Knowledge framework automatic generation method and device, computer equipment and storage medium

Abstract

The invention discloses a knowledge framework automatic generation method and device, computer equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: analyzing a target video to obtain a video frame sequence and a video voice segment; identifying key frames in the video frame sequence, and performing image feature extraction on the key frames to obtain video image features; converting the video voice segment into a voice text, and performing text feature extraction on the voice text to obtain voice text features; in a knowledge base system, performing knowledge association on the video image features and the voice text features to generate an associated knowledge graph; and obtaining associated knowledge tags from the associated knowledge graph, and combining the associated knowledge tags to construct a target knowledge framework. The invention also relates to the technical field of block chains, and the target video is stored in a block chain network. According to the method, the construction efficiency of the knowledge framework is improved, the dependence on manual intervention in the construction process is reduced, and the construction accuracy of the knowledge framework is improved.

Description

Technical Field

[0001] The present application belongs to the field of artificial intelligence technology, and specifically relates to a method, apparatus, computer equipment and storage medium for automatically generating a knowledge framework. Background Art

[0002] Knowledge framework refers to the classification, organization and arrangement of knowledge in a certain field to form a structured system, which can help people better understand and learn knowledge, thereby improving their ability to apply knowledge. Through classification and arrangement, the knowledge framework presents a clear hierarchical structure of knowledge, with logical connections between various parts, forming an organic whole, covering all aspects and levels of knowledge in a certain field, providing people with comprehensive learning resources and references, enabling people to systematically learn and master knowledge, and facilitating people's understanding and memory.

[0003] The current knowledge framework is usually built by staff in related fields based on their own knowledge reserves and needs. Take insurance training as an example. In the current insurance training process, insurance agents need to find relevant materials for expansion and mining, and then build a knowledge framework for insurance training based on their own knowledge reserves. However, this method is highly dependent on the staff's own knowledge reserves, and has low construction efficiency and poor accuracy. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a method, device, computer equipment and storage medium for automatically generating a knowledge framework, so as to solve the technical problems that the construction of the existing knowledge framework depends on the staff's own knowledge reserves, and has low construction efficiency and poor accuracy.

[0005] In order to solve the above technical problems, the present application embodiment provides a method for automatically generating a knowledge framework, which adopts the following technical solution:

[0006] A method for automatically generating a knowledge framework, comprising:

[0007] Obtain the target video, parse the target video, and obtain the video frame sequence and video voice clips;

[0008] Identify key frames in a video frame sequence, and extract image features from the key frames to obtain video image features;

[0009] Convert the video voice clip into speech text, and extract text features from the speech text to obtain speech text features;

[0010] In the preset knowledge base system, the video image features and the speech text features are associated with each other to generate an associated knowledge graph;

[0011] Obtain associated knowledge tags from the associated knowledge graph, combine the associated knowledge tags, and construct the target knowledge framework.

[0012] Furthermore, the step of identifying key frames in the video frame sequence and extracting image features from the key frames to obtain video image features specifically includes:

[0013] Identifying histogram data and grayscale image data of video frames in a video frame sequence, and determining key frames in the video frame sequence based on the histogram data and the grayscale image data;

[0014] Based on the preset image target recognition algorithm, the image target of the key frame is recognized, and the feature of the image target is extracted to obtain the video image feature.

[0015] Further, the step of identifying the histogram data and the grayscale image data of the video frames in the video frame sequence, and determining the key frames in the video frame sequence based on the histogram data and the grayscale image data specifically includes:

[0016] Calculate the histogram data and grayscale image data of two adjacent video frames in the video frame sequence;

[0017] Based on the histogram data and the grayscale image data, the weighted Euclidean distance between two adjacent video frames is calculated;

[0018] Determining the shot transition boundary of the video frame sequence based on the weighted Euclidean distance;

[0019] Key frames in a video frame sequence are determined based on shot transition boundaries.

[0020] Furthermore, the step of identifying the image target of the key frame based on a preset image target recognition algorithm and extracting features of the image target to obtain the video image features specifically includes:

[0021] The key frame is divided into several grid cells, where each grid cell is responsible for detecting an initial prediction target in the key frame;

[0022] Generate a bounding box for each grid cell and predict the confidence and class probability of the bounding box;

[0023] Based on the confidence of the bounding box and the probability of the category to which it belongs, the initial prediction target is screened to obtain the image target of the key frame;

[0024] Perform convolution feature extraction on image targets to obtain video image features.

[0025] Furthermore, the step of converting the video voice clip into voice text and extracting text features from the voice text to obtain voice text features specifically includes:

[0026] Perform audio-to-text processing on the video voice clip to obtain the voice text;

[0027] Perform word segmentation on the speech text to obtain text segmentation;

[0028] The text features are encoded for the text segmentation to obtain the speech text features.

