Corpus-Based Paraphrase Detection Experiments and Review

TL;DR

This paper reviews and compares the performance of various corpus-based models, especially deep learning approaches, for paraphrase detection across multiple datasets, highlighting their competitiveness and potential for further development.

Contribution

It provides a comprehensive performance overview of eight models on three datasets, identifying effective preprocessing and model configurations for paraphrase detection.

Findings

Deep learning models are highly competitive with traditional methods.

Optimal preprocessing and hyper-parameter choices improve model performance.

DL models show significant potential for future research in paraphrase detection.

Abstract

Paraphrase detection is important for a number of applications, including plagiarism detection, authorship attribution, question answering, text summarization, text mining in general, etc. In this paper, we give a performance overview of various types of corpus-based models, especially deep learning (DL) models, with the task of paraphrase detection. We report the results of eight models (LSI, TF-IDF, Word2Vec, Doc2Vec, GloVe, FastText, ELMO, and USE) evaluated on three different public available corpora: Microsoft Research Paraphrase Corpus, Clough and Stevenson and Webis Crowd Paraphrase Corpus 2011. Through a great number of experiments, we decided on the most appropriate approaches for text pre-processing: hyper-parameters, sub-model selection-where they exist (e.g., Skipgram vs. CBOW), distance measures, and semantic similarity/paraphrase detection threshold. Our findings and those of other researchers who have used deep learning models show that DL models are very competitive with traditional state-of-the-art approaches and have potential that should be further developed.