Video content creation method and device based on new media data analysis

Abstract

The invention discloses a video content creation method and device based on new media data analysis, and a server side method comprises the steps: receiving a video content creation request sent by a client side, wherein the video content creation request carries a content keyword; through a preset BI system, retrieving new media accounts related to the content keywords and comment area contents corresponding to videos published by the new media accounts as new media data, and retrieving e-commerce platform data corresponding to products related to the content keywords; generating user preference information and market trend information corresponding to the content keywords according to the new media data and the e-commerce platform data; and generating a target video based on the user preference information and the market trend information, and sending the target video to the client for playing. Therefore, by adopting the embodiment of the invention, the limitation of the traditional manual creation and single automatic editing technology is overcome, so that the automation, individuation and high-quality output of video content creation are realized.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and device for creating video content based on new media data analysis. Background Art

[0002] In the new media era, short video platforms have become the main channel for the public to obtain information and entertainment. Enterprises and individual creators are faced with the challenge of attracting audiences, increasing brand exposure and user engagement through video content. Therefore, efficiently and accurately creating video content that meets the preferences of the target audience has become a key task in the field of new media marketing.

[0003] In the related art, the creation of video content mainly relies on manual editing and the creativity of the creator. This process usually involves market research and the collection of audience feedback to determine the theme and style of the video. In addition, the existing automatic editing technology can perform some single functions, such as adding special effects. The process of manually creating video content is time-consuming and labor-intensive. The automatic editing technology has a single function and cannot fully analyze user preferences, which limits the ability to create high-quality video content. Summary of the invention

[0004] The embodiments of the present application provide a method and device for creating video content based on new media data analysis. In order to have a basic understanding of some aspects of the disclosed embodiments, a simple summary is given below. This summary is not a general review, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a preface to the detailed description that follows.

[0005] In a first aspect, an embodiment of the present application provides a video content creation method based on new media data analysis, which is applied to a server, and the method includes: Receive a video content creation request sent by a client, where the video content creation request carries content keywords; Through the preset BI system, new media accounts related to content keywords and the comments corresponding to each video published by each new media account are retrieved as new media data, and e-commerce platform data corresponding to products related to content keywords are retrieved; Generate user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data; user preference information is generated through a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area; the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data; the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment labels; the market trend information is generated based on product sales information and product sales reviews contained in the e-commerce platform data; Based on user preference information and market trend information, the target video is generated and sent to the client for playback.

[0006] In a second aspect, an embodiment of the present application provides a video content creation device based on new media data analysis, the device comprising: A request receiving module, used to receive a video content creation request sent by a client, where the video content creation request carries content keywords; A data retrieval module is used to retrieve new media accounts related to content keywords and the content of the comment area corresponding to each video published by each new media account as new media data through a preset BI system, and to retrieve e-commerce platform data corresponding to products related to content keywords; An information generation module is used to generate user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data; the user preference information is generated through a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area, the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data, the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment labels, and the market trend information is generated based on product sales information and product sales reviews contained in the e-commerce platform data; The video generation module is used to generate target videos based on user preference information and market trend information, and send them to the client for playback.

[0007] The technical solution provided by the embodiments of the present application may have the following beneficial effects: In the embodiments of the present application, on the one hand, by presetting the BI system and the content keywords carried by the video content creation request, the relevant new media accounts and comment area content, as well as the e-commerce platform data, can be automatically retrieved. The automatically retrieved data can provide a data basis for video creation, thereby overcoming the time-consuming and labor-intensive problem of manually creating videos. On the other hand, through the pre-trained emotional tendency recognition model, the system can comprehensively analyze user preferences. This model is trained with positive emotional data, negative emotional data, and neutral emotional data. These data come from social media comments and product reviews with different emotional tendencies. Therefore, the system can identify the preferences of different user groups by analyzing the emotional tendencies in user comments, thereby customizing personalized video content for different users. At the same time, the system combines user preferences and market trend information to generate target videos. These videos can not only meet user needs, but also adapt to market changes and improve the attractiveness and competitiveness of videos.

[0008] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0010] Figure 1 This is a method flow diagram of a video content creation method based on new media data analysis provided by an embodiment of the present application; Figure 2 It is a user interface diagram displayed by a client provided in an embodiment of the present application; Figure 3 This is a schematic diagram of interaction between a server and a client provided in an embodiment of the present application; Figure 4 It is a flowchart of a method for training a sentiment tendency recognition model provided in an embodiment of the present application; Figure 5 It is a schematic diagram of a model architecture of a sentiment tendency recognition model provided in an embodiment of the present application; Figure 6 It is a structural schematic diagram of a video content creation device based on new media data analysis provided by an embodiment of the present application; Figure 7 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The following description and the accompanying drawings sufficiently illustrate specific embodiments of the present application to enable those skilled in the art to practice them.

[0012] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0013] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0014] In the description of the present application, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0015] Currently, the creation of video content mainly relies on human editing and the creativity of creators. This process usually involves market research and the collection of audience feedback to determine the theme and style of the video. In addition, existing automatic editing technologies are able to perform some single functions, such as adding special effects.

