Intelligent question and answer processing method and device for old people, storage medium and computer equipment

Abstract

The invention relates to the technical field of computers, and discloses an intelligent question and answer processing method and device for old people, a storage medium and computer device.The method comprises the steps that an original medical field document is converted into texts in various forms, and a multi-source database is constructed, the form text comprises a term and explanation form text, a question and answer pair form text and a layered summary form text, and the text content in each form text is a spoken language; receiving a medical problem proposed by the elderly user, and obtaining a problem sentence vector based on the medical problem; and based on the multi-source database and the question sentence vector, generating an answer statement corresponding to the medical question, performing oral rewriting on the answer statement, and returning the rewritten answer statement. The multi-source database is formed by effectively integrating health knowledge of old people dispersed in different sources and forms and carrying out oral improvement, intelligent question answering is carried out through the multi-source database, oral questioning and professional term understanding obstacles can be effectively dealt with, and the user experience of the old people is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and device for intelligent question-answering processing for the elderly, a storage medium, and a computer device. Background Art

[0002] As the global population continues to age, the health of the elderly has become a focus of widespread concern in all sectors of society. The health needs of the elderly are growing, and the demand for health consultation and knowledge answers is becoming more urgent. However, traditional health consultation methods often have many inconveniences, such as limited medical resources and poor consultation channels, which make it difficult to meet the actual needs of the elderly. Summary of the invention

[0003] In view of this, the present application provides a method and device for intelligent question and answer processing for the elderly, a storage medium, and a computer device. By effectively integrating the health knowledge of the elderly scattered in different sources and forms and improving it in colloquial terms, a multi-source database is formed. Intelligent question and answering through the multi-source database can effectively deal with oral questions and obstacles to understanding professional terms, thereby improving the user experience of the elderly.

[0004] According to one aspect of the present application, a method for processing intelligent questions and answers for the elderly is provided, the method comprising: Collect multiple original medical documents containing health knowledge about the elderly; For any original medical field document, convert the original medical field document into multiple forms of text, wherein the form text includes terminology and explanation form text, question-answer form text and hierarchical summary form text, and the text content in each form text is colloquial language; Construct a multi-source database based on multiple forms of text corresponding to multiple original medical documents; receiving a medical question raised by an elderly user, and obtaining a question sentence vector based on the medical question; Based on the multi-source database and the question sentence vector, an answer sentence corresponding to the medical question is generated, and the answer sentence is rewritten in colloquial language and the rewritten answer sentence is returned.

[0005] According to another aspect of the present application, a device for intelligent question-answering processing for the elderly is provided, the device comprising: The medical document collection module is used to collect multiple original medical documents containing health knowledge for the elderly; A formal text conversion module is used to convert any original medical field document into multiple formal texts, wherein the formal texts include terminology and explanation formal texts, question-answer formal texts and hierarchical summary formal texts, and the text content in each formal text is colloquial language; A multi-source database construction module is used to construct a multi-source database based on multiple forms of text corresponding to multiple original medical documents; A medical question processing module, used for receiving medical questions raised by elderly users, and obtaining question sentence vectors based on the medical questions; The answer statement generation and return module is used to generate an answer statement corresponding to the medical question based on the multi-source database and the question sentence vector, rewrite the answer statement into colloquial language, and return the rewritten answer statement.

[0006] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the above-mentioned intelligent question and answer processing method for the elderly is implemented.

[0007] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, and the processor implements the above-mentioned intelligent question-and-answer processing method for the elderly when executing the program.

[0008] Through the above-mentioned technical scheme, the present application provides a method and device for intelligent question and answer processing for the elderly, a storage medium, and a computer device. By effectively integrating the health knowledge of the elderly scattered in different sources and forms and improving it in colloquial terms, a multi-source database is formed. Intelligent question and answering is performed through the multi-source database, which can effectively deal with oral questions and obstacles to understanding professional terms, thereby improving the user experience of the elderly.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] 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: Figure 1 A schematic diagram of a process flow of an intelligent question-answering method for the elderly provided in an embodiment of the present application is shown; Figure 2 A schematic diagram of a process of collecting medical documents provided in an embodiment of the present application is shown; Figure 3 A schematic diagram of a multi-source database construction method provided in an embodiment of the present application is shown; Figure 4 A schematic diagram of a process flow of a formal text conversion method provided in an embodiment of the present application is shown; Figure 5 A schematic diagram of a process for handling medical problems provided in an embodiment of the present application is shown; Figure 6 A schematic diagram showing a flow chart of a method for generating an answer statement provided in an embodiment of the present application is shown; Figure 7 A schematic diagram showing a flow chart of another answer sentence generating method provided in an embodiment of the present application; Figure 8 A schematic diagram of the structure of an intelligent question-answering processing device for the elderly provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0011] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0012] In this embodiment, a method for processing intelligent questions and answers for the elderly is provided. Figure 1 As shown, the method includes: Step 101, collect multiple original medical field documents containing health knowledge for the elderly.

[0013] At present, when the elderly use the elderly health knowledge Q&A robot to answer medical questions, they face a series of challenges (the elderly health knowledge Q&A robot is an emerging intelligent auxiliary tool that can provide health consultation and knowledge answering services for the elderly), mainly including: Fragmented knowledge: Knowledge related to the health of the elderly is scattered in various sources such as e-books, public account articles, blogs, etc. These data are diverse in form and inconsistent in structure, making it difficult to integrate knowledge.

[0014] Users’ questions are colloquial: As the main user group, the elderly often ask questions in a colloquial and lifelike manner, which is significantly different from the written professional literature in terms of expression style.

[0015] Barriers to professional terminology: The elderly generally lack knowledge of professional medical terminology, and existing health knowledge materials are filled with a large number of professional terms, which increases the difficulty for the elderly to understand and use these resources.

[0016] Limitations of the retrieval system: The traditional retrieval system database may not have an answer that completely matches the question, and the answer is unnatural. The large language model gives natural answers, but is prone to incorrect answers. For example, the RAG+ (Retrieval-Augmented Generation) large language model has relatively low recall and precision.

