Work

Évaluation de la pertinence des citations dans les articles scientifiques

Year: 2023
Type: dissertation
Abstract: Citations play an important role in scientific research. However, recent studies have unveiled a concerning issue : numerous inaccurate citations are present within scientific papers. These erroneous references can result in misinterpretations, distortion of the original author’s intentions, and potentially even more serious consequences. To address this concern, our paper introduces two methods designed to assess citation accuracy using Natural Language Processing (NLP) techniques. The first method involves measuring text similarities through vectorized text representations. The second method entails employing a fine-tuned text classification approach to distinguish between reliable and flawed citations. Both of these methodologies adopt language models. Based on our experimental findings, it appears that the fine-tuned text classification method demonstrated the most optimal performance.
Source: HAL
Author: Liu, Qinyue
Language: English
Open Access status: green
OpenAlex ID: W7009620867
Cites: 19
Cited by: 1
Related to: 10