, and the keywords in each recommended creative can be identified through word frequency statistics. The keywords of each recommended creative are matched with platform topic keywords to determine the corresponding platform topic tag and identifier for each recommended creative. Simultaneously, different colors are matched according to the popularity of the platform topic tags, using red (most popular), orange-red (medium popularity), and pink (low popularity) to represent the weight of each recommended creative. The creative recommendation page displays highly popular videos related to different creatives on the platform, and users can also click on the identifier to jump to the relevant topic page for more creative references.

[0097] In summary, the creative recommendation method provided by this invention designs different style text generation models for different types of users, and then generates different style texts based on user characteristic information, platform historical hot topic information, and upward trend information. Then, the style text is input into a chat generation pre-trained converter to obtain reply text containing creative ideas through dialogue. Finally, information is extracted from the reply text to obtain creative ideas for recommendation to users. This achieves automatic generation and recommendation of creative ideas, saving significant manpower and operating costs. Furthermore, the generated creative ideas, based on user characteristic information, platform historical hot topic information, and upward trend information, resonate with users while ensuring a certain level of promotion and attention.

[0098] See Figure 3 , Figure 3 This is a block diagram illustrating the functional modules of a creative recommendation device provided for an embodiment of this application. For example... Figure 3 As shown, the creative recommendation device 300 includes:

[0099] Analysis module 301 is used to determine the user type based on the user profile;

[0100] The extraction module 302 is used to extract historical hot topics, upward trend information, and user characteristic information of the platform based on user type and current time.

[0101] The question-and-answer module 303 is used to input user characteristic information, platform historical hot information and upward trend information into the text generation model to generate paradigm text. The text generation model is determined by the user type. The paradigm text is then input into the chat generation pre-trained converter and the module receives the reply text from the chat generation pre-trained converter.

[0102] The extraction module 302 is also used to extract information from the paradigm text to obtain at least one recommended idea;

[0103] The recommendation module 304 is used to recommend at least one creative idea to the user.

[0104] In an embodiment of the present invention, in determining the user type based on the user profile, the analysis module 301 is specifically used for:

[0105] Based on user profiles, determine whether a user is an opinion leader.

[0106] If the user is an opinion leader, then the user type is determined to be type A; otherwise, the user profile is used to determine whether the user's registration time is less than n days, where n is an integer greater than or equal to 7.

[0107] If a user's registration time is less than n days, then the user type is determined to be type B. Otherwise, based on the user profile, determine whether the number of notes published by the user is greater than m and whether the average note publishing frequency is greater than q per week, where m is an integer greater than or equal to 5 and q is an integer greater than or equal to 1.

[0108] If a user publishes more than m notes and their average note publishing frequency is greater than q notes per week, then the user type is determined to be type C; otherwise, the user type is determined to be type D.

[0109] In an embodiment of the present invention, the extraction module 302 is specifically used for extracting historical hotspot information, upward trend information, and user characteristic information of the platform based on user type and current time, in order to:

[0110] Determine the extraction duration and feature information tags based on user type;

[0111] Using the current time as the endpoint, extract historical data of the same duration as the extracted time from the platform's historical database;

[0112] Analyze historical data to identify historical hot topics and upward trends on the platform;

[0113] Based on the feature information tags, extract user feature information from the user database.

[0114] In an embodiment of the present invention, when the user type is type A, the feature information tags include: opinion leader fan category preference distribution, characteristic group tags, interest point preference in the past p days, and posting preference, where p is an integer greater than or equal to 56;

[0115] When the user type is B, the feature information tags include: registration interests, occupation information, followed opinion leader tags, past day interest preferences, note category preferences, and preferred opinion leader tags, where w is an integer greater than or equal to 7;

[0116] When the user type is C, the feature information tags include: posting preferences, preferred opinion leader tags, recent p-day interest preferences, and followed opinion leader tags;

[0117] When the user type is D, the feature information tags include: followed opinion leader tags, preferred opinion leader tags, note category preferences, interest preferences in the past p days, and interest preferences in the past p days.