[0029] Furthermore, in the preset knowledge base system, before the step of performing knowledge association on the video image features and the speech text features to generate the associated knowledge graph, the following steps are also included:

[0030] Perform feature normalization processing on video image features and speech text features;

[0031] Use the preset task classifier to classify the normalized video image features and speech text features to obtain the target task classification features;

[0032] In the preset knowledge base system, the steps of performing knowledge association on the video image features and the speech text features to generate the associated knowledge graph specifically include:

[0033] Perform feature vector transformation on the knowledge ontology in the knowledge base system to obtain knowledge feature vector;

[0034] In the knowledge base system, the Euclidean distance between the knowledge ontology and the target task classification features is calculated based on the knowledge feature vector and the target task classification feature vector;

[0035] Knowledge association is performed based on the Euclidean distance between the knowledge ontology and the target task classification features to generate an associated knowledge graph.

[0036] Furthermore, the steps of obtaining associated knowledge tags from the associated knowledge graph and combining the associated knowledge tags to construct a target knowledge framework specifically include:

[0037] By using a preset keyword matching algorithm, the approximate knowledge tags of the associated knowledge tags are calculated based on the distance priority principle;

[0038] Sort the approximate knowledge labels to obtain an approximate knowledge label sequence;

[0039] A knowledge framework template is obtained, and associated knowledge tags and approximate knowledge tag sequences are inserted into the knowledge framework template to generate a target knowledge framework.

[0040] In order to solve the above technical problems, the embodiment of the present application also provides a knowledge framework automatic generation device, which adopts the following technical solution:

[0041] A knowledge framework automatic generation device, comprising:

[0042] The video parsing module is used to obtain the target video and parse the target video to obtain the video frame sequence and the video voice clip;

[0043] An image feature module is used to identify key frames in a video frame sequence and extract image features from the key frames to obtain video image features;

[0044] A text feature module is used to convert the video voice clip into voice text and extract text features from the voice text to obtain voice text features;

[0045] The knowledge association module is used to associate the video image features and the speech text features in a preset knowledge base system to generate an associated knowledge graph;

[0046] The framework construction module is used to obtain associated knowledge tags from the associated knowledge graph and combine the associated knowledge tags to construct the target knowledge framework.

[0047] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0048] A computer device comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for automatically generating a knowledge framework as described in any one of the above are implemented.

[0049] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0050] A computer-readable storage medium having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by a processor, implement the steps of any one of the above-described methods for automatically generating a knowledge framework.

[0051] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0052] The present application discloses a method, device, computer equipment and storage medium for automatically generating a knowledge framework, and belongs to the field of artificial intelligence technology. The present application automatically identifies key frames and voice content in a target video by analyzing and processing the target video, extracts image features and text features, and uses a preset knowledge base system to associate these features with knowledge, generates an associated knowledge graph, and constructs a target knowledge framework by combining knowledge tags. Compared with the traditional knowledge framework construction method that relies on artificial knowledge reserves, the present application greatly improves the construction efficiency and reduces the reliance on manual intervention. At the same time, through automated feature extraction and knowledge association, the accuracy of knowledge framework construction is significantly improved, effectively solving the technical problems of low efficiency and poor accuracy of manual construction, and has a higher level of automation and intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 An exemplary system architecture diagram is shown in which the present application can be applied;

[0055] Figure 2 A flow chart showing an embodiment of a method for automatically generating a knowledge framework according to the present application;

[0056] Figure 3 A schematic diagram showing the structure of an embodiment of a knowledge framework automatic generation device according to the present application is shown;

[0057] Figure 4 A schematic structural diagram of an embodiment of a computer device according to the present application is shown. DETAILED DESCRIPTION

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0059] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0060] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0061] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0062] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0063] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to a laptop computer 1011, a tablet computer 1012 or a mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0064] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0065] It should be noted that the knowledge framework automatic generation method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the knowledge framework automatic generation device is generally set in the server / terminal device.

[0066] It should be understood that Figure 1 The numbers of terminal devices, networks and servers in the embodiment are merely illustrative. The above system may have any number of terminal devices, networks and servers according to implementation requirements.

[0067] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for automatically generating a knowledge framework according to the present application. The method for automatically generating a knowledge framework comprises the following steps:

[0068] S201, obtaining a target video, and parsing the target video to obtain a video frame sequence and a video voice segment.

[0069] Specifically, the target video can be an audio and video file from various multimedia resources, such as teaching videos, promotional videos, documentaries, etc. The system first uses multimedia processing technology to parse the input video file and extract its basic components, such as video frame sequences and audio clips. A video frame sequence is a collection of static images of a video, arranged in chronological order, and each frame represents the image state of the video at a certain moment. The audio clip is the sound signal in the video, including character dialogue, background music, narration, etc. The system separates these audio clips from the video frame sequence through video decoding technology. In addition, during the video parsing process, the system can also identify and adjust parameters such as video resolution and frame rate to ensure the efficiency and accuracy of subsequent processing.

[0070] S202, identifying key frames in the video frame sequence, and extracting image features from the key frames to obtain video image features.

[0071] Specifically, a video frame sequence contains a large number of repeated or unimportant frames. Therefore, in order to improve analysis efficiency, the system will use a key frame extraction algorithm to select representative or informative frames as key frames. Common key frame extraction methods include algorithms based on frame differences, motion information, or scene changes. These algorithms can automatically determine the degree of change in video content and select frames that appear at plot turning points, important actions, or scene changes. After identifying the key frames, the system extracts image features for each key frame and uses deep learning models or computer vision technology to analyze the image content in the key frames, such as scenes, characters, objects, colors, textures, etc. The extracted image features can be represented as high-dimensional vectors. Image feature extraction not only helps to understand the visual content of the video, but also provides visual feature information for the associated knowledge graph.

