The Research On The Consultation And Question Answering Technology Of College Entrance Examination Based On Knowledge Graph

The college entrance examination is an important way for students to enter university.It is very important for examinees to know about various universities and majors after taking the college entrance examination.With the rapid growth of network information,it is very difficult to find the information among the large amount of information.In addition,application will begin immediately after the results of the college entrance examination are released.Therefore,it is important to be able to understand this information accurately and quickly.At present,with the continuous improvement of intelligence,NLP technology is also constantly improved,and the Q&A system has therefore been developed.Nowadays,intelligent question-answering systems are everywhere.Since the Google Knowledge Graph was proposed in 2012,it has quickly become widely known.Its characteristic is,store the data in the graph structure,make the database more concise,and can reason.Especially in the specific domain,making domain knowledge map can clearly reflect the relationship between data.Therefore,it is possible to develop a knowledge map based Q&A system for consulting universities and majors.Information is stored in a graphical structure and returned in the form of card information to facilitate,efficient and intuitive retrieval of relevant information.Therefore,this paper focuses on the making of knowledge map in the field of college entrance examination consultation and the questions used by examinees and parents in the consultation.The main work includes:1.Build a database of college entrance examination consulting fields.Firstly,the crawler technology was used to collect data from major college entrance examination platforms and official websites of colleges and universities.The data includes the data of 25 undergraduate universities in Yunnan Province in recent 3 years.The data include admission minimum score,average score,lowest ranking,batch,and number of plans.Also include the enrolment of colleges and universities general rules,institutional Settings.So as to form the data set needed to construct the knowledge graph.2.To build knowledge map of college entrance examination consulting field.The acquired data are used to construct the pattern layer of the knowledge graph.Define the nodes and edges of the knowledge graph.The Python programming language is used to call the Py2 neo module to connect to the Neo4 j diagram database.Complete the construction of knowledge graph.3.Innovate the classification model of college entrance examination questions.Based on CNN-BiLSTM-Attention model,the C&W-CNN-BiLSTM-Attention model is proposed.Explore on the basis of the original model.The model based on word vector is extended to word vector and character vector two-channel model.By adding character vector branch on the basis of semantic feature extraction by word vector,the model can mine the key information of data in a deeper level.4.Optimization of question entity recognition in the field of college entrance examination consulting.It is the first time to use the Bert-base model in the field of college entrance examination consultation.It was fused with the BILSTM-CRF model to form the BERT-BILSTM-CRF model.The traditional Word2vec method is replaced by the Bert-base model.Solve the problem that traditional Word2vec word vector cannot represent a polysemy and cannot consider the context semantics.And it has a strong ability to transfer,so that you can skip the tedious word vector training link.Save time.