[0118] In an embodiment of the present invention, in extracting information from paradigm text to obtain at least one recommended creative idea, the extraction module 302 is specifically used for:

[0119] Information is extracted from the paradigm text to obtain at least one first idea;

[0120] For each of the at least one first creative ideas, feature extraction is performed to obtain at least one first feature, wherein the at least one first feature corresponds one-to-one with the at least one first creative idea;

[0121] Each of the at least one first feature is compared with the historical creative features in the first work library. Based on the comparison results, at least one recommended creative is determined from at least one first creative. The first work library is the collection of all the user's creative works whose play count / fan count is in the bottom 20%.

[0122] In an embodiment of the present invention, the extraction module 302, which compares each of the at least one first feature with historical creative features in the first work library, and determines at least one recommended creative aspect from the at least one first creative based on the comparison result, is specifically used for:

[0123] Calculate the similarity between each first feature and the historical creative features in the first work library, and take the maximum value of the calculated similarity as the creative value of each first feature to obtain at least one creative value, wherein at least one creative value corresponds one-to-one with at least one first creative.

[0124] At least one recommended creative is determined from at least one first creative based on at least one creative value, wherein the creative value of each of the at least one recommended creative is greater than a preset threshold.

[0125] In an embodiment of the present invention, similarity can be expressed by formula ③:

[0126]

[0127] Where d(x,y) represents the similarity between each first feature x and the historical creative feature y in the first work library, xi represents the i-th element in each first feature x, yi represents the i-th element in the historical creative feature y in the first work library, h represents the total number of elements in each first feature x, and i is an integer greater than or equal to 1.

[0128] See Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided for an embodiment of this application. For example... Figure 4 As shown, the electronic device 400 includes a transceiver 401, a processor 402, and a memory 403. These are connected via a bus 404. The memory 403 stores computer programs and data, and can transfer data stored in the memory 403 to the processor 402.

[0129] Processor 402 is used to read the computer program in memory 403 and perform the following operations:

[0130] Determine the user type based on user profiles;

[0131] Based on user type and current time, extract historical hot topics, upward trend information, and user characteristic information from the platform;

[0132] User characteristic information, platform historical hot information, and upward trend information are input into the text generation model to generate paradigm text. The text generation model is determined by the user type.

[0133] Input the paradigm text into the chat-generated pre-trained converter and receive the reply text from the chat-generated pre-trained converter;

[0134] Extract information from the paradigm text to obtain at least one recommended creative;

[0135] Recommend at least one creative idea to the user.

[0136] In an embodiment of the present invention, in determining the user type based on a user profile, the processor 402 is specifically configured to perform the following operations:

[0137] Based on user profiles, determine whether a user is an opinion leader.

[0138] If the user is an opinion leader, then the user type is determined to be type A; otherwise, the user profile is used to determine whether the user's registration time is less than n days, where n is an integer greater than or equal to 7.

[0139] If a user's registration time is less than n days, then the user type is determined to be type B. Otherwise, based on the user profile, determine whether the number of notes published by the user is greater than m and whether the average note publishing frequency is greater than q per week, where m is an integer greater than or equal to 5 and q is an integer greater than or equal to 1.

[0140] If a user publishes more than m notes and their average note publishing frequency is greater than q notes per week, then the user type is determined to be type C; otherwise, the user type is determined to be type D.

[0141] In an embodiment of the present invention, in extracting historical hotspot information, upward trend information, and user characteristic information of the platform based on user type and current time, the processor 402 is specifically configured to perform the following operations:

[0142] Determine the extraction duration and feature information tags based on user type;

[0143] Using the current time as the endpoint, extract historical data of the same duration as the extracted time from the platform's historical database;

[0144] Analyze historical data to identify historical hot topics and upward trends on the platform;

[0145] Based on the feature information tags, extract user feature information from the user database.

[0146] In an embodiment of the present invention, when the user type is type A, the feature information tags include: opinion leader fan category preference distribution, characteristic group tags, interest point preference in the past p days, and posting preference, where p is an integer greater than or equal to 56;

[0147] When the user type is B, the feature information tags include: registration interests, occupation information, followed opinion leader tags, past day interest preferences, note category preferences, and preferred opinion leader tags, where w is an integer greater than or equal to 7;

[0148] When the user type is C, the feature information tags include: posting preferences, preferred opinion leader tags, recent p-day interest preferences, and followed opinion leader tags;

[0149] When the user type is D, the feature information tags include: followed opinion leader tags, preferred opinion leader tags, note category preferences, interest preferences in the past p days, and interest preferences in the past p days.

