Artificial intelligence-based cultural relic science popularization question and answer system and method
By designing a cultural relics popularization question and answer system based on artificial intelligence, the problem of lack of interactivity and personalization of traditional cultural relics popularization methods is solved, and convenient, efficient and accurate cultural relics popularization services are achieved to meet the diverse needs of users.
Patent Information
- Application Number
- CN202310957174.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-01
- Publication Date
- 2025-02-14
AI Technical Summary
Traditional cultural relics popular science methods have a single form, lack of interactiveness and personalized experience, which is difficult to meet people's needs for new knowledge, and is limited by time and space, so knowledge cannot be obtained anytime and anywhere.
Design a cultural relics popular science question and answer system based on artificial intelligence, including speech recognition module, natural language processing module, knowledge base module, question and answer matching module, answer generation module and user feedback module, to provide personalized cultural relics popular science services through intelligent means.
It has achieved convenient, efficient and accurate cultural relics popular science services to users, meeting the needs of different users, and continuously optimized and improved through interactive interfaces and user feedback.
Smart Images
Figure CN119441412A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of cultural tourism data management technology, and in particular to an artificial intelligence-based cultural relics science popularization question-and-answer system and method. Background Art
[0002] As people attach more importance to cultural heritage and cultural relics protection work continues to deepen, more and more people are beginning to pay attention to popular science knowledge about cultural relics. The traditional way of popularizing cultural relics is mainly through exhibitions, explanations, publications, etc. However, these methods have the following defects: the traditional way of popularizing cultural relics is single in form, lacks interactivity and personalized experience, and is difficult to attract the interest of young people. The content of popular science about cultural relics is usually static, difficult to update and expand in time, and cannot meet people's demand for new knowledge. Popular science about cultural relics is limited by time and space, and can only be carried out in specific places and times, which cannot meet people's needs to obtain knowledge anytime and anywhere. Summary of the invention
[0003] The present application provides a cultural relics science popularization question and answer system and system and method based on artificial intelligence, aiming to provide people with more convenient, efficient and accurate cultural relics science popularization services in an intelligent way.
[0004] This application provides a cultural relic science question-answering system based on artificial intelligence, including:
[0005] A speech recognition module is used to convert the user's speech input into text through speech recognition technology;
[0006] A natural language processing module is used to perform natural language processing on the text input by the user, including but not limited to word segmentation, part-of-speech tagging, and named entity recognition;
[0007] A knowledge base module is used to establish a cultural relic knowledge base, which includes but is not limited to basic information, historical background, cultural connotation of cultural relics, and common questions and answers related to cultural relics;
[0008] The question-answer matching module is used to match the questions input by the user with the questions in the knowledge base to find the best answer;
[0009] The answer generation module is used to generate the best answer based on the matching results and convert the answer into speech output through speech synthesis technology;
[0010] The user feedback module is used for users to evaluate and provide feedback on the system's answers, so that the system can be continuously optimized and improved.
[0011] Furthermore, the speech recognition module includes:
[0012] The sound wave acquisition module is used to obtain audio data through a microphone or other sensors during the speech recognition process;
[0013] The audio preprocessing module performs preprocessing steps after collecting the original audio, including but not limited to removing noise, filtering interference and enhancing sound quality;
[0014] The data cleaning module cleans and analyzes the data when there is enough data, removes unnecessary noise or noise, and corrects erroneous or incomplete information;
[0015] The word segmentation conversion module is used to convert speech into words using a rule-based approach;
[0016] Feature extraction module, extracts useful features from the classified word sequences to facilitate subsequent analysis;
[0017] The model training module is used to select a suitable deep learning model for model training based on the selected feature extraction method and task requirements;
[0018] The prediction output module is used to output the speech conversion results.
[0019] Furthermore, the knowledge base module includes:
[0020] The knowledge collection module is used to collect knowledge related to cultural relics through various channels, including but not limited to literature, expert explanations, and museum exhibitions;
[0021] The knowledge organization module is used to organize and classify the collected cultural relics knowledge to form a structured knowledge system;
[0022] The knowledge summarization module is used to summarize the organized knowledge and extract the core content including but not limited to the characteristics, historical background and cultural connotation of cultural relics;
[0023] The database building module is used to store and manage the organized knowledge in the form of a database.
