Acceptance Decision Prediction in Peer-Review Through Sentiment Analysis

• Type: Article in International Conference Proceedings Book
• Date: 2021

Authors

Ana Carolina Ribeiro (Author) Other; Amanda Sizo (Author) Other; Henrique Lopes Cardoso (Author) FEUP; Luís Paulo Reis (Author) FEUP

Conference proceedings International

Title: PROGRESS IN ARTIFICIAL INTELLIGENCE (EPIA 2021)

Pages: 766-777

20th EPIA Conference on Artificial Intelligence (EPIA)

ELECTR NETWORK, SEP 07-09, 2021

Indexing

ISI Web of Knowledge - 1 Citation

Scopus - 1 Citation

Other information

Authenticus ID: P-00V-CPG

DOI: 10.1007/978-3-030-86230-5_60

Abstract (EN): Peer-reviewing is considered the main mechanism for quality control of scientific publications. The editors of journals and conferences assign submitted papers to reviewers, who review them. Therefore, inconsistencies between reviewer recommendations and reviewer comments are a problem that the editor needs to handle. However, few studies have explored whether it is possible to predict the reviewer recommendation from review comments based on NLP techniques. This study aims to predict reviewer recommendation of the scientific papers they review (accept or reject) and predict reviewers' final scores. We used a dataset composed of 2,313 review texts from two computer science conferences to test our approach, based on seven ML algorithms on regression and classification tasks and VADER application. SVM and MLP Classifier achieved the best performance in the classification task. In the regression task, the best performance was achieved by Nearest Neighbors. One of the most interesting results is the positive classification of most reviews by VADER: reviewers present constructively written reviews without highly negative comments land; therefore, VADER cannot detect reviews with a negative score.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 12