[0150] In an embodiment of the present invention, in extracting information from paradigm text to obtain at least one recommended idea, processor 402 is specifically configured to perform the following operations:

[0151] Information is extracted from the paradigm text to obtain at least one first idea;

[0152] For each of the at least one first creative ideas, feature extraction is performed to obtain at least one first feature, wherein the at least one first feature corresponds one-to-one with the at least one first creative idea;

[0153] Each of the at least one first feature is compared with the historical creative features in the first work library. Based on the comparison results, at least one recommended creative is determined from at least one first creative. The first work library is the collection of all the user's creative works whose play count / fan count is in the bottom 20%.

[0154] In an embodiment of the present invention, the processor 402 is specifically configured to perform the following operations: comparing each of at least one first feature with historical creative features in a first creative library, and determining at least one recommended creative aspect from at least one first creative based on the comparison results:

[0155] Calculate the similarity between each first feature and the historical creative features in the first work library, and take the maximum value of the calculated similarity as the creative value of each first feature to obtain at least one creative value, wherein at least one creative value corresponds one-to-one with at least one first creative.

[0156] At least one recommended creative is determined from at least one first creative based on at least one creative value, wherein the creative value of each of the at least one recommended creative is greater than a preset threshold.

[0157] In an embodiment of the present invention, similarity can be expressed by formula ④:

[0158]

[0159] Where d(x,y) represents the similarity between each first feature x and the historical creative feature y in the first work library, xi represents the i-th element in each first feature x, yi represents the i-th element in the historical creative feature y in the first work library, h represents the total number of elements in each first feature x, and i is an integer greater than or equal to 1.

[0160] It should be understood that the creative recommendation device in this application may include smartphones (such as Android phones, iOS phones, Windows Phones, etc.), tablet computers, PDAs, laptops, mobile internet devices (MIDs), robots, or wearable devices, etc. The above-mentioned creative recommendation devices are merely examples and not exhaustive, and include, but are not limited to, the creative recommendation devices described above. In practical applications, the above-mentioned creative recommendation devices may also include: intelligent in-vehicle terminals, computer equipment, etc.

[0161] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software combined with a hardware platform. Based on this understanding, all or part of the technical solution of the present invention that contributes to the background art can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of the present invention.

[0162] Therefore, embodiments of this application also provide a computer-readable storage medium storing a computer program that is executed by a processor to implement some or all of the steps of any of the creative recommendation methods described in the above method embodiments. For example, the storage medium may include a hard disk, floppy disk, optical disk, magnetic tape, magnetic disk, USB flash drive, flash memory, etc.

[0163] This application also provides a computer program product comprising a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of any of the creative recommendation methods described in the above method embodiments.

[0164] It should be noted that, for the sake of simplicity, the aforementioned methods are described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are optional, and the actions and modules involved are not necessarily essential to this application.

[0165] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0166] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0167] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0168] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software program module.

[0169] If the integrated unit is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0170] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0171] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A creative recommendation method, characterized in that, The method includes: Determine the user type based on user profiles; Based on the user type and the current time, extract historical hot topic information, upward trend information, and user characteristic information of the user from the platform; The user characteristic information, the platform's historical hot topic information, and the upward trend information are input into the text generation model to generate paradigm text, wherein the text generation model is determined by the user type; Input the paradigm text into the chat generation pre-trained converter, and receive the reply text from the chat generation pre-trained converter; Extracting information from the reply text to obtain at least one recommended creative includes: extracting information from the reply text to obtain at least one first creative; extracting features from each of the at least one first creative to obtain at least one first feature, wherein each of the at least one first feature corresponds one-to-one with the at least one first creative; comparing each of the at least one first feature with historical creative features in a first works library, and determining the at least one recommended creative from the at least one first creative based on the comparison results, wherein the first works library is a collection of all the user's creative works whose play count / fan count falls within a preset range; The at least one recommended idea is recommended to the user.