[0017] In the above embodiments of the present application, it can be applied to the health knowledge question-and-answer robot for the elderly. Through the large model offline corpus organization, structured storage, multi-source matching and large model reordering and other technical means, the data utilization rate, answer coverage and accuracy can be improved, thereby providing more accurate and convenient health knowledge question-and-answer services for the elderly. Specifically, based on different sources (e-books, public accounts, blogs, etc.), multiple original medical field documents containing health knowledge for the elderly are collected to prepare for the establishment of a subsequent multi-source database.

[0018] Optionally, in step 101, in collecting a plurality of original medical field documents containing health knowledge for the elderly, refer to Figure 2 As shown, specifically including: Step 1011, collecting multiple original candidate medical field documents containing health knowledge for the elderly.

[0019] Step 1012, for any original candidate medical field document, obtain the evaluation value of the original candidate medical field document according to the preset evaluation indicators and the indicator weights corresponding to the preset evaluation indicators, and obtain the original medical field document corresponding to the evaluation value greater than the preset evaluation threshold, wherein the preset evaluation indicators include the document collection source, the document release time and the authority of the document source.

[0020] In the above-mentioned embodiments of the present application, original candidate medical field documents containing health knowledge for the elderly are widely collected through various channels. These channels may include: academic databases, such as PubMed, CNKI, etc., which include a large number of medical papers and research reports. Government and medical institution websites, such as the official websites of the World Health Organization and major hospitals, will publish authoritative health guidelines and policy documents. Professional medical websites and forums, such as Medical World, Dingxiangyuan, etc., bring together many medical professionals and patients for communication and discussion. It can also be collected from channels such as e-books, public accounts, blogs, etc. to expand the scope of collection and enrich the original data.

[0021] Next, set the evaluation indicators. The preset evaluation indicators are used to evaluate the quality and reliability of the original candidate medical documents. According to actual needs, the following evaluation indicators and indicator weights can be set: document collection source (30%), document release time (20%), considering the timeliness of the document, newer documents often contain more cutting-edge and accurate information, document source authority (50%), evaluate the authority and credibility of the document publishing agency, such as government agencies, well-known medical institutions, etc. are usually more authoritative.

[0022] For each original candidate medical field document, the evaluation value is obtained by calculating the preset evaluation index and the corresponding index weight. The specific steps are as follows: Scoring the sources of document collection: for example, papers in academic databases get 30 points (full score), documents published on government and medical institution websites get 25 points, and documents on professional medical websites and forums get 20 points.

[0023] Score the document release time: Score based on the difference between the document release time and the current time. For example, if it was released within the past year, it will get 20 points; if it was released within one to three years, it will get 15 points; and if it was released more than three years ago, it will get 10 points.

[0024] The authority of the document source is scored: documents released by government agencies are given 50 points (full score), documents released by well-known medical institutions are given 45 points, authoritative papers in academic databases are given 40 points, and other reliable sources are given 35 points.

[0025] For example, if a research paper on diabetes management for the elderly is collected, assuming that the document collection source is an academic database, the release time is within the past year, and the document source authority is an authoritative paper in the academic database, the evaluation value is calculated as follows: Evaluation value = 30 (document collection source) × 30% + 20 (release time) × 20% + 40 (document source authority) × 50% = 27 + 4 + 20 = 51 (points). Set a preset evaluation threshold, such as 50 points, and then filter out original medical documents with evaluation values ​​greater than the preset evaluation threshold. These documents are high-quality, reliable original medical documents containing health knowledge for the elderly.

[0026] In the above example, the evaluation value of the research paper on diabetes management for the elderly is 51 points, which is higher than the preset evaluation threshold of 50 points. Therefore, the research paper on diabetes management for the elderly is selected as the original medical field document. Through this process, multiple high-quality original medical field documents containing health knowledge for the elderly can be effectively collected and screened, improving the accuracy of subsequent answer matching.

[0027] Step 102, for any original medical field document, convert the original medical field document into multiple formal texts, wherein the formal texts include terminology and explanation formal texts, question-answer formal texts and hierarchical summary formal texts, and the text content in each formal text is spoken language.

[0028] Step 103: construct a multi-source database based on the multiple forms of text corresponding to the multiple original medical field documents.

[0029] Then, if Figure 3 As shown, for any original medical domain document, the original medical domain document is preprocessed, including extracting medical term entity words and their explanations, questions (questions that can be asked) and their answers (answer sentences), document fragments and their summaries (summary content), so as to convert the original medical domain document into multiple forms of text, including term and explanation formal text, question and answer formal text and hierarchical summary formal text. Based on the multiple forms of text corresponding to multiple original medical domain documents, a multi-source database is constructed to prepare for the subsequent generation of answer sentences.

[0030] In particular, the text content in each formal text is a colloquial language, so that the elderly health knowledge question-and-answer robot can better "understand" the colloquial questions of the elderly and generate colloquial answers that the elderly can understand. Specifically, after obtaining the formal text, the natural language processing technology can be used to analyze and process the formal text. For example, the text can be preprocessed using tools such as word segmentation and part-of-speech tagging, and then the content in the formal text can be improved in colloquial language in combination with the rules and strategies of colloquial conversion. After the improvement is completed, manual review and correction can be performed to ensure that the improved text meets the requirements of colloquial language. You can also learn how to make better improvements by finding examples of colloquial conversion. These examples can come from resources such as daily conversations, oral teaching materials, or colloquial writing guidance. It should be noted that when performing colloquial conversion, appropriate adjustments need to be made according to the target audience and context.

[0031] Optionally, in step 102, for any original medical field document, the original medical field document is converted into a plurality of forms of text, referring to Figure 4 As shown, specifically including: Step 1021, for any original medical field document, perform word segmentation on the original medical field document, determine multiple medical term entity words from the segmented multiple words, obtain the corresponding explanations of each medical term entity word, construct term explanation pairs based on the medical term entity words and the explanations corresponding to the medical term entity words, rewrite the term explanation pairs in colloquial language, and obtain the term and explanation form text corresponding to the original medical field document based on the rewritten term explanation pairs.