[0016] The inventors realized that the process of manually creating video content is time-consuming and labor-intensive. Automatic editing technology has a single function and cannot fully analyze user preferences, which limits the ability to create high-quality video content.

[0017] In order to solve the above problems, the present application provides a video content creation method and device based on new media data analysis to solve the problems existing in the above-mentioned related technical problems. In the embodiment of the present application, on the one hand, by presetting the BI system and the content keywords carried by the video content creation request, the relevant new media accounts and comment area content, as well as the e-commerce platform data can be automatically retrieved. The automatically retrieved data can provide a data basis for video creation, thereby overcoming the time-consuming and labor-intensive problem of manually creating videos. On the other hand, through the pre-trained emotional tendency recognition model, the system can comprehensively analyze user preferences. This model is obtained by training positive emotional data, negative emotional data and neutral emotional data. These data come from social media comments and product reviews with different emotional tendencies. Therefore, the system can identify the preferences of different user groups by analyzing the emotional tendencies in user comments, thereby customizing personalized video content for different users. At the same time, the system combines user preferences and market trend information to generate target videos. These videos can not only meet user needs, but also adapt to market changes and improve the attractiveness and competitiveness of videos. The following exemplary embodiments are used for detailed description.

[0018] The following will be combined with the attached Figure 1 -Attached Figure 5 , a video content creation method based on new media data analysis provided by an embodiment of the present application is introduced in detail. The method can be implemented by a computer program and can be run on a video content creation device based on new media data analysis based on the von Neumann system. The computer program can be integrated into an application or run as an independent tool application.

[0019] See also Figure 1 , which is a flow chart of a video content creation method based on new media data analysis, applied to a server. Figure 1 As shown, the method of the embodiment of the present application may include the following steps: S101, receiving a video content creation request sent by a client, where the video content creation request carries content keywords; The client refers to the device or software used by the user, such as a mobile app, web browser, or professional video editing software, through which the user interacts with the server. Video content creation requests refer to requests sent by the client to the server to request the server to help create video content. Content keywords refer to words related to video content, which are used to describe the theme, style, or other important features of the video.

[0020] In some embodiments of the present application, users use a video creation application on their mobile phones to enter the theme or keywords of the video they want to create. For example, they enter keywords such as "healthy diet", "exercise", "natural food", etc. These keywords will be used to guide the creation of video content. The application encapsulates the keywords into a video content creation request and sends it to the server. Among them, the user interface for users to enter keywords is as follows: Figure 2 shown.

[0021] S102, through a preset BI system, searching for new media accounts related to the content keywords and the content of the comment area corresponding to each video published by each new media account as new media data, and searching for e-commerce platform data corresponding to the products related to the content keywords; Among them, the preset BI system is a tool for analyzing and storing new media platform data and e-commerce platform data. New media accounts refer to personal or corporate accounts on new media platforms, such as Weibo, Douyin, Instagram, etc., which are used to publish content and interact with the audience. Comment area content refers to the comments of users under the videos posted by new media accounts. These comments reflect the audience's feedback and emotions on the video content. E-commerce platform data refers to data collected from e-commerce platforms, including product information, sales data, user reviews, etc.

[0022] In some embodiments of the present application, the BI system first collects data from new media platforms and e-commerce platforms. These data include account information, video content, comment area content, and product sales information and evaluations. The collected data is stored in the database of the BI system. These databases may use distributed database technologies such as Hadoop, HBase, or Cassandra to ensure data reliability, scalability, and high performance. For new media data, the BI system may extract text features such as keywords, themes, or emotional tendencies. For e-commerce platform data, the system may extract product features such as categories, prices, or user reviews.

[0023] For example, if the content keyword is "eco-friendly fashion", the BI system will calculate the similarity of this keyword with the new media accounts and video content stored in the database, as well as the similarity with related products on the e-commerce platform. Based on the similarity score, the BI system retrieves the most relevant new media accounts and videos, as well as related products on the e-commerce platform. The content of the comment area corresponding to each video is used as new media data, and the product features are used as e-commerce platform data.

[0024] In some embodiments of the present application, the specific process of data generation of the preset BI system is: obtaining each registered new media account from the new media platform; storing the comment area content corresponding to each new media account and each video it publishes; obtaining product description information of each product from the e-commerce platform; storing the product description information of each product and its e-commerce platform data, and using the database storing the above content as the preset BI system.

[0025] S103, generating user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data; the user preference information is generated through a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area, the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data, the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment labels, and the market trend information is generated based on product sales information and product sales evaluations contained in the e-commerce platform data; Among them, new media data includes content in multiple comment areas, and e-commerce platform data includes product sales information and product sales evaluations.

