TRANSFROMATIONAL ANALYSIS OF ARABIC SENTENCES: APPLICATION TO AUTOMATICALLY EXTRACTED BIOMEDICAL SYMPTOMS INES BOUJELBEN ABDELMAJID BEN HAMADOU Abstract The transformational analysis has been a topic of many discussions concerning the semantic analysis. It is considered as one of the biggest challenges among natural language processing NLP tasks. This paper deals with the transformation analysis of Arabic automatically extracted sentences. Its main goal is to go beyond syntactic parsing in order to fully implement a transformational system that can both analyze texts and generate paraphrases. Given these purposes, we will propose a detailed generic approach that can be applied to different application domains. In fact, it can be used to other tasks of NLP such as the automatic summarization, the question / answering applications, and machine translation as well as text categorization process. Experimental results show that the proposed transformational process presents encouraging results. Introduction The transformational analysis requires additional meaning comprehension of texts. In fact, many sentences using different structures can explicit the same idea. These sentences can be enriched with adverbs, redundant expressions, adjectives and useless words. Through our task, we intend to reduce the semantic ambiguity and go beyond syntactic parsing. That is why we should be able to easily summarize what the sentence is about and decipher eventual hidden meanings in order to produce the most accurate paraphrases. We intend to simplify complicated and long sentences.