[0024] Furthermore, the question-answer matching module performs matching based on the calculation method of sentence similarity, and the calculation formula is:
[0025]
[0026] Among them, SIM(A, B) represents the similarity between sentence A and sentence B, comment(A, B) represents the commonality between text A and text B, and description(A, B) represents the complete information set of text A and text B.
[0027] This application also provides a cultural relic science question-answering method based on artificial intelligence, comprising the following steps:
[0028] Through speech recognition technology, the user's voice input is converted into text;
[0029] Perform natural language processing on the text input by the user, including but not limited to word segmentation, part-of-speech tagging, and named entity recognition;
[0030] Establishing a cultural relic knowledge base, which includes but is not limited to basic information, historical background, cultural connotations of cultural relics, and common questions and answers related to cultural relics;
[0031] Match the questions entered by the user with the questions in the knowledge base to find the best answer;
[0032] Generate the best answer based on the matching results, and convert the answer into speech output through speech synthesis technology;
[0033] Users evaluate and provide feedback on the system's answers, allowing the system to be continuously optimized and improved.
[0034] Furthermore, the user's voice input is converted into text through voice recognition technology, specifically:
[0035] The sound wave acquisition module is used to obtain audio data through a microphone or other sensors during the speech recognition process;
[0036] The audio preprocessing module performs preprocessing steps after collecting the original audio, including but not limited to removing noise, filtering interference and enhancing sound quality;
[0037] The data cleaning module cleans and analyzes the data when there is enough data, removes unnecessary noise or noise, and corrects erroneous or incomplete information;
[0038] The word segmentation conversion module is used to convert speech into words using a rule-based approach;
[0039] Feature extraction module, extracts useful features from the classified word sequences to facilitate subsequent analysis;
[0040] The model training module is used to select a suitable deep learning model for model training based on the selected feature extraction method and task requirements;
[0041] The prediction output module is used to output the speech conversion results.
[0042] Furthermore, the establishment of the cultural relics knowledge base is specifically as follows:
[0043] Collect knowledge related to cultural relics through various channels, including but not limited to literature, expert explanations, and museum exhibitions;
[0044] Organize and classify the collected cultural relics knowledge to form a structured knowledge system;
[0045] Summarize the knowledge that has been sorted out and extract the core content that is not limited to the characteristics, historical background, and cultural connotation of cultural relics;
[0046] The organized knowledge is stored and managed in the form of a database.
[0047] Furthermore, the question input by the user is matched with the question in the knowledge base, and the calculation formula is:
[0048]
[0049] Among them, SIM(A, B) represents the similarity between sentence A and sentence B, comment(A, B) represents the commonality between text A and text B, and description(A, B) represents the complete information set of text A and text B.
[0050] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:
[0051] The present invention adopts an intelligent approach to provide users with cultural relics science popularization services more conveniently, efficiently and accurately. Through the intelligent recommendation system and user portrait system, more personalized services can be provided to users. Through the diversification of interactive interfaces, the needs of different users can be met. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0053] Figure 1 This is a flowchart of a specific embodiment of the present application;
[0054] Figure 2 A workflow diagram for question and answer matching for this application;
[0055] Figure 3 This is the workflow diagram of the answer query module for this application;
[0056] Figure 4 This is the question and answer interface diagram of the cultural relics question and answer system for this application. DETAILED DESCRIPTION
[0057] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.
[0058] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.
[0059] like Figure 1 As shown, the present application provides a cultural relic science question-answering system based on artificial intelligence, comprising:
[0060] Speech recognition module: Converts user voice input into text through speech recognition technology.
[0061] Natural language processing module: performs natural language processing on the text input by the user, including word segmentation, part-of-speech tagging, named entity recognition, etc., for subsequent processing.
[0062] Knowledge base module: Establish a cultural relic knowledge base, including the basic information of the cultural relics, historical background, cultural connotation, etc., as well as common questions and answers related to cultural relics.
[0063] Question-answer matching module: matches the questions entered by the user with the questions in the knowledge base to find the best answer.
[0064] Answer generation module: Generates the best answer based on the matching results and converts the answer into speech output through speech synthesis technology.
[0065] User feedback module: Users can evaluate and provide feedback on the system's answers so that the system can be continuously optimized and improved.
