Design And Implementation Of Teaching Queston Answering System Based On Lucene

Teaching Q&A is a very important part of teachers' teaching work.The traditional way of answering questions is usually to ask the teacher to answer the question in person,or to ask the teacher for questions through email,QQ,WeChat and other channels.In the traditional question-answering process,teachers will inevitably encounter a large number of repeated and similar problems,which is a waste of teachers' instruction.At the same time,the teacher's question answering information is not well recorded,which leads to a low knowledge sharing rate.Based on the above background,this paper designs and implements a teaching question answering system.In the process of question and answer,the system needs to match the most similar question from the question and answer database according to the user's question,and then return the corresponding answer to the user.First,the system will use full-text retrieval technology to quickly recall a batch of candidate question sets from the question and answer database,thereby reducing the computational pressure of subsequent semantic matching.Then use semantic matching technology to perform semantic matching on the candidate question set to find the questions with the highest similarity and greater than the threshold.In the above process,it mainly includes the key technologies of question retrieval and semantic matching.For question retrieval,the system uses the open source full-text search toolkit Lucene to achieve question retrieval.By analyzing the shortcomings of its default scoring mechanism in question retrieval,this paper proposes a re-score that combines synonyms information and keyword information mechanism.And design experimental comparison,when using the improved scoring mechanism,the recall rate and accuracy of search results have improved.For semantic matching,this paper uses deep learning technology to construct a sentence similarity matching model,investigates the CNN-based sentence semantic similarity matching model and the LSTM-based sentence semantic similarity matching model,and proposes a sentence semantic similarity matching model based on CNN and LSTM and includes attention mechanism,And the design experiment compares the effects of each model.The model proposed in this paper has improved the accuracy and F1-Score of the semantic matching task.This paper first expounds the background and significance of the subject and the current situation of research at home and abroad,and determines the main research content of the thesis.Then this paper introduces the relevant techniques and theories of the teaching question answering system.Then this paper studies the key technologies of Lucene-based question retrieval and deep-learning-based sentence semantic similarity matching model in the question answering system.Then,according to the practical application scenario of the teaching question answering system,this paper makes a demand analysis of the system and determines the functional requirements and non-functional requirements of the system.Then the system is designed in outline,the system's network topology and software-level architecture are determined,and the division of the system functional modules and the database design are completed.Then according to the requirements analysis and outline design,the functional modules of the system are designed in detail,and the implementation details of each module are introduced in detail.Finally,the system was tested for function and performance according to the requirements analysis,and the temporary test results were analyzed.Based on the actual needs,this paper designs and implements an easy-to-use and fully functional teaching question answering system by using Lucene,deep learning and Web front-end back-end development.