A Comparative Study Of Embedding-Based Textual Semantic Similarity techniques For Urdu Language

Abstract

Textual Semantic Similarity (TSS) evaluates the degree to which two sentences or short texts are semantically proportional to one another. It plays an increasingly important role in tasks such as machine translation, information retrieval, and textual forgery detection. TSS is one of the significant problems in the field of Natural Language Processing (NLP). Text reuse and plagiarism detection are famous examples of TSS. TSS can be found on several levels, for example, word, sentence, and document level. Existing approaches have relied upon word and sentence level embedding for various languages (English, Arabic, Hindi, Turkish, etc.) to retrieve similarity indexes. Our research focuses on studying the existing approaches and comparing these for Urdu TSS tasks by using Word, Sentence, and Document-level embedding respectively.