Research And Implementation Of Abstact-Based Paper Classification And Recommendation Model

Researchers often need to consult a large number of scientific papers in their daily research activities.Therefore,the rapid acquisition of scientific research papers related to research directions has become a common demand of researchers.Thus,the classification and recommendation of paper has become an urgent problem to be solved.The abstract of paper is a kind of short text data with great research value.Therefore,it is of great significance to perform short text processing on data and to implement abstract-based paper classification and recommendation.At this stage,short text processing problems are mainly solved by deep learning techniques,such as convolutional neural networks and recurrent neural networks.However,the drawback of these methods is that the mining of short text feature spaces is not sufficient.Therefore,this paper studies and improves the existing short text processing method,and implements an abstract-based paper classification and recommendation model as well as a prototype system for scientific paper recommendation.The main contents are as follows:(1)Research of abstract-based paper classification model.In view of the shortcomings of existing short text classification methods,this paper adds topic vector concatenation method to convolutional neural network,completes the introduction of text topic information,and proposes a topic-based convolutional neural network model.Moreover,this paper integrates the model into the paper classification model,which solves the problem of insufficient mining of text feature space caused by existing methods,and achieves better classification accuracy.(2)Research of abstract-based paper recommendation model.In this paper,a new feature engineering method is proposed,which adds classification results obtained in the paper classification model to the paper recommendation model in the form of classification features.This method solves the problem that the traditional feature engineering method based on statistics and semantics has limitations in the breadth of feature extraction.Thus,an abstract-based paper recommendation model with better classification accuracy is proposed.(3)Design and implementation of prototype system for scientific paper recommendation.Based on the models above,this paper implements a prototype system for scientific paper recommendation with high-quality paper recommendation function,completing the application verification of the model proposed in this paper.In summary,this paper analyzes the shortcomings of existing short text classification and recommendation models for solving abstract-based paper classification and recommendation problems,proposes an-based paper classification and recommendation models as well as designs and implements a prototype system for scientific paper recommendation systems.Through the model verification experiment and system test,the model proposed in this paper is superior to the existing short text classification and recommendation model in terms of accuracy and other indicators.The system implemented in this paper can meet the functional and non-functional requirements of users.