[0072] S203, converting the video voice clip into voice text, and performing text feature extraction on the voice text to obtain voice text features.

[0073] Specifically, the system converts the voice clips in the video into corresponding text through speech recognition technology (ASR, Automatic Speech Recognition). Speech recognition technology relies on acoustic models, language models and pronunciation dictionaries. It analyzes the speech features of the audio signal, such as frequency, pitch, rhythm, etc., identifies the speech content and generates a text representation. After conversion to text, the system further analyzes the text content and performs natural language processing (NLP, Natural Language Processing), including word segmentation, part-of-speech tagging, syntactic analysis and other steps. The process of text feature extraction not only includes the extraction of keywords, but also involves the understanding of text semantics, such as topic modeling, sentiment analysis, etc. In addition, the system can also recognize and translate multilingual speech in the video, and convert speech text into text in multiple languages ​​to enhance the applicability of the system. The obtained text features will be used to complete knowledge association with image features.

[0074] S204: In a preset knowledge base system, knowledge association is performed on video image features and speech text features to generate an associated knowledge graph.

[0075] Specifically, the knowledge base system is a pre-built database containing rich domain knowledge, usually represented in the form of a graph. In this system, knowledge is presented in the form of nodes (entities) and edges (relationships) to form a knowledge graph. After the video image features and speech text features are extracted, the system will match and associate these features with the existing knowledge in the knowledge base. First, the system will identify objects, scenes or characters in the video through image features, and match these visual elements with relevant entities in the knowledge base. Secondly, the system finds related knowledge nodes through keywords or semantic information extracted from text features, and forms knowledge links based on the semantic correlation between features. This process involves knowledge reasoning and knowledge fusion technology to ensure that the generated knowledge associations are reasonable and contextual. Through these associations, the system can construct a knowledge graph related to the video content, thereby providing a multi-dimensional knowledge interpretation for the video content.

[0076] S205, obtaining associated knowledge tags from the associated knowledge graph, and combining the associated knowledge tags to construct a target knowledge framework.

[0077] Specifically, the system first extracts the associated knowledge tags from the associated knowledge graph generated in the previous step. Knowledge tags are concise descriptions of nodes or edges in the graph, usually identifiers of concepts, objects, events or attributes. These knowledge tags help to simplify the complex relationships and information in the knowledge graph. The system screens out the knowledge tags that are most relevant to the video content by analyzing the nodes and connection relationships in the graph. For example, if there are tags such as "car", "transportation", and "driving" in the graph, the system will extract these tags as knowledge representations closely related to the video content. Next, the system combines and optimizes the knowledge tags based on their relevance and hierarchical structure to construct a target-oriented knowledge framework. This framework is a high-level abstraction of video information, integrating multimodal information (images, voice, etc.) in the video and its associated information with the knowledge base into a structured knowledge system. The knowledge framework finally generated can not only fully reflect the core information of the video, but also has logic and interpretability, which is convenient for subsequent application scenarios, such as automatic summarization, content retrieval, and intelligent question and answer. This step significantly improves the efficiency and accuracy of information extraction through automated knowledge integration and framework construction.

[0078] In the above embodiments, the present application automatically identifies key frames and voice content in the target video through analysis and processing of the target video, extracts image features and text features, uses a preset knowledge base system to associate these features with knowledge, generates an associated knowledge graph, and constructs a target knowledge framework by combining knowledge tags. Compared with the traditional knowledge framework construction method that relies on artificial knowledge reserves, the present application greatly improves the construction efficiency and reduces the reliance on manual intervention. At the same time, through automated feature extraction and knowledge association, the accuracy of knowledge framework construction is significantly improved, effectively solving the technical problems of low efficiency and poor accuracy of manual construction, and has a higher level of automation and intelligence.

[0079] Furthermore, the step of identifying key frames in the video frame sequence and extracting image features from the key frames to obtain video image features specifically includes:

[0080] Identifying histogram data and grayscale image data of video frames in a video frame sequence, and determining key frames in the video frame sequence based on the histogram data and the grayscale image data;

[0081] Based on the preset image target recognition algorithm, the image target of the key frame is recognized, and the feature of the image target is extracted to obtain the video image feature.

[0082] In this embodiment, the content of the frame is first analyzed by identifying the histogram data and grayscale data of each frame in the video frame sequence. The histogram data can reflect the statistical information of the distribution of different colors in the image, while the grayscale data provides the brightness distribution information of the image. The combination of the two can effectively detect the degree of change between video frames, help identify those frames that are significantly different from other frames, that is, key frames. By selecting key frames, the amount of data can be greatly reduced, and the important segments in the video can be analyzed intensively. Next, the image targets (such as people, objects, etc.) in these key frames are identified based on the preset image target recognition algorithm, and the features of the image targets are extracted. The image target recognition algorithm can use a deep learning model, such as the Yolo (You Only Look Once) target detection algorithm, a convolutional neural network (CNN), etc., to classify and detect objects. The feature extraction process generates a series of feature vectors describing the image targets, which are used for association and analysis, and can better describe the visual content in the video.