[0032] Step 1022, for any original medical field document, generate answer sentences that can be used to answer in the original medical field document and the askable questions that can be answered by the answer sentences based on the preset medical field question and answer model, obtain question and answer pairs based on the askable questions and the answer sentences corresponding to the askable questions, rewrite the question and answer pairs in colloquial language, and obtain the question and answer pair formal text corresponding to the original medical field document based on the rewritten question and answer pairs, wherein the preset medical field question and answer model is established based on LLaMA-13B.

[0033] Step 1023, for any original medical field document, the original medical field document is divided into multiple fragment contents based on a text segmentation algorithm, and each fragment content is summarized based on a text summarization algorithm to obtain the summary content corresponding to each fragment content, and a fragment summary content pair is obtained based on the fragment content and the summary content corresponding to the fragment content. After the fragment summary content pair is colloquially rewritten, a hierarchical summary text corresponding to the original medical field document is obtained based on the rewritten fragment summary content pair.

[0034] In the above-mentioned embodiments of the present application, specifically, the formal text includes terminology and explanation formal text, question-answer formal text and hierarchical summary formal text.

[0035] For the terminology and explanation form text, the original medical domain document contains text content such as: "Heart disease is a common circulatory system disease, which is mainly manifested by symptoms such as palpitations and chest pain." The original medical domain document is segmented to obtain: "Heart disease is a common circulatory system disease, which is mainly manifested by palpitations, chest pain, etc. symptoms." Among the multiple segmented words, multiple medical term entity words are determined, such as: "heart disease", "circulatory system", "palpitations", and "chest pain". Obtain the explanation of each medical term entity word: "heart disease" refers to a disease caused by abnormal heart structure or function; "circulatory system" refers to a system composed of the cardiovascular system and the lymphatic system; "palpitations" refers to the feeling of accelerated, strengthened or irregular heartbeats; "chest pain" refers to the symptom of chest pain. Then, construct the term explanation pair: ["heart disease-disease caused by abnormal heart structure or function", "circulatory system-system composed of the cardiovascular system and the lymphatic system", "palpitations-feeling of accelerated, strengthened or irregular heartbeats", "chest pain-symptoms of chest pain"]. After rewriting the term explanation into colloquial language, we get: "Heart disease means that there is something wrong with the heart, either the structure is wrong or the function is not good; the circulatory system is the cardiovascular and lymphatic systems in our body; palpitations mean that you feel your heart beats very fast, very hard, or irregularly; chest pain means pain in the chest." The final generated term and explanation form text can be: "Heart disease is a common disease, which means that there is something wrong with the heart, which may be a structural problem or a functional problem. The main manifestations are palpitations, which means that you feel your heart beats very fast, very hard, or irregularly, and chest pain, which means pain in the chest." For question-answer pair text, the original medical document is such as: "Diabetes is a chronic disease, which is mainly manifested as high blood sugar. It is usually caused by insufficient insulin secretion or weakened cell response to insulin. Long-term high blood sugar may cause damage to blood vessels, nerves and other tissues." Using the preset medical field question-answering model, the content of the original document can be analyzed and multiple possible answer sentences can be generated, as well as the questions that these answer sentences can answer. For example: Answer 1: Chronic disease. Answer 2: High blood sugar. Answer 3: Usually caused by insufficient insulin secretion or weakened cell response to insulin. Answer 4: May cause damage to blood vessels, nerves and other tissues. These answers can be used to answer: Question 1, what type of disease is diabetes? Question 2, what are the main manifestations of diabetes? Question 3, what causes diabetes? Question 4, what are the consequences of long-term high blood sugar? For each question, combine the question and answer to form a question-answer pair. For example: Question-answer pair 1: Question, what type of disease is diabetes? Answer, chronic disease. In order to make the question-answer pair easier to understand and communicate, the question-answer pair can be rewritten in a colloquial way. For example: Rewritten question-answer pair 1: Question: What type of disease is diabetes? Answer: It is a chronic disease. Finally, combine the rewritten question-answer pairs according to a certain format (such as list, paragraph, etc.) to form a question-answer pair text.

[0036] In particular, the medical field question-answering model can be built based on LLaMA-13B, which is a large natural language processing model with powerful language understanding and generation capabilities. It is one of the Llama series models released by the Meta AI team and contains 13 billion parameters. LLaMA-13B adopts the Transformer Decoder architecture and has made many improvements to optimize performance. It can be incrementally pre-trained on a large corpus, thus having a wide range of knowledge backgrounds. LLaMA-13B has high performance and scalability, and can be applied to various specific fields. Through customized training for domain knowledge, it can improve the understanding and question-answering capabilities in the field. For example, in the medical field, the MedicalGPT (Medical Generative Pre-trainedTransformer, the full name of which is Medical Generative Pre-trained Transformer) model built on LLaMA-13B can accurately understand and answer medical-related questions through multiple stages of training such as secondary pre-training, supervised fine-tuning, reward modeling, and reinforcement learning training, providing a convenient communication bridge for doctors and patients.

[0037] For hierarchical summary text, the original medical field document is as follows: "Diabetes is a chronic metabolic disease that mainly affects the blood sugar level in the body. It is divided into two types: type 1 diabetes and type 2 diabetes. Type 1 diabetes usually occurs in children and adolescents, due to an absolute lack of insulin secretion. Type 2 diabetes is more common and occurs mostly in adults, due to a relative lack of insulin secretion or a weakened response of the body to insulin (insulin resistance). Symptoms of diabetes include polyuria, thirst, weight loss, etc. If not treated in time, diabetes can cause a variety of complications, such as cardiovascular disease, retinopathy, neuropathy and kidney disease." First, use the text segmentation algorithm to segment the original document into multiple fragments. These fragments can be segmented based on sentences, paragraphs, or topics. In this example, for example, by segmenting by paragraph, a total of 6 fragments can be obtained, two of which are, for example: Clip 1: Diabetes is a chronic metabolic disease that primarily affects blood sugar levels in the body.

[0038] Clip 2: It is divided into two types: type 1 diabetes and type 2 diabetes.

[0039] Next, use a text summarization algorithm to summarize the content of each segment. These summaries should convey the main information of the segment succinctly and clearly. For example: Segment 1 Summary (summary content): Diabetes affects blood sugar levels. Segment 2 Summary (summary content): Diabetes is divided into type 1 and type 2.

