LyS at TASS 2014A Prototype for Extracting and Analysing Aspects from Spanish tweets
- David Vilares 1
- Yerai Doval 1
- Miguel A. Alonso 1
- Carlos Gómez-Rodríguez 1
-
1
Universidade da Coruña
info
Publisher: Sociedad Española para el Procesamiento del Lenguaje Natural
Year of publication: 2014
Congress: Sociedad Española para el Procesamiento del Lenguaje Natural. Congreso (30. 2014. