Construction Of Intelligent Q&A System For Medicinal Plant Based On Multimodal Knowledge Graph

Medicinal plants refer to plants that are used in traditional medicine,which have the functions of treating diseases,relieving symptoms and maintaining health.China is one of the countries with the richest medicinal plant resources.The history of Chinese people cultivating,picking and using medicinal plants is long and reflects the invaluable historical value of medicinal plants.In addition,medicinal plants are one of the important sources of raw materials for modern pharmaceutical industry.The development and utilization of medicinal plants have broad economic value.Moreover,medicinal plants are an important part of biodiversity and have important significance for the balance and maintenance of ecosystems.This paper selects 265 kinds of medicinal plants from the Chinese Pharmacopoeia,integrates multi-source heterogeneous data of medicinal plants,constructs a multi-modal knowledge graph of medicinal plants,realizes the question answering function by using various convolutional neural networks,and finally completes the research and implementation of the intelligent question answering system of medicinal plants.The main research contents of this paper are as follows:(1)Construction of multi-modal knowledge graph of medicinal plants.Structured text data,semi-structured text data and image data are obtained from different professional databases in the field of traditional Chinese medicine and processed.After knowledge extraction and knowledge fusion,entities,relations and attributes of medicinal plant knowledge graph are obtained.Finally,the association between medicinal plant entities and images is established by Latin name,and image entities and corresponding relations are added to the medicinal plant knowledge graph.The above entities,relations and attributes are stored in Neo4 j graph database by knowledge storage,thus completing the construction of multi-modal knowledge graph of medicinal plants.(2)Implementation of intelligent question answering function.Question answering function is the core function of intelligent question answering system for medicinal plants.According to different modalities of user question data,question answering function is divided into text question answering function and image question answering function.The key steps to realize text question answering function are question entity recognition and question intention recognition.This paper implements the above steps based on Aho-Corasick automaton and TextCNN model.For image answer function,this paper implements medicinal plant image recognition function based on Efficient Net-B3 model,and improves the recognition accuracy to 83.53%.(3)Implementation of intelligent question answering system for medicinal plants.This paper analyzes the characteristics and system development requirements of multi-source heterogeneous data of medicinal plants,designs the system architecture,and realizes the question answering process by using python language PyQt library.Finally,the running process of each page of the system is introduced,and the effectiveness of the system is verified by testing.