[0083] Through the above steps, the number of video frames to be processed is effectively reduced, the focus is on extracting representative information in the video, the efficiency and accuracy of feature extraction are improved, and the automatic processing capability of the system is enhanced.

[0084] Further, the step of identifying the histogram data and the grayscale image data of the video frames in the video frame sequence, and determining the key frames in the video frame sequence based on the histogram data and the grayscale image data specifically includes:

[0085] Calculate the histogram data and grayscale image data of two adjacent video frames in the video frame sequence;

[0086] Based on the histogram data and the grayscale image data, the weighted Euclidean distance between two adjacent video frames is calculated;

[0087] Determining the shot transition boundary of the video frame sequence based on the weighted Euclidean distance;

[0088] Key frames in a video frame sequence are determined based on shot transition boundaries.

[0089] In this embodiment, first, the system needs to calculate the histogram data and grayscale data of two adjacent video frames in the video frame sequence. The histogram data can reflect the color distribution of the image, and the grayscale data describes the brightness distribution of the image. These data can well capture the visual changes between video frames. Then, based on these histogram and grayscale data, the system calculates the weighted Euclidean distance between adjacent video frames. The weighted Euclidean distance is calculated by combining the differences of each channel in the histogram and the grayscale image, and can more comprehensively measure the similarity between two frames. The weighted method can adjust the weights of different features according to actual needs to ensure that the color and brightness information obtain appropriate weight distribution in the comparison. Through this calculation method, the degree of change between frames can be more accurately identified. Subsequently, the system determines the shot transition boundary in the video frame sequence based on the calculated weighted Euclidean distance. Shot transition is usually a sign of scene change in the video. The identification of shot boundaries helps to divide the video into multiple different segments. Finally, the system determines the key frame in the video based on these shot transition boundaries, that is, the frame that best represents the content of each shot.

[0090] Through the above steps, the shot transition points in the video can be efficiently and accurately identified, and key frames can be extracted based on this, which reduces the processing volume of the video frames while ensuring the complete extraction of important information.

[0091] Furthermore, the step of identifying the image target of the key frame based on a preset image target recognition algorithm and extracting features of the image target to obtain the video image features specifically includes:

[0092] The key frame is divided into several grid cells, where each grid cell is responsible for detecting an initial prediction target in the key frame;

[0093] Generate a bounding box for each grid cell and predict the confidence and class probability of the bounding box;

[0094] Based on the confidence of the bounding box and the probability of the category to which it belongs, the initial prediction target is screened to obtain the image target of the key frame;

[0095] Perform convolution feature extraction on image targets to obtain video image features.

[0096] In this embodiment, first, the system divides the key frame into several grid units, each grid unit corresponds to a different area in the key frame. Such grid division is to better locate and detect the initial predicted target in the key frame. Each grid unit will be responsible for detecting potential targets within its own range. This processing method is similar to the regional division strategy in target detection, which can effectively improve the detection accuracy and reduce the amount of calculation. For each grid unit, the system generates a bounding box, which is used to describe the specific position, size and shape of the target that may exist in the grid unit. Subsequently, the system predicts a confidence for each bounding box, which indicates the probability that the bounding box contains the target; at the same time, it also predicts the probability of the category to which the bounding box belongs, which indicates the possibility that the object detected in the grid belongs to a specific category. Based on the confidence and category probability of the bounding box, the system will screen the initial predicted target, that is, retain those targets with high confidence and clear category recognition. Next, the filtered image target is subjected to convolution feature extraction. The convolution operation further extracts the features (such as edges, textures, shapes, etc.) in the target area through a deep neural network (such as a convolutional neural network, CNN), and finally obtains a high-dimensional feature vector or feature map.

[0097] Specifically, Yolo transforms the target detection task into a regression problem of a single neural network, achieving simultaneous detection of multiple targets in an image. The Yolo network divides the input image into S×S grid cells, each of which is responsible for detecting the target in the image. If the center point of the target falls within a grid cell, the grid cell is responsible for predicting the target.

[0098] For each grid cell, the network outputs a bounding box and the class probability corresponding to each bounding box. The bounding box is described by 5 parameters: the center coordinates of the target (x, y), width w, height h, and confidence c of the target. The confidence reflects the probability of the target existing in the bounding box and the accuracy of the bounding box.

[0099] Yolo uses convolutional neural networks (CNNs) to extract image features. The convolution layer extracts local features in the image through convolution operations, and reduces the dimension and computational complexity of the feature map through the pooling layer. As the network goes deeper, the resolution of the feature map gradually decreases, but the expressiveness of the features gradually increases.

[0100] In addition to the bounding box prediction, each grid cell also predicts a category probability value, which represents the probability that the target in the bounding box predicted by the grid cell belongs to each category. These probability values ​​are conditional probabilities under each bounding box confidence.

[0101] During the test phase, Yolo screens the predicted bounding boxes based on their confidence and category probability, and uses the bounding boxes with confidence higher than the set threshold as the final target detection results. At the same time, the non-maximum suppression algorithm (NMS) is used to remove overlapping bounding boxes to ensure that each target is detected only once.