[0026] In some embodiments of the present application, the specific process of generating user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data includes: performing data cleaning and data segmentation processing on the content of each comment area to obtain the vocabulary sequence and phrase sequence corresponding to the content of each comment area; generating user preference information corresponding to the content keywords based on the vocabulary sequence and phrase sequence corresponding to the content of each comment area; counting the sales volume of products related to the content keywords from the product sales information contained in the e-commerce platform data; analyzing the changing trend of product sales over time according to time series; using the time series analysis method to analyze the changing trend to predict the product sales trend within a preset period of time in the future, the time series analysis method being the ARIMA model; predicting user demand information for the product based on the product sales evaluation contained in the e-commerce platform data; using the changing trend of product sales over time and the user demand information for the product as the market trend information corresponding to the content keywords.

[0027] Among them, data cleaning refers to removing or correcting erroneous, duplicated or incomplete information in the content of each comment area. Data segmentation processing is to divide the content of each comment area into separate words or phrases. Word sequence and phrase sequence refer to a series of words or phrases divided from the content of each comment area. Product sales information includes data such as product sales volume and sales revenue, which is used to analyze the market performance of products. Time series analysis is a statistical technique used to analyze data points arranged in chronological order to identify trends, seasonality and periodicity. The ARIMA (Autoregressive Integrated Moving Average) model is a statistical model used for time series forecasting. Product sales evaluation refers to users' evaluation of purchased products, which can provide additional information about product satisfaction and demand.

[0028] In an embodiment of the present application, user preference information is extracted from the comment area content of the new media platform, and product sales trends are counted and predicted from the product sales information of the e-commerce platform. Combined with the time series analysis of the ARIMA model, the system can accurately predict the product sales trend in the future. In addition, by analyzing the product sales evaluation, the system can predict the user's demand information for the product. This comprehensive information provides in-depth market trend insights for content keywords, allowing video content creators to produce video content that is more in line with market demand and user preferences, thereby improving the attractiveness and market competitiveness of the video.

[0029] In some embodiments of the present application, the specific process of generating user preference information corresponding to content keywords based on the vocabulary sequence and phrase sequence corresponding to the content of each comment area includes: inputting the vocabulary sequence and phrase sequence corresponding to the content of each comment area into a pre-trained sentiment tendency recognition model, and outputting multiple sentiment words corresponding to the content of each comment area and the strength of each sentiment word; calculating the sentiment intensity value of the content of each comment area based on the strength of each sentiment word and the number of sentiment words; performing topic analysis on the vocabulary sequence and phrase sequence corresponding to the content of each comment area, and extracting the tendency topics involved in the content of each comment area; identifying the user's preference for different topics based on the sentiment intensity value of the content of each comment area and the tendency topics involved in the content of each comment area, as the user preference information corresponding to the content keywords.

[0030] Among them, the sentiment tendency recognition model is a machine learning model used to analyze text data and identify the sentiment tendencies expressed in it, such as positive, negative or neutral. Sentiment words refer to words with emotional colors, and intensity indicates the degree or strength of the emotions of these sentiment words.

[0031] For example, the content of the comment area is input into the sentiment tendency recognition model. The model outputs the sentiment words in each comment, such as "satisfied" and "disappointed", as well as their respective strengths. For each comment, the sentiment intensity value is calculated based on the strength and number of sentiment words, such as 0.8 for positive sentiment intensity and 0.2 for negative sentiment intensity. The word sequence and phrase sequence of the comments are subjected to topic analysis to extract tendency themes, such as "sustainable development" and "product effect". Combining the sentiment intensity value and tendency themes, it is identified that users have a stronger positive preference for the theme of "sustainable development" and more negative feedback on "product effect". These analysis results are integrated to form user preference information about "environmentally friendly products" videos, which can be used to guide the creation and optimization of video content to better meet user needs and expectations.

[0032] S104, generating a target video based on the user preference information and the market trend information, and sending the target video to the client for playback.

[0033] In some embodiments of the present application, based on user preference information and market trend information, the specific process of generating a target video includes: extracting target audience characteristics related to content keywords from user preference information, the target audience characteristics include age characteristics, gender characteristics and interest characteristics; extracting product characteristics related to content keywords from market trend information, the product characteristics include product name, feature information, and advantage information; integrating the target audience characteristics and product characteristics into a semantically coherent description text to obtain a text prompt for prompting a preset large language model, the text prompt is used to guide the preset large language model to generate a product introduction text corresponding to the content keywords; inputting the text prompt into the preset large language model to output the product introduction text; the preset large language model is ChatGPT; retrieving video materials matching the content keywords from a preset video material library, the video materials include pictures, special effects, and music; using video editing software to integrate the pictures, special effects, music and product introduction text to obtain the target video.

[0034] Among them, target audience characteristics refer to specific attributes of the target audience, such as age, gender, and interests. Product features refer to specific attributes of the product, such as name, features, and advantages, which help distinguish the product and attract potential consumers. Semantically coherent description text refers to text that is semantically related and logically clear. ChatGPT is a large language model developed by OpenAI that is capable of conducting conversations and generating text. The video material library is a database that stores the materials required for video production, including pictures, special effects, and music. Target videos are video content customized according to specific target audiences and market trends.

