A context and ad hoc lexicon proposal to an interdisciplinary approach to opinion analysis

This paper addresses the question of how to determine polarity of opinion among texts and, more specifically, those published on Twitter. The analysis is performed from multiple perspectives, including information science, and communication and information ethics with emphasis on the analysis of natural language processing and knowledge organization. A prototype is developed that proposes annotation of opinion conveying words as adjectives and adverbs contextualized by grammatical core, i.e., nouns and verbs, respectively. These nuclei also serve to model the terminology of the domain, and the development an ad hoc lexicon is proposed.