Research On The Evaluation Of Academic Papers Based On Open Access Text

With the important role of peer review in the evaluation of high quality achievements"proposed by Ministry of science and technology of Ministry of education,how to give full play to peer review in academic paper evaluation has become an important research direction in the field of scientometrics.At the same time,the continuous development of open access platform brings more and more open access texts,forming a multi-source open academic evaluation model,which provides the conditions for the quantitative research of qualitative evaluation of academic papersIn order to study the validity of peer review text as the research data of academic paper quality evaluation,this paper collects the data of the review text of ICLR conference from the open review website,and proposes an sentiment analysis model of the review text based on the attention mechanism of the opinion sentence.According to the positive and negative sentiment prediction of the candidates' accepting states,this paper conducts experiments on three traditional sentiment analysis models,which verifies the effectiveness of peer review text and model.Then,the paper acceptance index obtained by this model is used to complete the task of prediction of the decision-making status of the paper,and the relationship between the acceptance index and the average score and citations is analyzed.Base on this,Twitter review are added to obtain papers and propose a paper quality evaluation model integrating qualitative evaluation.The paper information,Twitter review and peer review are used as model input,and the average score of paper acceptance is taken as the actual score.The sentiment analysis model is used to mining and open to get the sentiment polarity of text,and the paper prediction score is obtained.The effectiveness of the quantitative model is analyzed by comparing the predicted score with the actual score,predicted ranking and the actual ranking.The relationship between the predicted score and the traditional literature measurement index is analyzed,and a new dataset is added to explore the other use possibilities of the model.This paper obtains quantitative index of qualitative evaluation by mining the sentiment of qualitative evaluation,which provides a new index for the quality evaluation system of academic papers.