[0035] For example, extract target audience characteristics from user preference information, such as "young adults", "health conscious", and "interested in technology products". Extract product characteristics from market trend information, such as "fitness tracker", "real-time heart rate monitoring", and "long battery life". Integrate target audience characteristics with product characteristics into description text, such as "Introducing a fitness tracker with real-time heart rate monitoring and long battery life for young health-conscious technology enthusiasts". Use the description text as a text prompt to guide ChatGPT to generate product introduction text. Input the text prompt into ChatGPT and output the product introduction text. Retrieve pictures, special effects, and music matching "fitness tracker" from the video material library. Use video editing software to integrate the retrieved pictures, special effects, music, and product introduction text to create a target video. Finally, a fitness tracker introduction video for young health-conscious technology enthusiasts is obtained, which combines attractive visual materials with attractive product introduction text generated by ChatGPT.

[0036] For example Figure 3 As shown, the server 110 can be a server, which can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms, such as a server device running a preset BI system. When it is necessary to create video content based on new media data analysis, the server 110 receives a video content creation request sent by the client 120, and the video content creation request carries content keywords; the server 110 retrieves new media accounts related to content keywords and the comment area content corresponding to each video posted by each new media account as new media data through the preset BI system, and retrieves e-commerce platform data corresponding to products related to content keywords; the server 110 generates user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data; the server 110 generates a target video based on user preference information and market trend information, and sends it to the client 120 for playback.

[0037] It should be noted that the client 120 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited thereto. The server 110 and the client 120 may be connected via Bluetooth, USB (Universal Serial Bus) or other communication connection methods, and the present invention does not limit this.

[0038] In the embodiments of the present application, on the one hand, by presetting the BI system and the content keywords carried by the video content creation request, the relevant new media accounts and comment area content, as well as the e-commerce platform data, can be automatically retrieved. The automatically retrieved data can provide a data basis for video creation, thereby overcoming the time-consuming and labor-intensive problem of manually creating videos. On the other hand, through the pre-trained emotional tendency recognition model, the system can comprehensively analyze user preferences. This model is trained with positive emotional data, negative emotional data, and neutral emotional data. These data come from social media comments and product reviews with different emotional tendencies. Therefore, the system can identify the preferences of different user groups by analyzing the emotional tendencies in user comments, thereby customizing personalized video content for different users. At the same time, the system combines user preferences and market trend information to generate target videos. These videos can not only meet user needs, but also adapt to market changes and improve the attractiveness and competitiveness of videos.

[0039] See also Figure 4 , which is a flow chart of a method for training a sentiment tendency recognition model according to an embodiment of the present application. Figure 4 As shown, the method of the embodiment of the present application may include the following steps: S201, collect social media comments and product reviews covering different fields and sentiment tendencies to obtain text data; Among them, comments or feedback posted by users on social media platforms on specific content (such as posts, videos, pictures, etc.). Feedback provided by consumers on purchased products or services on e-commerce platforms or review websites usually includes ratings and text comments. Sentiment tendency is the emotional direction expressed in the text, which is usually divided into positive, negative and neutral.

[0040] In some embodiments of the present application, comments and reviews are obtained from multiple social media platforms and e-commerce platforms. Data from different fields can be collected, including but not limited to technology products, health foods, fashion apparel, and online education services.

[0041] S202, sentiment annotating the collected text data to divide the text data into positive sentiment data, negative sentiment data, and neutral sentiment data; positive sentiment data is data annotated to represent the user's optimistic sentiment, negative sentiment data is data annotated to represent the user's negative sentiment, and neutral sentiment data is data annotated to represent the user's lack of sentiment; In some embodiments, the sentiment tendency of each comment is identified and marked, positive sentiment may be represented by "1", negative sentiment by "-1", and neutral sentiment by "0", thereby obtaining positive sentiment data, negative sentiment data, and neutral sentiment data.

[0042] S203, creating a sentiment tendency recognition model; For example Figure 5 As shown in the figure, the sentiment tendency recognition model includes a word embedding network, a sequence embedding layer, a position embedding layer, a self-attention layer, a feature fusion module and a loss function.

[0043] S204, inputting the positive emotion data, the negative emotion data and the neutral emotion data into the emotion tendency recognition model, and outputting the loss value of the model; In some embodiments of the present application, positive sentiment data, negative sentiment data and neutral sentiment data are input into a sentiment tendency recognition model, and the specific process of outputting the loss value of the model includes: performing data cleaning and text segmentation on the positive sentiment data, negative sentiment data and neutral sentiment data to obtain text data after segmentation; inputting the text data after segmentation into a word embedding network to capture the semantic relationship and context information between words for vocabulary conversion to obtain word vectors in a text sequence; inputting the word vectors in the text sequence into a sequence embedding layer to capture the order information and context relationship in the text sequence to obtain a sequence vector; obtaining the position information of the text data after segmentation, and inputting the obtained position information of the text data into a position embedding layer to capture the position information of the words in the text to obtain a position vector; inputting the sequence vector and the position vector into a self-attention layer to capture the relationship between different words to obtain global features; inputting the word vector, sequence vector, position vector and global features into a feature fusion module to integrate feature information at different levels to obtain a comprehensive feature vector; outputting the loss value of the model based on the comprehensive feature vector and the loss function.

