SubTLe. Subjectivity tagging and learning

Begin - Einde 
2010 - 2014 (afgewerkt)
Type 
Vakgroep(en) 
Vakgroep Vertalen, Tolken en Communicatie
Onderzoeksgebied 

Tabgroup

Abstract

The SubTLe project investigates two different methodologies for sentiment mining of online text types (blogs, chats, tweets, etc.). We contrast the traditional lexicon-based approach with a corpus-based approach based on a deeply annotated corpus of user generated content.

In cooperation with the Suicide Prevention Centre, the resulting systems are applied to detect suicidal messages in online content.

Onderzoekers

Promotor(en)

Doctoraatsstudent(en)

Publicaties