[0066] The above modules specifically include:
[0067] 1) Speech Recognition Module
[0068] Speech recognition refers to the process of converting human language into computer-readable digital signals. In the system of intelligent question-answering robots, the core of speech recognition is to convert sound into text information and combine it with natural language processing algorithms to generate answers to user questions. The following are the main working principles of speech recognition:
[0069] Acoustic wave acquisition: In the process of speech recognition, audio data needs to be acquired through microphones or other sensors. These devices emit sound waves into the surrounding environment and record them to obtain the required data.
[0070] Audio preprocessing: After collecting the raw audio, some preprocessing steps are required, such as removing noise, filtering out interference, and enhancing sound quality, etc. This helps improve the accuracy and reliability of the data.
[0071] Data cleaning: Once you have enough data, you need to clean it and analyze it. This includes removing unnecessary noise or clutter and correcting erroneous or incomplete information.
[0072] Word segmentation: Word segmentation is a technique for converting speech into words. This is usually done using a rule-based approach. The advantage of this approach is that it can quickly classify large amounts of speech data, but the disadvantage is that it cannot handle problems such as ambiguity well.
[0073] Feature extraction: Extract useful features from the classified word sequences to facilitate subsequent analysis. Common methods include TF-IDF (term frequency-inverse document frequency) vectorization, BagofWords, etc.
[0074] Model training: According to the selected feature extraction method and task requirements, select a suitable deep learning model (such as recurrent neural network (RNN) or convolutional neural network, etc.) to train the model. This process can be evaluated by cross-validation, and the hyperparameters can be adjusted according to the evaluation results to improve the performance of the model.
[0075] Prediction output: When a new voice is input, after a series of processing, the corresponding answer can be obtained. Specifically, each question can be split into multiple sub-questions for analysis, and then these sub-questions are assigned to different categories, and finally the final result is obtained by combining them.
[0076] 2) Knowledge base module
[0077] It is the core part of the entire cultural relic intelligent question-answering system. It mainly builds a database containing rich cultural relic knowledge by collecting, organizing, classifying and summarizing cultural relic-related knowledge. The principles of the knowledge base module mainly include the following aspects:
[0078] Collection of knowledge: Collect knowledge related to cultural relics through various channels, including literature, expert explanations, museum exhibitions, etc.
[0079] Knowledge organization: Organize and classify the collected cultural relics knowledge to form a structured knowledge system to facilitate subsequent query and application.
[0080] Summarization of knowledge: Summarize and summarize the organized knowledge to extract the core content of the characteristics, historical background, cultural connotation, etc. of the cultural relics.
[0081] Database construction: Store and manage the organized knowledge in the form of a database for subsequent query and application.
[0082] In actual applications, the content of the knowledge base module includes basic information of cultural relics, historical background, cultural connotation, production process, preservation and restoration, etc. At the same time, in order to improve the intelligence level of the system, the knowledge base can also be combined with natural language processing technology to achieve intelligent answers to user questions.
[0083] 3) Question and answer matching module
[0084] The matching is based on the calculation method of sentence similarity. The string-based method calculates the proportion of common strings in all words in two texts at the literal level, which depends on the number of common words. The calculation of common words can be divided into two granularities: characters and words. English characters can use character-based calculation methods, and Chinese texts are suitable for string-based calculation methods. This patent uses cosine distance to calculate vector similarity.
[0085] The following two formulas are both formulas for calculating text vectors, which are total-score relationships. Formula 2 is a generalized concept of similarity and is an explanation of the calculation principle. Formula 1 is the calculation method of Formula 2 based on strings.
[0086] Cosine distance is the calculation of the similarity between two vectors after the text is vectorized, using the cosine value of the angle between two vectors in a vector space. The closer the two vectors are to parallel, the closer the cosine value is to 1, indicating that the vectorized representations of the two texts are more similar. Otherwise, it indicates that they are very different and have little similarity.
[0087]
[0088] In formula 1, x i ,y i It means multiplying each component vector in sentence A and sentence B.
[0089] The quality of a question-answering system lies in the efficient and accurate matching of questions and the updating of question pairs. Therefore, the quality of text similarity calculation is an important factor in measuring the performance of a question-answering system. Text similarity can be calculated by a certain rule method to obtain a specific numerical value representing the similarity between sentences. The larger the value, the smaller the difference between text entities and the more similarities they have; conversely, the larger the difference, the smaller the numerical value, and the smaller the similarity between text entities.