[0044] Specifically, according to the comprehensive feature vector and the loss function, the specific process of outputting the loss value of the model includes: performing a linear transformation on the comprehensive feature vector to obtain a score vector for each emotion category; wherein the calculation formula of the score vector is: in, is the score vector, is the weight matrix, is the comprehensive feature vector, is the bias vector; the score vector of each emotion category is converted into a probability distribution to obtain the predicted emotion category probability distribution; the calculation formula of the probability distribution is:

[0045] in, It is The probability distribution of emotion categories, It is The score vector of sentiment categories, is the total number of sentiment categories, It is the sum of the score vectors of all sentiment categories; the predicted sentiment category probability distribution and the labeled sentiment label corresponding to the predicted sentiment category probability distribution are input into the loss function, and the loss value of the model is output.

[0046] Specifically, the loss function is:

[0047] in, is the sentiment tendency recognition model at time step The forward propagation result is Update the parameters for the model, The optimization goal is to maximize the set of parameters The loss value, is the predicted probability distribution of sentiment categories, is the labeled emotion label corresponding to the predicted emotion category probability distribution, is the probability distribution of all predicted emotion categories, Indicates the emotional label The quantitative value of is the logarithmic probability, are the original parameters of the model, are the updated model parameters, is the time step .

[0048] S205, when the loss value reaches the minimum, a pre-trained sentiment tendency recognition model is generated; or, when the loss value has not reached the minimum, the model loss value is forward propagated to update the model update parameters of the sentiment tendency recognition model, and the steps of inputting positive sentiment data, negative sentiment data and neutral sentiment data into the sentiment tendency recognition model are continued until the loss value reaches the minimum.

[0049] In the embodiments of the present application, on the one hand, by presetting the BI system and the content keywords carried by the video content creation request, the relevant new media accounts and comment area content, as well as the e-commerce platform data, can be automatically retrieved. The automatically retrieved data can provide a data basis for video creation, thereby overcoming the time-consuming and labor-intensive problem of manually creating videos. On the other hand, through the pre-trained emotional tendency recognition model, the system can comprehensively analyze user preferences. This model is trained with positive emotional data, negative emotional data, and neutral emotional data. These data come from social media comments and product reviews with different emotional tendencies. Therefore, the system can identify the preferences of different user groups by analyzing the emotional tendencies in user comments, thereby customizing personalized video content for different users. At the same time, the system combines user preferences and market trend information to generate target videos. These videos can not only meet user needs, but also adapt to market changes and improve the attractiveness and competitiveness of videos.

[0050] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0051] See also Figure 6 , which shows a schematic diagram of the structure of a video content creation device based on new media data analysis provided by an exemplary embodiment of the present application. The video content creation device based on new media data analysis can be implemented as all or part of an electronic device through software, hardware or a combination of both. The device 1 includes a request receiving module 10, a data retrieval module 20, an information generating module 30, and a video generating module 40.

[0052] The request receiving module 10 is used to receive a video content creation request sent by a client, where the video content creation request carries content keywords; The data retrieval module 20 is used to retrieve new media accounts related to the content keywords and the comment area content corresponding to each video published by each new media account as new media data through a preset BI system, and retrieve e-commerce platform data corresponding to products related to the content keywords; The information generation module 30 is used to generate user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data; the user preference information is generated by a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area, the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data, the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment labels, and the market trend information is generated based on product sales information and product sales evaluations contained in the e-commerce platform data; The video generation module 40 is used to generate a target video based on user preference information and market trend information, and send it to the client for playback.

[0053] It should be noted that the video content creation device based on new media data analysis provided in the above-mentioned embodiment only uses the division of the above-mentioned functional modules as an example when executing the video content creation method based on new media data analysis. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the video content creation device based on new media data analysis provided in the above-mentioned embodiment and the video content creation method based on new media data analysis belong to the same concept. The implementation process thereof is detailed in the method embodiment and will not be repeated here.

[0054] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0055] In the embodiments of the present application, on the one hand, by presetting the BI system and the content keywords carried by the video content creation request, the relevant new media accounts and comment area content, as well as the e-commerce platform data, can be automatically retrieved. The automatically retrieved data can provide a data basis for video creation, thereby overcoming the time-consuming and labor-intensive problem of manually creating videos. On the other hand, through the pre-trained emotional tendency recognition model, the system can comprehensively analyze user preferences. This model is trained with positive emotional data, negative emotional data, and neutral emotional data. These data come from social media comments and product reviews with different emotional tendencies. Therefore, the system can identify the preferences of different user groups by analyzing the emotional tendencies in user comments, thereby customizing personalized video content for different users. At the same time, the system combines user preferences and market trend information to generate target videos. These videos can not only meet user needs, but also adapt to market changes and improve the attractiveness and competitiveness of videos.