[0102] Through the above steps, the combination of grid division, bounding box prediction and convolutional feature extraction can accurately locate and identify image targets in video key frames. By screening the confidence and category probability, the accuracy of target recognition is ensured, and the video image features are enriched through convolutional feature extraction, which effectively improves the overall recognition performance and processing efficiency of the system.

[0103] Furthermore, the step of converting the video voice clip into voice text and extracting text features from the voice text to obtain voice text features specifically includes:

[0104] Perform audio-to-text processing on the video voice clip to obtain the voice text;

[0105] Perform word segmentation on the speech text to obtain text segmentation;

[0106] The text features are encoded for the text segmentation to obtain the speech text features.

[0107] In this embodiment, first, the system performs audio-to-text processing on the video voice clip, which can be achieved through automatic speech recognition technology (ASR). The ASR system converts the voice content in the audio into corresponding text information by performing spectrum analysis and acoustic modeling on the audio signal. This process ensures that the voice content in the video can be presented in text form. Next, the system performs word segmentation processing on the obtained voice text. Word segmentation is to divide a continuous text string into words or phrases with independent meanings. In languages ​​such as Chinese, word segmentation is an important preprocessing step that can parse the text in a more fine-grained manner. Subsequently, the system performs text feature encoding on these word segmentation results. Common encoding methods include word vectors (such as Word2Vec, BERT, etc.). The text after word segmentation is converted into a feature vector that can be used for machine learning models through encoding, retaining the semantics and contextual information of the text.

[0108] Through the above steps, text information can be efficiently extracted and encoded from video speech, speech-to-text conversion can be achieved, and the text can be characterized to provide accurate semantic information support for subsequent knowledge association and analysis.

[0109] Furthermore, in the preset knowledge base system, before the step of performing knowledge association on the video image features and the speech text features to generate the associated knowledge graph, the following steps are also included:

[0110] Perform feature normalization processing on video image features and speech text features;

[0111] Use the preset task classifier to perform feature classification on the normalized video image features and speech text features to obtain the target task classification features.

[0112] In this embodiment, first, the system performs feature normalization on the extracted video image features and speech text features. Feature normalization is to eliminate the dimensional differences between different features, so that image features and text features can be compared and analyzed on the same scale to avoid a certain feature from having too much influence on the results. Then, the normalized features are classified using a preset task classifier. The task classifier can be a model based on deep learning or traditional machine learning, which can classify the input features according to the requirements of different tasks, and finally obtain classification features related to the target task.

[0113] Through the above steps, the consistency of multimodal features is ensured, the accuracy and efficiency of the classification model are improved, and it helps to achieve more accurate task classification and analysis.

[0114] In the preset knowledge base system, the steps of performing knowledge association on the video image features and the speech text features to generate the associated knowledge graph specifically include:

[0115] Perform feature vector transformation on the knowledge ontology in the knowledge base system to obtain knowledge feature vector;

[0116] In the knowledge base system, the Euclidean distance between the knowledge ontology and the target task classification features is calculated based on the knowledge feature vector and the target task classification feature vector;

[0117] Knowledge association is performed based on the Euclidean distance between the knowledge ontology and the target task classification features to generate an associated knowledge graph.

[0118] In this embodiment, first, the system converts the knowledge ontology in the knowledge base system into a feature vector. The knowledge ontology is a structured representation of different concepts, entities and their relationships in the knowledge base. By converting it into a feature vector, the system can use a unified vectorized representation to process the knowledge. The conversion process usually uses an embedding method, such as a word vector or a graph-based embedding technology, to map various concepts, relationships and other information in the knowledge ontology into a vector space. Then, the system will calculate the Euclidean distance between the preset task classification feature vector and these knowledge feature vectors in the knowledge base system. The Euclidean distance is used to measure the similarity between two vectors in the feature space. The smaller the distance, the higher the correlation between the two in semantics or features. Through this calculation method, the system can determine which knowledge ontologies in the knowledge base are closer to the target task classification features. Subsequently, based on the Euclidean distance, the knowledge association is performed. The system associates the knowledge ontology with high similarity with the task classification features, thereby generating an associated knowledge graph. This graph can show the relationship between the task features and the relevant knowledge points in the knowledge base.

[0119] Through the above steps, the classification features of the target task can be effectively matched with the relevant knowledge in the knowledge base, and the correlation can be measured by Euclidean distance to generate a high-quality associated knowledge graph. This greatly improves the efficiency and accuracy of knowledge retrieval and provides more efficient knowledge support for applications such as intelligent question and answer and recommendation systems.

[0120] Furthermore, the steps of obtaining associated knowledge tags from the associated knowledge graph and combining the associated knowledge tags to construct a target knowledge framework specifically include:

[0121] By using a preset keyword matching algorithm, the approximate knowledge tags of the associated knowledge tags are calculated based on the distance priority principle;

[0122] Sort the approximate knowledge labels to obtain an approximate knowledge label sequence;

[0123] A knowledge framework template is obtained, and associated knowledge tags and approximate knowledge tag sequences are inserted into the knowledge framework template to generate a target knowledge framework.

