Design And Implementation Of Intelligent Question Answering System For Academic Career Based On Semantic Analysis

The traditional retrieval question answering system can quickly retrieve the corresponding content through literal similarity matching,but the accuracy is not high.We hope that the Question answering system can maintain good operation,requiring its stable performance and strong concurrency.The question answering system based on deep learning can effectively solve the problem of accuracy.However,the training stage usually depends on a large number of data resources,and more calculation time is needed in prediction.This paper proposes a semantic analysis question answering system with text matching as the core.The system is a question answering system based on deep vector matching and deep learning sorting based on information retrieval with context understanding ability.The system adopts two-stage sorting algorithm and uses a simple retrieval model to retrieve multiple related problems in the first stage,In order to reduce the data that the training depends on and the time spent in recall,the second stage uses the sorting algorithm to use the pre training model for fine sorting again on the basis of the retrieved multiple questions,so as to obtain the best answer.The model also needs less data resources and is better than most models in accuracy.It is mainly composed of retrieval module and sorting module.Firstly,the user’s questions are processed by word segmentation,entity recognition and long and difficult sentence compression.This module mainly uses AC automata to extract keywords.Secondly,according to the preprocessed questions,the retrieval module queries the positions of multiple similar answers in the knowledge base through literal retrieval,semantic retrieval and knowledge map,and quickly selects the sentences with the highest scores.This module mainly uses BM25 statistical model based on word bag and word2 vec language model based on depth vector matching.Finally,the sorting module merges according to the retrieved results,filters some unreasonable answers through the entity alignment strategy,and uses the rearrangement technology to reorder the filtered answers to get the most appropriate answers.This module mainly uses the entity alignment method based on keywords and the learningtorank rearrangement model based on pairwise.The optimization and supplement of knowledge base and knowledge map are obtained from information events,dialogue logs,terms knowledge and other texts through knowledge collection and screening,orderly organization of knowledge,knowledge mining analysis,reading comprehension and other methods.This topic realizes the intelligent academic career evaluation question and answer system based on text matching,which solves the literal matching,depth vector matching and knowledge map text matching system with logical reasoning respectively.Through the further integration and combination of the three,the answer is more accurate and comprehensive,and solves the practical application of natural language question and answer processing technology in the field of academic career evaluation.