[0056] The present application also provides a computer-readable medium having program instructions stored thereon, which, when executed by a processor, implement the video content creation method based on new media data analysis provided by the above-mentioned various method embodiments.

[0057] The present application also provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the video content creation method based on new media data analysis of the above-mentioned various method embodiments.

[0058] See also Figure 7 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 7 As shown, the electronic device 1000 may include: at least one processor 1001 , at least one network interface 1004 , a user interface 1003 , a memory 1005 , and at least one communication bus 1002 .

[0059] The communication bus 1002 is used to realize the connection and communication between these components.

[0060] The user interface 1003 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0061] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0062] Among them, the processor 1001 may include one or more processing cores. The processor 1001 uses various interfaces and lines to connect various parts of the entire electronic device 1000, and executes various functions and processes data of the electronic device 1000 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1005, and calling data stored in the memory 1005. Optionally, the processor 1001 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 1001 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 1001, and it can be implemented separately through a chip.

[0063] Among them, the memory 1005 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 1005 includes a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1005 may also be optionally at least one storage system located away from the aforementioned processor 1001. As Figure 7 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a video content creation application based on new media data analysis.

[0064] exist Figure 7In the electronic device 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 1001 can be used to call the video content creation application based on new media data analysis stored in the memory 1005, and specifically perform the following operations: Receive a video content creation request sent by a client, where the video content creation request carries content keywords; Through the preset BI system, new media accounts related to content keywords and the comments corresponding to each video published by each new media account are retrieved as new media data, and e-commerce platform data corresponding to products related to content keywords are retrieved; Generate user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data; user preference information is generated through a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area; the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data; the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment labels; the market trend information is generated based on product sales information and product sales reviews contained in the e-commerce platform data; Based on user preference information and market trend information, the target video is generated and sent to the client for playback.

[0065] In one embodiment, when the processor 1001 generates user preference information and market trend information corresponding to content keywords based on new media data and e-commerce platform data, the processor 1001 specifically performs the following operations: Perform data cleaning and word segmentation on the content of each comment area to obtain the vocabulary sequence and phrase sequence corresponding to the content of each comment area; Generate user preference information corresponding to content keywords based on the vocabulary sequence and phrase sequence corresponding to the content of each comment area; From the product sales information contained in the e-commerce platform data, count the sales volume of products related to the content keywords; Analyze the trend of product sales over time by time series; Use time series analysis method to analyze the change trend to predict the product sales trend in the future preset period of time. The time series analysis method is ARIMA model; Predict user demand for products based on product sales reviews contained in e-commerce platform data; The changing trend of product sales over time and user demand information for the product are used as market trend information corresponding to the content keywords.

[0066] In one embodiment, when the processor 1001 generates user preference information corresponding to content keywords according to the vocabulary sequence and phrase sequence corresponding to each comment area content, the processor 1001 specifically performs the following operations: Input the word sequence and phrase sequence corresponding to each comment area content into the pre-trained sentiment tendency recognition model, and output multiple sentiment words corresponding to each comment area content and the strength of each sentiment word; Calculate the sentiment intensity value of each comment area content according to the intensity of each sentiment word and the number of sentiment words; Perform topic analysis on the word sequence and phrase sequence corresponding to the content of each comment area to extract the tendency theme involved in the content of each comment area; According to the sentiment intensity value of each comment area content and the tendency topics involved in each comment area content, the user's preference for different topics is identified as the user preference information corresponding to the content keywords.

[0067] In one embodiment, when the processor 1001 generates a pre-trained emotion tendency recognition model, the processor 1001 specifically performs the following operations: Collect social media comments and product reviews covering different fields and sentiment tendencies to obtain text data; The collected text data is emotionally annotated to divide the text data into positive emotional data, negative emotional data and neutral emotional data; positive emotional data is data annotated to represent the user's optimistic emotions, negative emotional data is data annotated to represent the user's negative emotions, and neutral emotional data is data annotated to represent the user's lack of emotions; Create a sentiment tendency recognition model; Input positive emotion data, negative emotion data and neutral emotion data into the emotion tendency recognition model and output the loss value of the model; When the loss value reaches the minimum, a pre-trained sentiment tendency recognition model is generated; alternatively, when the loss value does not reach the minimum, the model loss value is forward propagated to update the model update parameters of the sentiment tendency recognition model, and the steps of inputting positive sentiment data, negative sentiment data, and neutral sentiment data into the sentiment tendency recognition model are continued until the loss value reaches the minimum.

