Paper Details
Authors: Anik Mahmud Shanto, Md. Sajid Alam Chowdhury, Mostak Mahmud Chowdhury, Udoy Das, Hasan Murad.
Venue: Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024), pages 859-864, Mexico City, Mexico, 2024.
Status: Published · DOI: 10.18653/v1/2024.semeval-1.123.
Abstract
Predicting semantic textual relatedness is one of the most challenging tasks in natural language processing. Semantic relatedness prediction has practical applications in search engines and modern text generation systems. SemEval 2024 organized a shared task on semantic textual relatedness, providing an English dataset under Shared Task 1, Track A3. We developed models to predict relatedness between pairs of English sentences by training and evaluating transformer-based architectures, deep learning models, and machine learning methods on the shared dataset. We also used existing semantic textual relatedness datasets such as the STS-B multilingual benchmark, SemEval 2014 Task 1, and SemEval 2015 Task 2. The fine-tuned STS-BERT model performed best, scoring 0.8103 on the test set and placing 25th among participants.
BibTeX
@inproceedings{shanto-etal-2024-fired,
title = {{F}ired{\_}from{\_}{NLP} at {S}em{E}val-2024 Task 1: Towards Developing Semantic Textual Relatedness Predictor - A Transformer-based Approach},
author = {Shanto, Anik Mahmud and Chowdhury, Md. Sajid Alam and Chowdhury, Mostak Mahmud and Das, Udoy and Murad, Hasan},
booktitle = {Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024)},
pages = {859--864},
year = {2024},
publisher = {Association for Computational Linguistics},
doi = {10.18653/v1/2024.semeval-1.123},
url = {https://aclanthology.org/2024.semeval-1.123/}
}