Hybrid system for recognizing Acronym expansions using heuristics And machine learning technique

Abstract

An acronym is a type of abbreviation made up of initial letter or letters of other words. An abbreviation is a short form (SF) of a phrase. The long-form (LF) of an abbreviation is called either a definition or an expansion. Abbreviations and acronyms are commonly used in biomedical literature, scientific and technical articles, information retrieval and web search, etc. Recognizing full forms that are associated with the acronym is important for identifying the meaning of an acronym that facilitates natural language processing and information retrieval from the literature. newlineSeveral research works are under practice to automate the recognition of acronym expansion pairs from text and web documents. Heuristics or Machine Learning approaches are prevalently pursued extracting acronym-definition from text or web. Existing heuristics and machine learning approaches recall rate (i.e. Number of retrieved acronym expansion pairs from document rate) is low. Hence, a hybrid model combining heuristics and machine learning is proposed in this work to retrieve more number of acronym expansion pairs from documents. The main objective of the work is to extract abbreviation definition pairs from text documents and also find the list of definitions of the acronym from the web. newlineFirstly, seven space reduction heuristics are applied to recognize acronyms from the text. Then, three mapping strategies are proposed for doing a sequence labeling task to recognize the expansion of the acronym. Since the usage of acronyms is more in biomedical literature, a biomedical dataset is created from Thalia semantic search engine. Then, the dataset is utilized to identify the potential abbreviation definition pairs. newline