[0068] In one embodiment, when the processor 1001 inputs the positive emotion data, the negative emotion data, and the neutral emotion data into the emotion tendency recognition model and outputs the loss value of the model, the following operations are specifically performed: Perform data cleaning and text segmentation on positive sentiment data, negative sentiment data, and neutral sentiment data to obtain text data after segmentation; The segmented text data is input into the word embedding network to capture the semantic relationship and context information between words for word conversion, and obtain the word vector in the text sequence; The word vectors in the text sequence are input into the sequence embedding layer to capture the order information and contextual relationship in the text sequence and obtain the sequence vector; Obtaining the position information of the text data after word segmentation, and inputting the obtained position information of the text data into the position embedding layer to capture the position information of the vocabulary in the text and obtain the position vector; The sequence vector and position vector are input into the self-attention layer to capture the relationship between different words and obtain global features; The word vector, sequence vector, position vector and global feature are input into the feature fusion module to integrate feature information at different levels and obtain a comprehensive feature vector. According to the comprehensive feature vector and loss function, the loss value of the model is output.

[0069] In one embodiment, when the processor 1001 generates a target video based on the user preference information and the market trend information, the processor 1001 specifically performs the following operations: Extract target audience characteristics related to content keywords from user preference information, including age characteristics, gender characteristics, and interest characteristics; Extract product features related to content keywords from market trend information, including product name, feature information, and advantage information; Integrate the target audience characteristics and product characteristics into a semantically coherent description text to obtain a text prompt for prompting a preset large language model. The text prompt is used to guide the preset large language model to generate a product introduction text corresponding to the content keywords; Input the text prompt into the preset large language model and output the product introduction text; Retrieve video materials that match the content keywords from the preset video material library, including pictures, special effects, and music; Use video editing software to integrate pictures, special effects, music and product introduction text to obtain the target video.

[0070] In the embodiments of the present application, on the one hand, by presetting the BI system and the content keywords carried by the video content creation request, the relevant new media accounts and comment area content, as well as the e-commerce platform data, can be automatically retrieved. The automatically retrieved data can provide a data basis for video creation, thereby overcoming the time-consuming and labor-intensive problem of manually creating videos. On the other hand, through the pre-trained emotional tendency recognition model, the system can comprehensively analyze user preferences. This model is trained with positive emotional data, negative emotional data, and neutral emotional data. These data come from social media comments and product reviews with different emotional tendencies. Therefore, the system can identify the preferences of different user groups by analyzing the emotional tendencies in user comments, thereby customizing personalized video content for different users. At the same time, the system combines user preferences and market trend information to generate target videos. These videos can not only meet user needs, but also adapt to market changes and improve the attractiveness and competitiveness of videos.

[0071] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program for creating video content based on new media data analysis can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium of the program for creating video content based on new media data analysis can be a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0072] The above disclosure is only the preferred embodiment of the present application, which certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A video content creation method based on new media data analysis, characterized in that: Applied to the server, the method includes: Receive a video content creation request sent by a client, wherein the video content creation request carries a content keyword; Through the preset BI system, new media accounts related to the content keywords and the content of the comment area corresponding to each video published by each new media account are retrieved as new media data, and e-commerce platform data corresponding to the products related to the content keywords are retrieved; Generate user preference information and market trend information corresponding to the content keywords based on the new media data and the e-commerce platform data; the user preference information is generated by a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area, the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data, the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment tags, and the market trend information is generated based on product sales information and product sales evaluations contained in the e-commerce platform data; Based on the user preference information and market trend information, a target video is generated and sent to the client for playback.

2. The method according to claim 1, characterized in that: The new media data includes the contents of multiple comment areas, and the e-commerce platform data includes product sales information and product sales evaluations; The generating of user preference information and market trend information corresponding to the content keywords according to the new media data and the e-commerce platform data includes: Performing data cleaning and data segmentation processing on the content of each comment area to obtain a vocabulary sequence and a phrase sequence corresponding to the content of each comment area; Generating user preference information corresponding to the content keywords according to the vocabulary sequence and phrase sequence corresponding to the content of each comment area; Counting the sales volume of products related to the content keywords from the product sales information included in the e-commerce platform data; Analyze the sales trend of the product over time by time series; A time series analysis method is used to analyze the change trend to predict the product sales trend within a preset period of time in the future. The time series analysis method is an ARIMA model. Predicting user demand information for products based on product sales reviews contained in the e-commerce platform data; The change trend of the product sales volume over time and the user's demand information for the product are used as the market trend information corresponding to the content keyword.

3. The method according to claim 2, characterized in that Generating user preference information corresponding to the content keywords according to the vocabulary sequence and phrase sequence corresponding to the content of each comment area includes: Inputting the vocabulary sequence and phrase sequence corresponding to each comment area content into a pre-trained sentiment tendency recognition model, and outputting a plurality of sentiment words corresponding to each comment area content and the strength of each sentiment word; Calculate the sentiment intensity value of each comment area content according to the intensity of each sentiment word and the number of sentiment words; Performing topic analysis on the word sequence and phrase sequence corresponding to the content of each comment area, and extracting the tendency theme involved in the content of each comment area; According to the sentiment intensity value of the content in each comment area and the tendency topics involved in the content in each comment area, the user's preference for different topics is identified as the user preference information corresponding to the content keyword.