2. The method according to claim 1, characterized in that, The process of determining the user type based on the user profile includes: Based on the user profile, determine whether the user is an opinion leader. If the user is the opinion leader user, then the user type is determined to be type A; otherwise, based on the user profile, it is determined whether the user's registration time is less than n days, where n is an integer greater than or equal to 7. If the user's registration time is less than n days, then the user type is determined to be type B; otherwise, based on the user profile, it is determined whether the number of notes published by the user is greater than m and whether the average note publishing frequency is greater than q per week, where m is an integer greater than or equal to 5 and q is an integer greater than or equal to 1. If the number of notes published by the user is greater than m and the average note publishing frequency is greater than q per week, then the user type is determined to be type C; otherwise, the user type is determined to be type D.

3. The method according to claim 2, characterized in that, The step of extracting historical hot topic information, upward trend information, and user characteristic information of the user based on the user type and current time includes: Based on the user type, determine the extraction duration and feature information tags; Using the current time as the endpoint, extract historical data from the platform's historical database that has the same duration as the extraction time. The historical data is analyzed to determine the platform's historical hotspot information and upward trend information; Based on the feature information tags, the user feature information is extracted from the user database.

4. The method according to claim 3, characterized in that, When the user type is type A, the feature information tags include: opinion leader fan category preference distribution, characteristic group tags, interest point preference in the past p days, and posting preference, where p is an integer greater than or equal to 56; When the user type is type B, the feature information tags include: registration interests, occupation information, followed opinion leader tags, interest preferences in the past w days, note category preferences, and preferred opinion leader tags, where w is an integer greater than or equal to 7; When the user type is type C, the feature information tags include: the posting preference, the preference opinion leader tag, the interest point preference in the past p days, and the opinion leader tag being followed; When the user type is type D, the feature information tags include: the tags of the opinion leaders followed, the tags of the opinion leaders preferred, the preferences for note categories, and the preferences for points of interest in the past p days.

5. The method according to any one of claims 1-4, characterized in that, The first work library is a collection of works from all of the user's creative works that rank in the bottom 20% in terms of play count / fan count.

6. The method according to claim 5, characterized in that, The step of comparing each of the at least one first feature with historical creative features in the first work database, and determining the at least one recommended creative from the at least one first creative based on the comparison results, includes: Calculate the similarity between each first feature and the historical creative features in the first work library, and take the maximum value of the calculated similarity as the creative value of each first feature to obtain at least one creative value, wherein the at least one creative value corresponds one-to-one with the at least one first creative; The at least one recommended creative is determined from the at least one first creative based on the at least one creative value, wherein the creative value of each of the at least one recommended creative is greater than a preset threshold.

7. The method according to claim 6, characterized in that, The similarity satisfies the following formula: Where d(x,y) represents the similarity between each first feature x and the historical creative feature y in the first work library, xi represents the i-th element in each first feature x, yi represents the i-th element in the historical creative feature y in the first work library, h represents the total number of elements in each first feature x, and i is an integer greater than or equal to 1.

8. A creative recommendation device, characterized in that, The device includes: The analysis module is used to determine the user type based on the user profile; The extraction module is used to extract historical hot topic information, upward trend information, and user characteristic information of the user based on the user type and the current time. The question-and-answer module is used to input the user characteristic information, the platform's historical hot topic information, and the upward trend information into the text generation model to generate paradigm text, wherein the text generation model is determined by the user type, and input the paradigm text into the chat generation pre-trained converter to receive the reply text from the chat generation pre-trained converter; The extraction module is further configured to extract information from the reply text to obtain at least one recommended creative, including: extracting information from the reply text to obtain at least one first creative; extracting features from each of the at least one first creative to obtain at least one first feature, wherein the at least one first feature corresponds one-to-one with the at least one first creative; comparing each of the at least one first feature with historical creative features in a first work library, and determining the at least one recommended creative from the at least one first creative based on the comparison results, wherein the first work library is a collection of all the user's creative works whose play count / fan count falls within a preset range; The recommendation module is used to recommend at least one creative idea to the user.

9. An electronic device, characterized in that, The method includes a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the one or more programs include instructions for performing the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement the method as described in any one of claims 1-7.

Images