[0040] Combine each segment with its corresponding summary (summary content) to form a segment-summary content pair. For example: Segment-summary content pair 1: Segment content: Diabetes is a chronic metabolic disease that mainly affects blood sugar levels in the body. Summary content: Diabetes affects blood sugar levels.

[0041] In order to make the fragment summary easier to understand and communicate, you can rewrite it in a colloquial way. For example: Rewritten fragment summary 1: Fragment content (colloquial): Diabetes is a chronic disease that affects blood sugar levels. Summary content (colloquial): That is, diabetes is related to blood sugar levels.

[0042] Finally, the rewritten fragment summary content pairs are combined in a logical order to form a hierarchical summary text, which can be organized according to the structure or theme of the original document, for example.

[0043] To this end, since the health knowledge of the elderly is scattered in a variety of documents with different formats and structures, the data format can be unified through preprocessing and structured storage, which facilitates subsequent retrieval and matching and improves the availability of data.

[0044] Step 104: receiving medical questions raised by elderly users, and obtaining question sentence vectors based on the medical questions.

[0045] Step 105, based on the multi-source database and the question sentence vector, generate an answer sentence corresponding to the medical question, rewrite the answer sentence into colloquial language, and return the rewritten answer sentence.

[0046] Next, the medical questions raised by elderly users are received, and the question sentence vectors are obtained. Based on the multi-source database and the question sentence vectors, the answer sentences corresponding to the medical questions are generated, and the answer sentences are rewritten in colloquial language and the rewritten answer sentences are returned.

[0047] Optionally, in step 104, a medical question raised by an elderly user is received, and based on the medical question, a question sentence vector is obtained, referring to Figure 5 , specifically including: Step 1041, receiving medical questions raised by elderly users, segmenting the medical questions to obtain multiple question words, performing vectorization operations on each question word to obtain a question word vector, and combining and encoding the question word vectors of each question word to obtain a question sentence vector.

[0048] In the above embodiment of the present application, medical problems raised by elderly users are received, the medical problems are segmented, and multiple problem words are obtained. After obtaining the problem word vectors corresponding to each problem word, each problem word vector is encoded and combined using the Transformer decoder architecture to obtain a problem sentence vector. Specifically, for example, a medical problem raised by an elderly user is received, such as "I have been dizzy recently. Is my high blood pressure recurring? What should I do?" The received medical problem is parsed to extract key information, such as segmenting the problem, removing stop words, and other pre-processing steps. For the above example problem, the following key information (i.e., problem words) can be parsed: dizziness, high blood pressure, what to do.

[0049] After obtaining the problem words, you can use the preset word vector model to match the problem word vector corresponding to each problem word. Specifically, you can choose a suitable word vector model, such as Word2Vec, GloVe, etc. These models have been trained on large-scale corpora, including word vectors of multiple words, and can accurately represent the semantic information of words.

[0050] That is, after dividing the sentence into several words A, B, and C, vectors a, b, and c are obtained respectively, and vectors a, b, and c are combined for encoding (for example, using a Transformer decoder for combined encoding) to obtain a vector x to represent the question sentence.

[0051] Optionally, in step 105, based on the multi-source database and the question sentence vector, an answer sentence corresponding to the medical question is generated, and the answer sentence is rewritten into colloquial language and then returned to the rewritten answer sentence, with reference to Figure 6 As shown, specifically including Step 1051, in a multi-source database, obtain the medical term entity word vectors corresponding to the medical term entity words in the term and explanation formal text, the questionable question word vectors corresponding to each questionable question word segmented from the questionable question in the question-answer formal text, and the answer sentence word vectors corresponding to each answer sentence word segmented from the answer sentence, and the fragment content word vectors corresponding to each fragment content word segmented from the fragment content in the hierarchical summary formal text.

[0052] In the above embodiment of the present application, in a multi-source database, the medical term entity word vectors corresponding to the medical term entity words in the term and explanation form text are obtained, the question word vectors corresponding to each questionable question word segmented from the questionable question in the question-answer form text, and the answer sentence word vectors corresponding to each answer sentence word segmented from the answer sentence, the fragment content word vectors corresponding to each fragment content word segmented from the fragment content in the hierarchical summary form text, and specifically the word vectors can be obtained through vectorization operations.

[0053] Step 1052, after adding a term explanation tag to the medical term entity word vector, adding a question and answer tag to the questionable question word vector and the answer sentence word vector, and adding a fragment summary content tag to the fragment content word vector, an answer word vector index library is constructed based on each word vector with the tag added.

[0054] Next, based on the medical term entity word vectors, question word vectors, answer sentence word vectors and fragment content word vectors, an answer word vector index library is constructed to prepare for subsequent answer matching. In particular, the answer word vector index library is divided into three types of labels, namely term explanation labels, question and answer labels and fragment summary content labels. Different types of word vectors are added with different labels.

[0055] Step 1053, for the word vectors under any label in the answer word vector index library, respectively calculate the similarity between each word vector and the question sentence vector, and based on the calculated similarity, determine multiple initial candidate answer word vectors under the label.

[0056] Step 1054, when the medical term entity word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the term interpretation pair corresponding to the medical term entity word vector; when the questionable question word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the question-answer pair corresponding to the questionable question vector; when the answer sentence word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the question-answer pair corresponding to the answer sentence word vector; when the fragment content word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the fragment summary content pair corresponding to the fragment content word vector.

[0057] Suppose the question asked by an elderly user is: "What is diabetes?" First, this question is segmented to obtain a word sequence such as [what / is / diabetes], and then these words are converted into corresponding word vectors. A pre-trained language model (such as Word2Vec, BERT, etc.) can be used to obtain word embedding representation (word vector), and then these word vectors are encoded and combined to obtain the question sentence vector.

[0058] For the word vectors under any label in the answer word vector index library, for example, for the various word vectors under the term explanation label, calculate the similarity between each word vector and the question sentence vector. The similarity calculation can be calculated by calculating cosine similarity, Euclidean distance, etc. Cosine similarity is a commonly used method. Its value range is [-1,1]. The larger the value, the more similar the two vectors are. The similarity calculation formula is as follows:

[0059] Among them, v1 and v2 represent the word vector and the question sentence vector respectively, v1·v2 represents the inner product (dot product) of vector v1 and vector v2, ||v1|| and ||v2|| represent the modulus (Euclidean norm) of vector v1 and vector v2 respectively.