[0090] The quality of a question-answering system lies in the efficient and accurate matching of questions and the updating of question pairs. Therefore, the quality of text similarity calculation is an important factor in measuring the performance of a question-answering system. Text similarity can be calculated by a certain rule method to obtain a specific numerical value representing the similarity between sentences. The larger the value, the smaller the difference between text entities and the more similarities they have; conversely, the larger the difference, the smaller the value, and the smaller the similarity between text entities. The text similarity theorem based on hypothesis reasoning is shown in Formula 1.
[0091]
[0092] Formula 2 is defined as a generalized concept of similarity, where SIM(A, B) represents the similarity between sentence A and sentence B, comment(A, B) represents the commonality between text A and text B, and description(A, B) represents the complete information set of text A and text B. In practical applications, common common information of texts includes synonyms, co-occurring words, and hyponymy.
[0093] Cosine distance is a calculation method based on the string in Theorem 1. After the text is vectorized, the cosine value of the angle between two vectors in a vector space is used to calculate the similarity. The closer the two vectors are to parallel, the closer the cosine value is to 1, indicating that the vectorized representations of the two texts are more similar. Otherwise, it indicates that they are very different and have little similarity.
[0094]
[0095] In formula 2, x i y i It means multiplying each component vector in sentence A and sentence B.
[0096] This application runs in the Python programming environment, uses the Tensorflow deep learning framework, uses Neo4j to store data in the database, and uses Html, Css, JavaScript and other technologies to develop front-end pages. This article uses the Neo4j database to complete the construction of the knowledge graph, and stores relevant data of the knowledge graph in the field of cultural relics in the database, and uses the query language Cypher provided by the Neo4j database to retrieve knowledge. The front end of the system is written using the MVC design pattern, where M stands for Model, which is responsible for database operations and the implementation of business logic, and is the main body of MVC. V stands for View, which represents the interactive interface. C stands for Controller, which controls the entire business process and selects interfaces according to the presentation layer to perform business logic processing. The system framework structure is as follows Figure 1 shown.
[0097] (1) Data Layer
[0098] The data layer is one of the important components of the question-answering system, and it operates on the database. The data processed includes the data of the knowledge graph, the question-answering text, and the data of the system operation. All operation steps are implemented through the background code.
[0099] (2) Logical layer
[0100] The intelligent question-answering system of this application has three main backend logic layers: question processing and analysis, template matching, and answer generation. Question analysis mainly analyzes and understands the questions input by the user, mainly by parsing natural language through operations such as word segmentation, part-of-speech tagging, and removing stop words, and understanding sentences in combination with contextual content; template matching mainly searches for similar questions in questions and question templates; answer generation mainly obtains the corresponding Cypher query statement through the matched template, and obtains the required answer by mobilizing the database.
[0101] (3) Presentation Layer
[0102] The presentation layer processes user requests and data responses when the user interacts with the visualized page. When the user enters the human-computer interaction page and enters a question, the front end will feed back the received information to the back end, which will process and parse the answer obtained from the database and present it on the front end page. The system is built using the lightweight web framework web.py.
[0103] Main functions of the question-answering system:
[0104] In the design of the main functions of the question-answering system, the question analysis and template matching methods are adopted. When the user enters a question on the front-end page, the first step is to analyze the question, which requires preprocessing such as word segmentation and part-of-speech tagging, and then entity recognition; the second step is template matching. The preprocessed question is abstracted and the similarity is calculated with the question template, and finally the correct answer is presented. Using the Cypher statement corresponding to the question template, the query statement is constructed through the identified entities, and the answer is queried and returned in the graph database.
[0105] (1) Problem Preprocessing
[0106] The main work of question preprocessing includes word segmentation, part-of-speech tagging, and stop word removal. When the user enters a question in the question-answering interface, question preprocessing is required first, and then entity recognition is performed through a neural network. The natural language after question preprocessing will be abstracted, and the question-answering template with the highest matching degree will be found through similarity calculation. Table 1 is an example of the input and output results of the abstraction after question preprocessing.
[0107]
[0108] Table 1 Examples of input and output results after problem preprocessing
[0109] (2) Question-Answer Matching Module
[0110] After preprocessing the user input question, the question is abstracted into a template form and matched with the question-answer template, and the template with the highest similarity is selected through similarity calculation. Figure 2 As shown in the figure, question similarity calculation is used to match question templates for abstract questions. The similarity calculation is performed by matching question templates, using the TF-IDF (termfrequency–inverse document frequency) method to represent sentence vectors, combined with word order similarity calculation, and weighted summation of the two algorithms to calculate the final score.
