Research And Implementation Of Intelligent Question Answering Algorithm Based On Geological Literature

In recent years,with the continuous development of national science and technology innovation and exploration in geological risk prevention and application demonstration,the Ministry of land and resources and China Geological Survey put forward innovative three-step plan on the basis of the national 13 th five year plan,centering on the national science and technology strategy to promote rapid economic development and scientific and technological service innovation in the geological field,and using scientific and technological forces to solve the problem of environmental issues and the safety of the earth system.At present,in the geological mining discovery,PB level electronic documents,texts and other unstructured discrete data are very unfavorable for geological retrieval and statistical discovery.At the same time,with the improvement of the national economic level,the public’s desire for high efficiency of service is increasing.Therefore,in order to meet the strategic deployment of national science and technology innovation and the people’s desire for quality of life,China Geological Survey Bureau is aware of geological problems Knowledge discovery proposes to build a knowledge discovery assistant decision-making platform based on geological documents.In the process of knowledge discovery,this thesis takes geological knowledge intelligent QA service as the starting point to study the decision support and knowledge service of geological knowledge association.The research work includes the following three aspects:(1)Named entity recognition from query statement information.According to the geological knowledge asked by users in the QA system,the grid based LSTM model is used to identify named entities in the geological field,including geological information from geochemistry to biological insects,so as to realize the node information mapping of the knowledge map.(2)In addition to geological entity,the user’s intention is judged for the information of inquiry statement.The character-based convolutional neural network(CNN)is used to classify the attributes of the information except the named entity recognition part of the user’s inquiry information,including 14 types from definition,attribute label,relation to knowledge inference,so as to realize the user’s intention.(3)QA platform architecture design and application implementation.By using Python’s Web lightweight development framework to process and store literature data,a service platform for decision support of geological QA is realized.Aiming at the time-consuming,laborious and weak expansibility of user intention recognition in question answering robots based on template matching,keyword cooccurrence or artificial feature set.The hybrid model of this thesis regards user’s intention recognition as a classification problem.Firstly,the grid memory network is used to identify the named entity,then the convolution neural network is used to classify the attributes of other text information input by users,and then the classification results are transformed into a structured way to meet the query of knowledge graph,and finally realizes the attribute mapping of user intention recognition.Experiments show that the hybrid model focuses on the characteristics of the research object,and effectively improves the accuracy of intention recognition of question answering system.