[0060] Next, select the word vectors with the highest similarity (which can be set to other thresholds such as 3, 5, etc.) as candidate answer word vectors. Specifically: For medical term entity word vectors: If the medical term entity word vector is determined as the initial candidate answer word vector (such as the word vector of "insulin"), then the explanation of the term (insulin) is found through the term explanation pair where insulin is located, and the word vectors corresponding to the medical term entity word and the explanation are used together as the candidate answer word vectors for subsequent answer generation.

[0061] For questionable question word vectors: If the initial candidate answer word vector corresponds to a questionable question (such as "What are the symptoms of diabetes?"), then the answer can be retrieved from the question-answer pairs in the multi-source database based on this question, and the word vectors corresponding to the question and the answer can be used together as the candidate answer word vector for subsequent answer generation.

[0062] For the answer sentence word vector: If the initial candidate answer word vector directly corresponds to the answer to a question (such as "diabetes is a chronic disease"), then the question can be retrieved from the question-answer pair in the multi-source database based on this answer, and the word vectors corresponding to the answer and the question can be used together as the candidate answer word vector for subsequent answer generation.

[0063] For the fragment content word vector: If the initial candidate answer word vector is a fragment content word vector, then the candidate answer word vector used to generate the answer can be determined based on the word vector corresponding to the fragment summary content pair of the fragment content corresponding to the fragment content word vector.

[0064] For example, we can find several question and answer pairs (Q&A pairs) that are highly relevant to "diabetes", such as "What are the symptoms of diabetes? (Answers: polyuria, thirst, weight loss, etc.)" and "What are the treatments for diabetes? (Answers: medication, diet control, exercise, etc.)". Then, we can select the most appropriate answer from these candidate answers and present it to the elderly user based on the specific context and intent of the elderly user's medical question. This process combines multiple technologies such as information retrieval, word vector matching, and semantic understanding to achieve accurate and efficient medical Q&A services.

[0065] Step 1055, obtain the candidate answer word vectors corresponding to various labels through a preset similarity scoring model, and score the similarity between them and the question sentence vector respectively, sort the candidate answer word vectors according to the obtained similarity scores, and determine the target answer word vector based on the sorting results.

[0066] Step 1056, fill the target answer word corresponding to the target answer word vector into the preset answer generation template to obtain the answer sentence corresponding to the medical question, rewrite the answer sentence into colloquial language, and return the rewritten answer sentence, wherein the preset answer generation template includes a plurality of connectives rewritten into colloquial language, and the connectives are used to connect the answer words.

[0067] Suppose an elderly user asks a medical question: "How to prevent diabetes?" The candidate answer word vectors are, for example: ["control diet", "increase exercise", "check blood sugar regularly", "avoid high-sugar foods"], etc.

[0068] In order to determine which candidate answers are more relevant and should be presented to elderly users first, the preset similarity scoring model can be used again to obtain the candidate answer word vectors corresponding to various labels, and the similarity scores between them and the question sentence vectors. To this end, the accuracy of answer sentence generation can be improved by scoring and sorting the three candidate answer word vectors again. After the preset similarity scoring model is trained, by inputting the candidate answer word vectors to be scored and the question sentence vector, the similarity scores of each candidate answer word vector can be output.

[0069] Next, the candidate answer word vectors are sorted according to the obtained similarity score, and the target answer word vectors corresponding to the scores of the preset number (such as 3 or other thresholds) before the score ranking are determined. In particular, for example, the top three with the highest scores are taken, and multiple candidate answer word vectors may have the same score, then all candidate answer vectors with the top three scores are used as target answer word vectors. In order to convert the target answer word vector into an answer sentence that is easy for elderly users to understand, a preset answer generation template can be used to generate an answer sentence. This preset answer generation template contains multiple connecting words that connect the answer words. These connecting words use colloquial language, such as "want...", "also...", "and", etc., to make the answer more natural and fluent.

[0070] For example, assuming that the target answer word vector is ["control diet", "increase exercise", "check blood sugar regularly"], the generated answer sentence may be: "To prevent diabetes, you need to control your diet, increase exercise, and check blood sugar regularly." For another example, a preset answer generation template is as follows: "The [symptom] you mentioned may be due to [reason 1], [reason 2] or [reason 3]. It is recommended that you [suggest action]." Among them, [Symptoms], [Reason 1], [Reason 2], [Reason 3], and [Suggested Action] are the slots that need to be filled.

[0071] Fill the corresponding slots in the template with the terms or phrases (target answer terms) corresponding to the target answer word vectors. For example, if the target answer word vectors are v{arrhythmia}, v{anemia}, and v{anxiety}, the corresponding terms "arrhythmia", "anemia", and "anxiety" can be filled into the template.

[0072] After filling the slots, to ensure that the answer sentence is colloquial, you can choose colloquial conjunctions (such as "because", "or", "suggest you") and appropriate sentence structures. For example, the generated answer sentence may be: "The panic you mentioned may be caused by arrhythmia, anemia or anxiety. It is recommended that you seek medical attention in time to determine the specific cause." Next, the answer sentence is rewritten in a colloquial way. For example, tools such as word segmentation and part-of-speech tagging can be used to pre-process the text, and then the content in the formal text can be improved in a colloquial way by combining the rules and strategies of colloquial conversion. Specifically, written words can be replaced with colloquial words using more natural, friendly language suitable for daily conversations. For example, "maybe because" can be changed to "maybe", and "it is recommended to seek medical attention in time" can be changed to "it is better to go to the hospital to have a look", so that the sentence is more fluent and easy to understand. For example, the sentence of cause and effect can be split into two independent sentences to make the expression clearer. At the same time, modal particles such as "ah" and "bar" can be added to make the conversation more natural. Finally, a colloquial answer sentence is generated and returned to the elderly user. The generated colloquial answer sentence is as follows: Your palpitations may be caused by arrhythmia, anemia or anxiety. It is best to go to the hospital to determine the specific cause.