4. The method according to claim 3, characterized in that: Follow these steps to generate a pre-trained sentiment recognition model, including: Collect social media comments and product reviews covering different fields and sentiment tendencies to obtain text data; Performing sentiment annotation on the collected text data to divide the text data into positive sentiment data, negative sentiment data and neutral sentiment data; the positive sentiment data is data annotated to represent the user's optimistic sentiment, the negative sentiment data is data annotated to represent the user's negative sentiment, and the neutral sentiment data is data annotated to represent the user's lack of sentiment; Create a sentiment tendency recognition model; Inputting the positive emotion data, negative emotion data and neutral emotion data into the emotion tendency recognition model, and outputting the loss value of the model; When the loss value reaches the minimum, a pre-trained emotion tendency recognition model is generated; or, when the loss value does not reach the minimum, the model loss value is forward propagated to update the model update parameters of the emotion tendency recognition model, and the step of inputting the positive emotion data, negative emotion data and neutral emotion data into the emotion tendency recognition model is continued until the loss value reaches the minimum.

5. The method according to claim 4, characterized in that The sentiment tendency recognition model includes a word embedding network, a sequence embedding layer, a position embedding layer, a self-attention layer, a feature fusion module and a loss function; The step of inputting the positive emotion data, the negative emotion data and the neutral emotion data into the emotion tendency recognition model and outputting the loss value of the model comprises: Performing data cleaning and text segmentation on the positive sentiment data, the negative sentiment data, and the neutral sentiment data to obtain text data after segmentation; Inputting the segmented text data into the word embedding network to capture the semantic relationship and context information between words for word conversion, and obtaining word vectors in the text sequence; Inputting the word vectors in the text sequence into the sequence embedding layer to capture the order information and contextual relationship in the text sequence to obtain a sequence vector; Acquire the position information of the text data after word segmentation, and input the acquired position information of the text data into the position embedding layer to capture the position information of the vocabulary in the text and obtain the position vector; Inputting the sequence vector and the position vector into the self-attention layer to capture the relationship between different words and obtain a global feature; Inputting the word vector, sequence vector, position vector and global feature into the feature fusion module to integrate feature information at different levels to obtain a comprehensive feature vector; According to the comprehensive feature vector and the loss function, the loss value of the model is output.

6. The method according to claim 5, characterized in that Outputting the loss value of the model according to the comprehensive feature vector and the loss function includes: Performing a linear transformation on the comprehensive feature vector to obtain a score vector for each emotion category; Convert the score vector of each emotion category into a probability distribution to obtain a predicted emotion category probability distribution; The predicted emotion category probability distribution and the annotated emotion label corresponding to the predicted emotion category probability distribution are input into the loss function, and the loss value of the model is output.

7. The method according to claim 1, characterized in that The generating a target video based on the user preference information and the market trend information comprises: Extracting target audience characteristics related to the content keywords from the user preference information, the target audience characteristics including age characteristics, gender characteristics and interest characteristics; Extracting product features related to the content keywords from the market trend information, the product features including product name, feature information, and advantage information; Integrate the target audience characteristics and the product characteristics into a semantically coherent description text to obtain a text prompt for prompting a preset large language model, wherein the text prompt is used to guide the preset large language model to generate a product introduction text corresponding to the content keyword; Input the text prompt into a preset large language model and output a product introduction text; Retrieving video materials matching the content keywords from a preset video material library, wherein the video materials include pictures, special effects, and music; The image, special effects, music and product introduction text are integrated using video editing software to obtain a target video.

8. The method according to claim 1, characterized in that: Before receiving the video content creation request sent by the client, the method further includes: Get each registered new media account from the new media platform; Store the comment area content corresponding to each new media account and each video it publishes; Get product description information of each product from the e-commerce platform; Storing product description information and e-commerce platform data of each product; The database storing the above contents is used as the preset BI system.

9. A video content creation device based on new media data analysis, characterized in that: The device comprises: A request receiving module, used to receive a video content creation request sent by a client, wherein the video content creation request carries a content keyword; A data retrieval module is used to retrieve new media accounts related to the content keywords and the comment area content corresponding to each video published by each new media account as new media data through a preset BI system, and retrieve e-commerce platform data corresponding to products related to the content keywords; An information generation module is used to generate user preference information and market trend information corresponding to the content keywords based on the new media data and the e-commerce platform data; the user preference information is generated by a pre-trained sentiment tendency recognition model, a vocabulary sequence and a phrase sequence corresponding to the content of each comment area, the pre-trained sentiment tendency recognition model is obtained by training positive sentiment data, negative sentiment data and neutral sentiment data, the positive sentiment data, negative sentiment data and neutral sentiment data are obtained by collecting social media comments and product reviews covering different sentiment tendencies and annotating them with sentiment tags, and the market trend information is generated based on product sales information and product sales evaluations contained in the e-commerce platform data; The video generation module is used to generate a target video based on the user preference information and market trend information, and send the target video to the client for playback.

Images