Polarity Detection of Named Entities and their

Abstract. This paper presents an opinion analysis application for the Spanish language based on proximity between polar words, entities and their coreferences. This is a named entity centered approach in compar-ison with a document or sentence level analysis. Examining the context of coreferences allows us to combine evidence to determine the polarity of entities with a higher confidence level. Making use of specific polarity lexicons for each class of entity allows us to improve precision in polarity detection.