[0073] For another example, you can also set a prompt template, fill in the question into the prompt template, and input the prompt into the spoken language rewriting model for rewriting. The prompt template is, for example: [Task] Rewrite this answer into a more natural and friendly language suitable for daily conversation; [Example] Your palpitations may be caused by arrhythmia, anemia or anxiety. It is best to go to the hospital to determine the specific cause; [Answer to be rewritten] It is recommended to seek medical attention promptly; [Answer rewrite result]: You'd better go to the hospital as soon as possible.

[0074] Through this process, the original answer sentences are rewritten into more colloquial, natural and friendly expressions that are more suitable for daily conversations, so that elderly users can understand their health conditions in a more understandable way and take corresponding actions. In this way, the most relevant and useful information can be selected from the candidate answers based on the medical questions raised by elderly users, and presented to elderly users in a natural and smooth way.

[0075] By applying the technical solution of this embodiment, the original medical field documents of health knowledge for the elderly scattered in different sources and forms are converted into various forms of text and effectively integrated into structured data to build a multi-source database, which can effectively search for efficiency and understanding ability, and when facing questions from the elderly, can effectively match the written and terminological content in professional literature. The professional terminology barriers encountered by the elderly when using the question-and-answer robot are reduced, and the understanding and acceptance of health knowledge are improved. The retrieval accuracy of the question-and-answer robot for questions asked by the elderly is improved, and high-quality and high-similarity answers are generated. Make full use of the health data of the elderly from different channels, including medical term entity words, questions, answers, document fragments and summaries, etc., to improve the coverage and accuracy of the question-and-answer robot's question-and-answer system, and finally, provide a health knowledge question-and-answer system that can better serve the elderly population, and improve the effectiveness and user satisfaction of the elderly in practical applications.

[0076] In a specific embodiment, for example Figure 7 As shown in the figure, according to the medical questions raised by the elderly users, the question sentence vector q is obtained, and the similarity is calculated with the word vectors of each label in the answer word vector index library, and the topk candidate answer word vectors that best match the question word vector q under each label are selected. The topk candidate answer word vectors corresponding to each label are scored for similarity, and re-sorted according to the scores to select the top3 most relevant target answer word vectors. In particular, prompt words (connecting words) can also be designed to combine the questions asked by the elderly users with the top3 candidate answer word vectors to generate answer sentences, and the answer sentences are rewritten in colloquial language and returned. Through structured storage, multi-source matching and re-sorting mechanisms, data utilization can be improved, matching clues can be found from multiple angles, and the most accurate information can be filtered out through large models to generate accurate answers.

[0077] Specifically, the medical questions raised by elderly users are processed by the semantic model to obtain the question sentence vector q, and [medical term entity word + explanation, term explanation pair] is processed by the semantic model to obtain the medical term word vector E. The questionable questions in [questionable questions + answer sentences, question-answer pairs] are processed by the semantic model to obtain the questionable question word vector Q. The answer sentences in [questionable questions + answer sentences, question-answer pairs] are processed by the semantic model to obtain the answer sentence word vector A. The fragment content in [fragment content + summary content, hierarchical summary] is processed by the semantic model to obtain the fragment content word vector D.

[0078] E, Q, A, and D are built into an answer word vector index library. The answer word vector index library can quickly convert the natural language questions of elderly users into vector form, match them with the word vectors in the multi-source database, and improve the efficiency and accuracy of retrieval. Calculate the similarity between the question sentence vector and the word vectors corresponding to each label in the answer word vector index library, and select the topk candidate answer word vectors closest to q under each label. If E is included in the candidate, the corresponding entity and explanation are placed in the candidate. If Q is included in the candidate, the corresponding A is also added to the candidate. If A is included in the candidate, the corresponding Q is also added to the candidate. If D is included in the candidate, the corresponding summary (summary content) is also added to the candidate. Through multi-source matching, data from different sources can be fully utilized to improve the coverage and diversity of answers.

[0079] The top k candidates under each label are scored for similarity, and re-ranked according to the scores to select the top 3 most relevant candidates, i.e., the targets. Re-ranking can further filter out the candidates most relevant to the elderly user's questions, i.e., the targets, and improve the accuracy of the answers.

[0080] Design prompt words, add q and the top 3 targets (E / Q / A / D) to the prompt words, and get the answer sentence. The steps of generating answer sentences ensure the coherence and accuracy of the answers, and can provide personalized answers according to the context. To this end, when faced with problems such as fragmented knowledge, colloquial questions from elderly users, and professional terminology barriers. Through these technical solutions, the integration and utilization of data can be improved, the accuracy of retrieval and matching can be improved, and finally health knowledge answers that are more in line with the needs of the elderly can be generated, thereby improving the user experience and service quality of the elderly.

[0081] By applying the technical solution of this embodiment, the efficiency and accuracy of the health knowledge question and answer system for the elderly are improved through structured storage and multi-source matching technology. The scattered data is effectively integrated and the data utilization rate is improved. At the same time, the user experience of the elderly is enhanced by understanding colloquial questions and reducing the barriers of professional terminology. In addition, the application of large models optimizes information screening and improves the accuracy of answers. It not only saves resources, but also promotes the updating and iteration of knowledge, improves the level of intelligence in the question and answer process, and provides more convenient and accurate health consultation services for the elderly.

[0082] Further, as Figure 1 The specific implementation of the method, the embodiment of the present application provides an intelligent question-answering processing device for the elderly, such as Figure 8 As shown, the device comprises: The medical field document collection module 201 is used to collect a plurality of original medical field documents containing health knowledge for the elderly; The formal text conversion module 202 is used to convert any original medical field document into multiple formal texts, wherein the formal texts include terminology and explanation formal texts, question-answer formal texts, and hierarchical summary formal texts, and the text content in each formal text is a spoken language; A multi-source database construction module 203 is used to construct a multi-source database based on multiple forms of text corresponding to multiple original medical field documents; A medical question processing module 204 is used to receive medical questions raised by elderly users and obtain a question sentence vector based on the medical questions; The answer statement generation and return module 205 is used to generate an answer statement corresponding to the medical question based on the multi-source database and the question sentence vector, rewrite the answer statement into colloquial language, and return the rewritten answer statement.