[0124] In this embodiment, first, the system calculates the approximate knowledge tags of the associated knowledge tags based on the distance priority principle through a preset keyword matching algorithm. The keyword matching algorithm determines which approximate knowledge tags are most relevant to the current task in the knowledge base by analyzing the text or semantic similarity between the knowledge tags and the preset keywords. The distance priority principle ensures that tags with a closer semantic distance to the associated knowledge tags are selected first, thereby ensuring the relevance and consistency between tags, so as to ensure that the system can accurately capture knowledge tags with similar semantics. Next, the system sorts these approximate knowledge tags. The sorting basis may include the degree of relevance of the tags to the task, the frequency of use, the historical application success rate, etc. The purpose of sorting is to ensure that the most valuable and relevant knowledge tags appear in the front position. Subsequently, the system obtains a knowledge framework template. The knowledge framework template is a preset structure used to help organize and present knowledge in different fields. At the corresponding position of the template, the system will automatically insert the associated knowledge tags and the sorted approximate knowledge tag sequence to ensure that the generated target knowledge framework can clearly and systematically reflect the knowledge content related to the task.

[0125] Through the above steps, intelligent matching and sorting technology can be used to ensure that the generated target knowledge framework is highly relevant and logical. The priority sorting of knowledge tags enables important information to be highlighted, thereby improving the practicality and accuracy of the knowledge framework and helping to achieve more efficient knowledge management and application.

[0126] For example, in insurance training scenarios, the use of the above technical solutions can significantly improve training efficiency and the accuracy of knowledge acquisition. For example, an insurance company plans to provide systematic training for newly hired insurance consultants, including insurance product knowledge, market analysis, customer service skills, etc. In this process, the system first obtains training videos related to insurance products and converts the voice information in the video into text through audio-to-text technology. Next, the system extracts key frames in the video and performs image feature analysis to identify and extract important image information, such as product display or customer service scenes. These extracted image features and voice text features will be normalized to ensure comparability between different data.

[0127] After that, the system classifies the normalized features through the set classifier and generates target task classification features related to insurance consultant training, such as "insurance product knowledge" and "customer communication skills". Next, the system converts the insurance knowledge ontology contained in the knowledge base into feature vectors and calculates the Euclidean distance between these knowledge ontologies and the target task classification features. This process can help the system identify the most relevant knowledge tags, such as "accident insurance", "liability insurance", "claims process", etc.

[0128] After identifying the related knowledge tags, the system uses a keyword matching algorithm to find tags similar to these knowledge tags based on the distance priority principle, such as taking into account tags related to "accident insurance", such as "accident claims" and "accidental injuries". These similar knowledge tags will be sorted to ensure that the most relevant knowledge is displayed first. When generating the target knowledge framework, the system will insert these tags into the preset framework template to form a comprehensive training knowledge framework. For example, the knowledge framework may include multiple modules under "insurance product knowledge", such as "introduction to liability insurance", "frequently asked questions about accident insurance", "how to handle claims", etc.

[0129] Throughout the training process, insurance consultants can quickly obtain the key information they need through such a knowledge framework, and can interact during the learning process, such as conducting in-depth learning and case analysis under specific knowledge tags. By combining video learning, image feature extraction, and text feature classification, the training content can be updated and optimized in real time to ensure that consultants are exposed to the latest market trends and product information. In addition, the system's automated generation capabilities have greatly reduced the workload of trainers, allowing trainers to devote more energy to personalized guidance and practical exercises.

[0130] Through the above process, the insurance training system not only improves the efficiency of training, but also enhances the accuracy of knowledge acquisition, providing a comprehensive, systematic and efficient learning platform for newly hired insurance consultants.

[0131] In the above embodiments, the present application discloses a method for automatically generating a knowledge framework, which belongs to the field of artificial intelligence technology. The present application automatically identifies key frames and voice content in the video through analysis and processing of the target video, extracts image features and text features, and uses a preset knowledge base system to associate these features with knowledge, generates an associated knowledge graph, and constructs a target knowledge framework by combining knowledge tags. Compared with the traditional knowledge framework construction method that relies on artificial knowledge reserves, the present application greatly improves the construction efficiency and reduces the dependence on manual intervention. At the same time, through automated feature extraction and knowledge association, the accuracy of knowledge framework construction is significantly improved, effectively solving the technical problems of low efficiency and poor accuracy of manual construction, and has a higher level of automation and intelligence.

[0132] In this embodiment, the electronic device (eg, Figure 1 The server shown in the figure) can receive instructions or obtain data through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.

[0133] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned target video information, the above-mentioned target video information can also be stored in a node of a blockchain.

[0134] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0135] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0136] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0137] 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 computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0138] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0139] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a knowledge framework automatic generation device, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0140] like Figure 3 As shown, the knowledge framework automatic generation device 300 described in this embodiment includes:

[0141] The video parsing module 301 is used to obtain a target video and parse the target video to obtain a video frame sequence and a video voice segment;

[0142] The image feature module 302 is used to identify key frames in the video frame sequence and extract image features from the key frames to obtain video image features;

[0143] The text feature module 303 is used to convert the video voice segment into voice text and extract text features from the voice text to obtain voice text features;

[0144] The knowledge association module 304 is used to perform knowledge association on video image features and speech text features in a preset knowledge base system to generate an associated knowledge graph;

[0145] The framework construction module 305 is used to obtain associated knowledge tags from the associated knowledge graph, and combine the associated knowledge tags to construct a target knowledge framework.