[0111] (3) Answer query
[0112] Use similarity calculation to match question templates, use the Cypher statement corresponding to the question template with the highest similarity, combine with the identified entities, construct a query statement to query the answer in the graph database and present it on the interactive interface. Specifically, through similarity calculation to match question templates, embed the identified entities into the Cypher statement corresponding to the question template with the highest similarity, query the answer in the Neo4j graph database, and return the query results to the interactive interface to present to the user. The module process is as follows Figure 3 .
[0113] (4) Answer Generation Module
[0114] The abstract question is matched with the question template to obtain the most similar question template, and the corresponding Cypher statement of the template is called to query the database Neo4j. In the query module, the query operation is implemented using the py2neo library in Python. For example, to query the attributes of an entity, the Cypher statement corresponding to its question template is: Match(a)u[ <RelationuName]u
[0115] System workflow:
[0116] 1. The user enters a question: What is the function of a pottery jar?
[0117] 2. Preprocess the question and identify entities. The recognition result is [pottery jar]
[0118] 3. Template the question and match the question template:
[0119] MATCH(m:cultural)-[r:function]->(n:use)wherem.name='{***}'r.name,
[0120] 4. Replace the identified entities, generate Cypher query statements, query in Neo4j and return the results.
[0121] Query statement:
[0122] MATCH(m:cultural)-[r:function]->(n:use)where m.name='{pottery jar}'return r.name
[0123] Return results: Daily necessities
[0124] After the user logs in to the Q&A system, a dialog box will automatically pop up saying "Hello, I am Xiaowen, a robot. How can I help you?" At this point, the user only needs to enter the question they want to query, and the Q&A system will call the Q&A model to generate the corresponding reply. Figure 4 Shown is the question and answer interface of the cultural relics question and answer system.
[0125] (5) System testing
[0126] To test the question-and-answer system of this application, we first let users use our cultural relic question-and-answer system, and then we counted 5 questions that no one asked among 30 users in the background for analysis. After the user asks a question, the cultural relic question-and-answer system will ask the user whether the question has been solved while giving the answer, so that we can do statistics in the background. Therefore, we use accuracy to describe the evaluation criteria of the cultural relic question-and-answer system. The accuracy of the cultural relic question-and-answer system is counted, and the results are shown in Table 3.
[0127] Accuracy\Number of questions total Deduplication issues Number of questions answered correctly 141 128 Accuracy 94％ 93.4％
[0128] Table 3 Test results of the cultural relics question-answering system
[0129] As can be seen from Table 3, the total number of correctly answered questions in the cultural relics question-and-answer system is 141, with an accuracy rate of 94%. After removing the 13 repeated questions, the accuracy rate is 93.4%, which shows that the cultural relics question-and-answer system in this paper can help users answer questions well.
[0130] The following is a specific embodiment of the present invention:
[0131] like Figure 1 , the basic process is:
[0132] Step S110, the user asks "What are the opening hours of the Palace Museum?";
[0133] Step S120, the voice recognition module converts the user's voice input into text: "What are the opening hours of the Palace Museum?";
[0134] Step S130, the natural language processing module performs natural language processing on the text input by the user, including word segmentation, part-of-speech tagging, named entity recognition, etc., to obtain the keywords "Palace Museum" and "opening hours";
[0135] Step S140: the question-answer matching module matches the keywords with the questions in the knowledge base and finds the best answer: "The opening hours of the Palace Museum are from 9:00 a.m. to 5:00 p.m. every day and closed on Mondays.";
[0136] In step S1450, the user can evaluate and provide feedback on the system's answer so that the system can be continuously optimized and improved, and then continue to step S110.
[0137] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity or device including the element.
[0138] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims
1. A cultural relic science question-and-answer system based on artificial intelligence, characterized in that: include: A speech recognition module is used to convert the user's speech input into text through speech recognition technology; A natural language processing module is used to perform natural language processing on the text input by the user, including but not limited to word segmentation, part-of-speech tagging, and named entity recognition; A knowledge base module is used to establish a cultural relic knowledge base, which includes but is not limited to basic information, historical background, cultural connotation of cultural relics, and common questions and answers related to cultural relics; The question-answer matching module is used to match the questions input by the user with the questions in the knowledge base to find the best answer; The answer generation module is used to generate the best answer based on the matching results and convert the answer into speech output through speech synthesis technology; The user feedback module is used for users to evaluate and provide feedback on the system's answers, so that the system can be continuously optimized and improved.