[0083] Optionally, the formal text conversion module 202 is further used for: Segmenting the original medical document, determining multiple medical term entity words from the segmented multiple words, obtaining explanations corresponding to each medical term entity word, constructing term explanation pairs based on the medical term entity words and the explanations corresponding to the medical term entity words, rewriting the term explanation pairs in colloquial language, and obtaining term and explanation form texts corresponding to the original medical document based on the rewritten term explanation pairs; Generate answer sentences that can be used for answering in the original medical field document and the questionable questions that can be answered by the answer sentences based on the preset medical field question-answering model, obtain question-answer pairs based on the questionable questions and the answer sentences corresponding to the questionable questions, rewrite the question-answer pairs in colloquial language, and obtain the question-answer pair format text corresponding to the original medical field document based on the rewritten question-answer pairs, wherein the preset medical field question-answering model is established based on LLaMA-13B; The original medical field document is segmented into multiple fragment contents based on a text segmentation algorithm, and each fragment content is summarized based on a text summarization algorithm to obtain the summary content corresponding to each fragment content. A fragment summary content pair is obtained based on the fragment content and the summary content corresponding to the fragment content. The fragment summary content pair is rewritten in colloquial language, and a hierarchical summary text corresponding to the original medical field document is obtained based on the rewritten fragment summary content pair.

[0084] Optionally, the medical problem processing module 204 is further used to: The medical question is segmented to obtain multiple question words, each question word is vectorized to obtain a question word vector, and the question word vectors of each question word are combined and encoded to obtain a question sentence vector.

[0085] Optionally, the answer statement generating and returning module 205 is further used to: In a multi-source database, obtain the medical term entity word vectors corresponding to the medical term entity words in the term and explanation form text, the questionable question word vectors corresponding to each questionable question word segmented from the questionable question in the question-answer pair form text, and the answer sentence word vectors corresponding to each answer sentence word segmented from the answer sentence, and the fragment content word vectors corresponding to each fragment content word segmented from the fragment content in the hierarchical summary form text; After adding a term explanation tag to the medical term entity word vector, adding a question and answer tag to the questionable question word vector and the answer sentence word vector, and adding a segment summary content tag to the segment content word vector, an answer word vector index library is constructed based on each of the labeled word vectors; Based on the answer word vector index library and the question sentence vector, an answer sentence corresponding to the medical question is generated.

[0086] Optionally, the answer statement generating and returning module 205 is further used to: For the word vectors under any label in the answer word vector index library, respectively calculate the similarity between each word vector and the question sentence vector, and determine multiple initial candidate answer word vectors under the label based on the calculated similarity; When the medical term entity word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the term interpretation pair corresponding to the medical term entity word vector; when the questionable question word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the question-answer pair corresponding to the questionable question vector; when the answer sentence word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the question-answer pair corresponding to the answer sentence word vector; when the fragment content word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the fragment summary content pair corresponding to the fragment content word vector; Based on the candidate answer word vectors of various tags, the target answer word vector is determined, and the answer sentence corresponding to the medical question is obtained based on the target answer word vector.

[0087] Optionally, the answer statement generating and returning module 205 is further used to: The candidate answer word vectors corresponding to various labels are obtained through a preset similarity scoring model, and the similarity scores between them and the question sentence vectors are scored. The candidate answer word vectors are sorted according to the obtained similarity scores, and the target answer word vector is determined based on the sorting results; The target answer words corresponding to the target answer word vector are filled into a preset answer generation template to obtain an answer sentence corresponding to the medical question, wherein the preset answer generation template includes a plurality of connecting words rewritten in colloquial language, and the connecting words are used to connect the answer words.

[0088] Optionally, the medical field document collection module 201 is further used for: Collect multiple original candidate medical field documents containing health knowledge of the elderly; For any original candidate medical field document, the evaluation value of the original candidate medical field document is obtained according to the preset evaluation indicators and the indicator weights corresponding to the preset evaluation indicators, and the original medical field document corresponding to the evaluation value greater than the preset evaluation threshold is obtained, wherein the preset evaluation indicators include the document collection source, the document release time and the authority of the document source.

[0089] It should be noted that for other corresponding descriptions of the functional units involved in the intelligent question-answering processing device for the elderly provided in the embodiment of the present application, reference can be made to Figure 1 , Figure 2 , Figures 4 to 6 The corresponding description in the method will not be repeated here.

[0090] Based on the above Figure 1 , Figure 2 , Figures 4 to 6 The method shown in the embodiment of the present application is accordingly provided with a storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned Figure 1 , Figure 2 , Figures 4 to 6 The intelligent question-answering method for the elderly is shown.

[0091] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of the present application.

[0092] Based on the above Figure 1 , Figure 2 , Figures 4 to 6 The method shown, and Figure 8 In order to achieve the above-mentioned purpose, the embodiment of the present application further provides a computer device, which can be a personal computer, a server, a network device, etc. The computer device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 , Figure 2 , Figures 4 to 6 The intelligent question-answering method for the elderly is shown.

[0093] Optionally, the computer device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a WI-FI interface), etc.

[0094] Those skilled in the art will appreciate that the computer device structure provided in this embodiment does not limit the computer device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0095] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and saves the hardware and software resources of the computer device, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to realize communication between the components inside the storage medium, and communication with other hardware and software in the physical device.

[0096] Through the description of the above implementation methods, technical personnel in this field can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, and can also be implemented through hardware by effectively integrating the health knowledge of the elderly scattered in different sources and forms and improving it in colloquial terms to form a multi-source database. Intelligent question and answering through the multi-source database can effectively deal with colloquial questions and obstacles to understanding professional terms, thereby improving the user experience of the elderly.

[0097] Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily necessary for implementing the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple submodules.

[0098] The above serial numbers of this application are only for description and do not represent the advantages and disadvantages of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of this application, but this application is not limited to them, and any changes that can be made by technicians in this field should fall within the scope of protection of this application.

