Keywords
How to Cite
Abstract
This thesis proposes a methodology for evaluating semantic search (embedding) models for Uzbek-language texts. In addition to measuring search accuracy, the methodology examines factors specific to the Uzbek language, including script type, apostrophe characters, word forms, and model robustness to paraphrasing.
References
1. Lewis P., Perez E., Piktus A. va boshq. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks // Advances in Neural Information Processing Systems (NeurIPS). – 2020. – Vol. 33. – P. 9459–9474.
2. Feng F., Yang Y., Cer D., Arivazhagan N., Wang W. Language-agnostic BERT Sentence Embedding // Proceedings of ACL. – 2022. – P. 878–891.
3. Wang L., Yang N., Huang X. va boshq. Multilingual E5 Text Embeddings: A Technical Report // arXiv preprint arXiv:2402.05672. – 2024.
4. Chen J., Xiao S., Zhang P. va boshq. BGE M3-Embedding: Multi-Lingual, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation // arXiv preprint arXiv:2402.03216. – 2024.
5. Bandarkar L., Liang D., Muller B. va boshq. The Belebele Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants // Proceedings of ACL. – 2024. – P. 749–775.
6. Thakur N., Reimers N., Rücklé A., Srivastava A., Gurevych I. BEIR: A Heterogeneous Benchmark for Zero-shot Evaluation of Information Retrieval Models // NeurIPS Datasets and Benchmarks Track. – 2021.
7. Robertson S., Zaragoza H. The Probabilistic Relevance Framework: BM25 and Beyond // Foundations and Trends in Information Retrieval. – 2009. – Vol. 3, No. 4. – P. 333–389.