2. The cultural relics science popularization question-and-answer system based on artificial intelligence according to claim 1 is characterized in that: The speech recognition module comprises: The sound wave acquisition module is used to obtain audio data through a microphone or other sensors during the speech recognition process; The audio preprocessing module performs preprocessing steps after collecting the original audio, including but not limited to removing noise, filtering interference and enhancing sound quality; The data cleaning module cleans and analyzes the data when there is enough data, removes unnecessary noise or noise, and corrects erroneous or incomplete information; The word segmentation conversion module is used to convert speech into words using a rule-based approach; Feature extraction module, extracts useful features from the classified word sequences to facilitate subsequent analysis; The model training module is used to select a suitable deep learning model for model training based on the selected feature extraction method and task requirements; The prediction output module is used to output the speech conversion results.
3. The cultural relics science popularization question-and-answer system based on artificial intelligence according to claim 1 is characterized in that: The knowledge base module includes: The knowledge collection module is used to collect knowledge related to cultural relics through various channels, including but not limited to literature, expert explanations, and museum exhibitions; The knowledge organization module is used to organize and classify the collected cultural relics knowledge to form a structured knowledge system; The knowledge summarization module is used to summarize the organized knowledge and extract the core content including but not limited to the characteristics, historical background and cultural connotation of cultural relics; The database building module is used to store and manage the organized knowledge in the form of a database.
4. The cultural relics science popularization question-and-answer system based on artificial intelligence according to claim 1 is characterized in that: The question-answer matching module performs matching based on the calculation method of sentence similarity, and the calculation formula is: Among them, SIM(A, B) represents the similarity between sentence A and sentence B, comment(A, B) represents the commonality between text A and text B, and description(A, B) represents the complete information set of text A and text B.
5. A cultural relic science question-answering method based on artificial intelligence, characterized in that: The following steps are involved: Through speech recognition technology, the user's voice input is converted into text; Perform natural language processing on the text input by the user, including but not limited to word segmentation, part-of-speech tagging, and named entity recognition; Establishing a cultural relic knowledge base, which includes but is not limited to basic information, historical background, cultural connotations of cultural relics, and common questions and answers related to cultural relics; Match the questions entered by the user with the questions in the knowledge base to find the best answer; Generate the best answer based on the matching results, and convert the answer into speech output through speech synthesis technology; Users evaluate and provide feedback on the system's answers, allowing the system to be continuously optimized and improved.
6. The method for cultural relics science popularization based on artificial intelligence according to claim 5 is characterized in that: The method of converting the user's voice input into text through voice recognition technology is as follows: The sound wave acquisition module is used to obtain audio data through a microphone or other sensors during the speech recognition process; The audio preprocessing module performs preprocessing steps after collecting the original audio, including but not limited to removing noise, filtering interference and enhancing sound quality; The data cleaning module cleans and analyzes the data when there is enough data, removes unnecessary noise or noise, and corrects erroneous or incomplete information; The word segmentation conversion module is used to convert speech into words using a rule-based approach; Feature extraction module, extracts useful features from the classified word sequences to facilitate subsequent analysis; The model training module is used to select a suitable deep learning model for model training based on the selected feature extraction method and task requirements; The prediction output module is used to output the speech conversion results.
7. The method for cultural relics science popularization based on artificial intelligence according to claim 5 is characterized in that: The establishment of the cultural relics knowledge base is specifically as follows: Collect knowledge related to cultural relics through various channels, including but not limited to literature, expert explanations, and museum exhibitions; Organize and classify the collected cultural relics knowledge to form a structured knowledge system; Summarize the knowledge that has been sorted out and extract the core content that is not limited to the characteristics, historical background, and cultural connotation of cultural relics; The organized knowledge is stored and managed in the form of a database.
8. The method for cultural relics science popularization based on artificial intelligence according to claim 5 is characterized in that: The question input by the user is matched with the question in the knowledge base, and the calculation formula is: Among them, SIM(A, B) represents the similarity between sentence A and sentence B, comment(A, B) represents the commonality between text A and text B, and description(A, B) represents the complete information set of text A and text B.