Design And Implementation Of A Search System Based On Multi-source Heterogeneous Academic Data

With the rapid development and progress of internet information,search engines have become an indispensable part of people’s lives.Among them,academic search engines are online tools that specialize in retrieving academic literature.They can help scholars,researchers,teachers and students quickly find relevant information,understand the latest developments in a certain field,and evaluate the quality and impact of literature.However,current academic search engines also process query words relatively simply and cannot support unified ranking of multi-source heterogeneous data.Therefore,this paper designs an academic search system that aims to improve the search experience and content quality of academic search engines by implementing deep processing of query words and multimodal content re-ranking,thereby improving the efficiency and quality of academic research and promoting knowledge innovation and dissemination.Therefore,the main research work of this paper is as follows:(1)Design and implement a query word understanding module based on natural language technology.By designing six sub-modules:query word preprocessing,error correction,intent recognition,segmentation,expansion to complete the processing of query words.Deeply understand the user’s input query words to more accurately recall and rank relevant documents.(2)Design and implement a multi-way recall engine based on ElasticSearch.Use ElasticSearch to build multi-way indexes for multisource heterogeneous academic data.Through ElasticSearch’s powerful search function,quickly and accurately recall documents.(3)Propose a ranking algorithm for multi-source heterogeneous data based on deep learning technology.Use gated recurrent neural network and convolutional neural network to learn text and image information respectively.Finally use inter-modal attention mechanism to fuse models to complete re-ranking task.And verify model effectiveness through a large number of comparative experiments and ablation experiments.(4)Build a microservice system for searching backend based on Spring Cloud.This paper follows software engineering design ideas and processes.Based on clarifying system functionality and performance requirements determine system architecture complete analysis design of each functional module interface meet user classification retrieval unified retrieval query word association requirements.Finally conduct sufficient testing on system prove correctness effectiveness design scheme proposed in this paper.