A semantic analyzer for detecting attitudes on SNs
Horia Nicolai Teodorescu, Navanath Saharia · 2016
Vie present the basic principles of an automated system for attitude detection in short messages and other texts. The core of the system is a 'semantic parser' that has three main components: the detector of attitude-pointing words, the syntax parsing for determining to whom the attitude-loaded words relate, and the determination of the relationship between the message writer, the message addressee, and the other persons on which the writer narrates. The social network messages include complexities due to incompleteness, poor grammar structure, use of slang, imprecise meaning, use of spatio-temporal references, use of contextual information, and use of various types of anaphors. We detail the ways of overcoming the difficulties.