Design And Implementation Of Automatic Question Answering System Based On Ontology For Traditional Chinese Medicine Coronary Artery Disease

The emergence of automatic question answering system(AQAS)has greatly satisfied the demand of people for accessing to relevant information efficiently.However,most of the modern QA systems are based on Frequently Asked Questions(FAQ)in open fields.It is inevitable that there are some inadequacies in the field of traditional Chinese medicine(TCM)coronary artery disease(CAD),such as poor professional answers,low accuracy,poor semantic comprehensions,inflexible forms.How to improve and design the Natural Language Processing(NLP)schemes to adapt to the automatic question and answering systems for CAD is a key problem in TCM domain.Each ontology node of TCMCAD likes a human brain neuron,which can reduce the drawbacks,to some extent,by replacing the traditional methods based on FAQ.Therefore,this paper focuses on designing and implementing an effective NLP scheme,which can verify the feasibility and high effectiveness for AQAS based on Ontology in TCMCAD.This thesis is based on the TCM literature authenticated by the experts to construct the knowledge network of Ontology in TCMCAD,which lays the foundation of the system to generate correct and professional answers.The related data construction involves the TCMCAD keyword thesaurus,common words thesaurus,question words thesaurus,question problem templates and word vector table,which can realize information matching and extraction from about 2962 keywords of CAD in TCM domain and 700 pieces of ancient Chinese literature.And the study provides the basic data for the system and the design methods and algorithms,especially for the word segmentation method,Reverse Maximum Matching algorithm based on binary search.After researching the key technology of domain-restricted QA system in NLP,an implementation schemes of AQAS in TCMCAD is designed.In this thesis,the system is divided into three processing phases on the basis of the question sentence characteristics.For each phase,there are different processing methods,which are based on question template matching,pattern matching and similarity calculation based on word vector.Moreover,the overall design schemes are expounded,and the TCMCAD keywords fuzzy matching algorithm are designed,and the similarity algorithm based on word vector from neural network modeling are built.It has greatly enhanced the system flexibility with taking the the answering accuracy as the prerequisite.This thesis uses Delphi7 to realize the AQAS based on Ontology for CAD in TCM field.Moreover,the word vectors are generated with Java and the data is stored with the files of Excel and TXT,such as keywords thesaurus of CAD in TCM and common words thesaurus,question words thesaurus,questions templates and word vector table.Through testing the system,it further verifies the applicability and intelligent of the design schemes.Therefore,the system can serve the doctors and patients by a more acceptable way.Furthermore,it is of great significance in TCM ancient literature developments,and provides a paradigm for researching on AQAS in TCM domain.