[0146] Furthermore, the image feature module 302 is specifically used for:

[0147] Identify the histogram data and grayscale data of the video frames in the video frame sequence, and determine the key frames in the video frame sequence based on the histogram data and the grayscale data; identify the image targets of the key frames based on a preset image target recognition algorithm, and extract features of the image targets to obtain video image features.

[0148] Furthermore, the image feature recognition module 302 is further used for:

[0149] Calculate the histogram data and grayscale image data of two adjacent video frames in the video frame sequence; calculate the weighted Euclidean distance of the two adjacent video frames based on the histogram data and the grayscale image data; determine the shot conversion boundary of the video frame sequence based on the weighted Euclidean distance; determine the key frame in the video frame sequence based on the shot conversion boundary.

[0150] Furthermore, the image feature recognition module 302 is further used to specifically include:

[0151] The key frame is divided into several grid units, where each grid unit is responsible for detecting an initial prediction target in the key frame; a bounding box is generated for each grid unit, and the confidence and category probability of the bounding box are predicted; based on the confidence and category probability of the bounding box, the initial prediction target is screened to obtain the image target of the key frame; convolutional feature extraction is performed on the image target to obtain video image features.

[0152] Furthermore, the text feature module 303 is specifically used for:

[0153] The video voice clip is processed into text to obtain the voice text; the voice text is segmented to obtain text segmentation; the text segmentation is encoded with text features to obtain the voice text features.

[0154] Furthermore, the knowledge framework automatic generation device 300 further includes a feature classification module, which is specifically used for:

[0155] The video image features and the speech text features are subjected to feature normalization processing; the normalized video image features and the speech text features are subjected to feature classification using a preset task classifier to obtain target task classification features.

[0156] Furthermore, the knowledge association module 304 is specifically used for:

[0157] The knowledge ontology in the knowledge base system is transformed into a feature vector to obtain a knowledge feature vector; within the knowledge base system, the Euclidean distance between the knowledge ontology and the target task classification feature is calculated based on the knowledge feature vector and the target task classification feature vector; knowledge association is performed based on the Euclidean distance between the knowledge ontology and the target task classification feature to generate an associated knowledge graph.

[0158] Furthermore, the framework building module 305 is specifically used for:

[0159] By using a preset keyword matching algorithm, the approximate knowledge tags of the associated knowledge tags are calculated based on the distance priority principle;

[0160] Sort the approximate knowledge labels to obtain an approximate knowledge label sequence;

[0161] A knowledge framework template is obtained, and associated knowledge tags and approximate knowledge tag sequences are inserted into the knowledge framework template to generate a target knowledge framework.

[0162] In the above embodiment, the present application discloses a knowledge framework automatic generation device, which belongs to the field of artificial intelligence technology. The present application automatically identifies the key frames and voice content in the video through analysis and processing of the target video, extracts image features and text features, and uses a preset knowledge base system to associate these features with knowledge, generates an associated knowledge graph, and constructs a target knowledge framework by combining knowledge tags. Compared with the traditional knowledge framework construction method that relies on artificial knowledge reserves, the present application greatly improves the construction efficiency and reduces the dependence on manual intervention. At the same time, through automated feature extraction and knowledge association, the accuracy of knowledge framework construction is significantly improved, effectively solving the technical problems of low efficiency and poor accuracy of manual construction, and has a higher level of automation and intelligence.

[0163] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0164] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 4 with a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0165] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0166] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk equipped on the computer device 4, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the knowledge framework automatic generation method, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0167] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the computer-readable instructions stored in the memory 41 or process data, such as running the computer-readable instructions of the knowledge framework automatic generation method.

[0168] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0169] In the above embodiment, the present application discloses a computer device, which belongs to the field of artificial intelligence technology. The present application automatically identifies the key frames and voice content in the video by analyzing and processing the target video, extracts the image features and text features, and uses the preset knowledge base system to associate these features with knowledge, generates an associated knowledge graph, and constructs the target knowledge framework by combining knowledge tags. Compared with the traditional knowledge framework construction method that relies on artificial knowledge reserves, the present application greatly improves the construction efficiency and reduces the dependence on manual intervention. At the same time, through automated feature extraction and knowledge association, the accuracy of knowledge framework construction is significantly improved, effectively solving the technical problems of low efficiency and poor accuracy of manual construction, and has a higher level of automation and intelligence.

[0170] The present application also provides another implementation, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the automatic knowledge framework generation method as described above.

[0171] In the above embodiments, the present application discloses a computer-readable storage medium, which belongs to the field of artificial intelligence technology. The present application automatically identifies key frames and voice content in the target video through analysis and processing of the target video, extracts image features and text features, and uses a preset knowledge base system to associate these features with knowledge, generates an associated knowledge graph, and constructs a target knowledge framework by combining knowledge tags. Compared with the traditional knowledge framework construction method that relies on artificial knowledge reserves, the present application greatly improves the construction efficiency and reduces the dependence on manual intervention. At the same time, through automated feature extraction and knowledge association, the accuracy of knowledge framework construction is significantly improved, effectively solving the technical problems of low efficiency and poor accuracy of manual construction, and has a higher level of automation and intelligence.