Claims

1. An intelligent question-answering processing method for the elderly, characterized in that: The method comprises: Collect multiple original medical documents containing health knowledge about the elderly; For any original medical field document, convert the original medical field document into multiple forms of text, wherein the form text includes terminology and explanation form text, question-answer form text and hierarchical summary form text, and the text content in each form text is colloquial language; Construct a multi-source database based on multiple forms of text corresponding to multiple original medical documents; receiving a medical question raised by an elderly user, and obtaining a question sentence vector based on the medical question; Based on the multi-source database and the question sentence vector, an answer sentence corresponding to the medical question is generated, and the answer sentence is rewritten in colloquial language and the rewritten answer sentence is returned.

2. The method according to claim 1, characterized in that The original medical field documents are converted into terminology and explanation form text, including: Segmenting the original medical document, determining multiple medical term entity words from the segmented multiple words, obtaining explanations corresponding to each medical term entity word, constructing term explanation pairs based on the medical term entity words and the explanations corresponding to the medical term entity words, rewriting the term explanation pairs in colloquial language, and obtaining term and explanation form texts corresponding to the original medical document based on the rewritten term explanation pairs; The original medical field document is converted into a question-answer pair text, including: Generate answer sentences that can be used for answering in the original medical field document and the questionable questions that can be answered by the answer sentences based on the preset medical field question-answering model, obtain question-answer pairs based on the questionable questions and the answer sentences corresponding to the questionable questions, rewrite the question-answer pairs in colloquial language, and obtain the question-answer pair format text corresponding to the original medical field document based on the rewritten question-answer pairs, wherein the preset medical field question-answering model is established based on LLaMA-13B; The original medical field document is converted into a hierarchical summary form text, including: The original medical field document is segmented into multiple fragment contents based on a text segmentation algorithm, and each fragment content is summarized based on a text summarization algorithm to obtain the summary content corresponding to each fragment content. A fragment summary content pair is obtained based on the fragment content and the summary content corresponding to the fragment content. The fragment summary content pair is rewritten in colloquial language, and a hierarchical summary text corresponding to the original medical field document is obtained based on the rewritten fragment summary content pair.

3. The method according to claim 1, characterized in that Based on the medical question, a question sentence vector is obtained, including: The medical question is segmented to obtain multiple question words, each question word is vectorized to obtain a question word vector, and the question word vectors of each question word are combined and encoded to obtain a question sentence vector.

4. The method according to claim 1, characterized in that The generating an answer sentence corresponding to the medical question based on the multi-source database and the question sentence vector includes: In a multi-source database, obtain the medical term entity word vectors corresponding to the medical term entity words in the term and explanation form text, the questionable question word vectors corresponding to each questionable question word segmented from the questionable question in the question-answer pair form text, and the answer sentence word vectors corresponding to each answer sentence word segmented from the answer sentence, and the fragment content word vectors corresponding to each fragment content word segmented from the fragment content in the hierarchical summary form text; After adding a term explanation tag to the medical term entity word vector, adding a question and answer tag to the questionable question word vector and the answer sentence word vector, and adding a segment summary content tag to the segment content word vector, an answer word vector index library is constructed based on each of the labeled word vectors; Based on the answer word vector index library and the question sentence vector, an answer sentence corresponding to the medical question is generated.

5. The method according to claim 4, characterized in that The generating of an answer sentence corresponding to the medical question based on the answer word vector index library and the question sentence vector includes: For the word vectors under any label in the answer word vector index library, respectively calculate the similarity between each word vector and the question sentence vector, and determine multiple initial candidate answer word vectors under the label based on the calculated similarity; When the medical term entity word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the term interpretation pair corresponding to the medical term entity word vector; when the questionable question word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the question-answer pair corresponding to the questionable question vector; when the answer sentence word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the question-answer pair corresponding to the answer sentence word vector; when the fragment content word vector is determined as the initial candidate answer word vector, the candidate answer word vector is determined based on the fragment summary content pair corresponding to the fragment content word vector; Based on the candidate answer word vectors of various tags, the target answer word vector is determined, and the answer sentence corresponding to the medical question is obtained based on the target answer word vector.

6. The method according to claim 5, characterized in that The step of determining a target answer word vector based on the candidate answer word vectors of each of the various tags, and obtaining an answer sentence corresponding to the medical question based on the target answer word vector includes: The candidate answer word vectors corresponding to various labels are obtained through a preset similarity scoring model, and the similarity scores between them and the question sentence vectors are scored. The candidate answer word vectors are sorted according to the obtained similarity scores, and the target answer word vector is determined based on the sorting results; The target answer words corresponding to the target answer word vector are filled into a preset answer generation template to obtain an answer sentence corresponding to the medical question, wherein the preset answer generation template includes a plurality of connecting words rewritten in colloquial language, and the connecting words are used to connect the answer words.

7. The method according to claim 1, characterized in that The collection includes multiple original medical documents on health knowledge for the elderly, including: Collect multiple original candidate medical field documents containing health knowledge of the elderly; For any original candidate medical field document, the evaluation value of the original candidate medical field document is obtained according to the preset evaluation indicators and the indicator weights corresponding to the preset evaluation indicators, and the original medical field document corresponding to the evaluation value greater than the preset evaluation threshold is obtained, wherein the preset evaluation indicators include the document collection source, the document release time and the authority of the document source.

8. An intelligent question-answering processing device for the elderly, characterized in that: The device comprises: The medical document collection module is used to collect multiple original medical documents containing health knowledge for the elderly; A formal text conversion module is used to convert any original medical field document into multiple formal texts, wherein the formal texts include terminology and explanation formal texts, question-answer formal texts and hierarchical summary formal texts, and the text content in each formal text is colloquial language; A multi-source database construction module is used to construct a multi-source database based on multiple forms of text corresponding to multiple original medical documents; A medical question processing module, used for receiving medical questions raised by elderly users, and obtaining question sentence vectors based on the medical questions; The answer statement generation and return module is used to generate an answer statement corresponding to the medical question based on the multi-source database and the question sentence vector, rewrite the answer statement into colloquial language, and return the rewritten answer statement.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the method for intelligent question-answering processing for the elderly according to any one of claims 1 to 7.

10. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, it implements the method for intelligent question-answering processing for the elderly described in any one of claims 1 to 7.

Images