Automatic Construction Method Of Historical Knowledge Base Based On Timeline

With the rapid development of artificial intelligence,machine learning and Natural Language Processing technology have made great progress,and the research of question answering system has attracted the attention of sch olars from all walks of life.People began to wonder whether the question answering system could be used as the answer to the college entrance examination questions like people,and the experiments were carried out on different subjects.Most of the existi ng question answering systems rely on a special knowledge base,knowledge base as a knowledge management tool for massive data management and organization,have very important practical significance for the confrontation problem solving "lack of knowledge" and "mass information".A well organized and knowledge based knowledge base is of great significance to the corresponding question answering system.However,the existing knowledge base is mostly aimed at the open domain,and the knowledge base for the specific task often needs to be constructed in a targeted way.This subject is mainly oriented to the history questions of Chinese college entrance examination,using machine learning,Natural Language Processing and other technologies to build a knowledge base of the history field.According to the characteristics of historical knowledge and the characteristics of the history short-answer question,the time information is relatively practical in the historical field,so the historical knowledge base is const ructed based on the time line.In the construction of the historical knowledge base,Baidu encyclopedia,Wikipedia were collected as a source of knowledge in the knowledge base.In the process of web page analysis on Wikipedia and history entries,proposes to use the page text density and text scope analysis to acquire the part of history knowledge in the content of the page,on account of the consumption problem of time for the main analysis for different web page tags.The collected Wikipedia contains all categories Chinese entires,we need to get the entries related to history from all the Wikipedia.Proposed to use the classification model based on convolutional neural network in the classification of Wikipedia,in order to enhance the quality of the hist orical knowledge base.After analyzing the time information in history knowledge,divided the time information into explicit time information and implicit time information.Using named entity recognition technology to solve the time information extraction task,a model combine conditional random field and deep learning is used to the time entites recognition task.After getting the time information in history knowledge,the knowledge base is organized according to the acquired time information,and the whole knowledge base is organized into a form based on time line according to the order of time.At the same time,in order to facilitate the knowledge view in the knowledge base,the display and retrieval system of the knowledge base is constructed.The knowledge base contains about 125 thousands entries,and the knowledge base is applied to the problem-solving process of the history entrance examination question answering system.