Summary of Umbclu at Semeval-2024 Task 1a and 1c: Semantic Textual Relatedness with and Without Machine Translation, by Shubhashis Roy Dipta and Sai Vallurupalli
UMBCLU at SemEval-2024 Task 1A and 1C: Semantic Textual Relatedness with and without machine translation
by Shubhashis Roy Dipta, Sai Vallurupalli
First submitted to arxiv on: 20 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The SemEval-2024 Task 1, “Semantic Textual Relatedness for African and Asian Languages,” aims to develop models that identify semantic textual relatedness between two sentences in multiple languages (14 African and Asian languages) and settings (supervised, unsupervised, and cross-lingual). Large language models have shown impressive performance on natural language understanding tasks like multilingual machine translation, semantic similarity, and encoding sentence embeddings. Using a combination of these LLMs, the authors developed two STR models, TranSem and FineSem, for supervised and cross-lingual settings. The effectiveness of various training methods and machine translation was explored. Results showed that direct fine-tuning on the task is comparable to using sentence embeddings, while translating to English leads to better performance for some languages. In the supervised setting, model performance was better than the official baseline for 3 languages, with the remaining 4 performing similarly. In the cross-lingual setting, model performance was better than the baseline for 3 languages, on par for 2, and poor for 7. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The goal of this research is to create models that understand how two sentences are related in different languages. The researchers used large language models that can translate texts and understand their meaning. They developed two new models, TranSem and FineSem, which work well in different situations. The team tried different methods to train the models and found that using sentence embeddings and translating to English helps for some languages. In most cases, the model performed as well or better than expected. |
Keywords
* Artificial intelligence * Fine tuning * Language understanding * Supervised * Translation * Unsupervised
