A method for automatic detection of acronyms in texts and building a dataset for acronym disambiguation

Nowadays, there is an increasing tendency for using acronyms in technical texts, which has led to ambiguous acronyms with different possible expansions. Diversity of expansions of a single acronym makes recognizing its expansion a challenging task. Replacing acronyms with incorrect expansions will lead to problems in text mining procedures, namely text normalization, summarization, machine translation, and tech-mining. Tech-mining involves exploring and analyzing technical texts to recognize the relations between technologies. This paper is aimed at proposing a method for building a dataset that meets the requirements for training acronym disambiguation models in technical texts. In this paper, challenges in automatic acronym disambiguation are presented. We have proposed a method for building the dataset and the accuracy of the acronym disambiguation model is 86%.