[0172] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0173] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0174] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for automatically generating a knowledge framework, characterized in that: include: Obtain a target video, and parse the target video to obtain a video frame sequence and a video voice segment; Identifying key frames in the video frame sequence, and performing image feature extraction on the key frames to obtain video image features; Convert the video voice clip into voice text, and extract text features from the voice text to obtain voice text features; In a preset knowledge base system, knowledge association is performed on the video image features and the speech text features to generate an associated knowledge graph; The associated knowledge tags are obtained from the associated knowledge graph, and the associated knowledge tags are combined to construct a target knowledge framework.

2. The method for automatically generating a knowledge framework according to claim 1, characterized in that: The step of identifying key frames in the video frame sequence and extracting image features from the key frames to obtain video image features specifically includes: Identifying histogram data and grayscale image data of video frames in the video frame sequence, and determining key frames in the video frame sequence based on the histogram data and the grayscale image data; The image target of the key frame is identified based on a preset image target recognition algorithm, and features of the image target are extracted to obtain video image features.

3. The method for automatically generating a knowledge framework as claimed in claim 2, characterized in that: The step of identifying the histogram data and the grayscale image data of the video frames in the video frame sequence, and determining the key frames in the video frame sequence based on the histogram data and the grayscale image data, specifically includes: Calculating histogram data and grayscale image data of two adjacent video frames in the video frame sequence; Calculating a weighted Euclidean distance between two adjacent video frames based on the histogram data and the grayscale image data; Determining a shot transition boundary of the video frame sequence based on the weighted Euclidean distance; A key frame in the video frame sequence is determined based on the shot transition boundary.

4. The method for automatically generating a knowledge framework as claimed in claim 2, characterized in that: The step of identifying the image target of the key frame based on a preset image target recognition algorithm, and extracting features of the image target to obtain video image features specifically includes: Dividing the key frame into a plurality of grid units, wherein each of the grid units is responsible for detecting an initial prediction target in the key frame; Generate a bounding box for each of the grid cells, and predict the confidence and category probability of the bounding box; Based on the confidence level and the category probability of the bounding box, the initial prediction target is screened to obtain the image target of the key frame; Convolution feature extraction is performed on the image target to obtain video image features.

5. The method for automatically generating a knowledge framework according to claim 1, characterized in that: The step of converting the video voice segment into voice text and extracting text features from the voice text to obtain voice text features specifically includes: Performing audio-to-text processing on the video audio segment to obtain the audio text; Performing word segmentation processing on the speech text to obtain text segmentation; The text segmentation is performed on the text feature encoding to obtain the speech text feature.

6. The method for automatically generating a knowledge framework according to claim 1, characterized in that: Before the step of performing knowledge association on the video image features and the speech text features in the preset knowledge base system to generate an associated knowledge graph, the method further includes: Performing feature normalization processing on the video image features and the speech text features; Using a preset task classifier to perform feature classification on the normalized video image features and the speech text features to obtain target task classification features; The step of performing knowledge association on the video image features and the speech text features in the preset knowledge base system to generate an associated knowledge graph specifically includes: Performing feature vector conversion on the knowledge ontology in the knowledge base system to obtain a knowledge feature vector; In the knowledge base system, according to the knowledge feature vector and the target task classification feature vector, the Euclidean distance between the knowledge ontology and the target task classification feature is calculated; Knowledge association is performed according to the Euclidean distance between the knowledge ontology and the target task classification features to generate the associated knowledge graph.

7. The method for automatically generating a knowledge framework according to claim 1, characterized in that: The step of acquiring associated knowledge tags from the associated knowledge graph and combining the associated knowledge tags to construct a target knowledge framework specifically includes: By using a preset keyword matching algorithm, the approximate knowledge tags of the associated knowledge tags are calculated based on the distance priority principle; Sorting the approximate knowledge tags to obtain an approximate knowledge tag sequence; A knowledge framework template is obtained, and the associated knowledge tags and the approximate knowledge tag sequence are inserted into the knowledge framework template to generate the target knowledge framework.

8. A knowledge framework automatic generation device, characterized in that: include: A video parsing module is used to obtain a target video and parse the target video to obtain a video frame sequence and a video voice segment; An image feature module, used for identifying key frames in the video frame sequence and performing image feature extraction on the key frames to obtain video image features; A text feature module, used for converting the video voice segment into voice text, and performing text feature extraction on the voice text to obtain voice text features; A knowledge association module, used to perform knowledge association on the video image features and the speech text features in a preset knowledge base system to generate an associated knowledge graph; The framework construction module is used to obtain associated knowledge tags from the associated knowledge graph, and combine the associated knowledge tags to construct a target knowledge framework.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for automatically generating a knowledge framework as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the method for automatically generating a knowledge framework according to any one of claims 1 to 7 are implemented.

Images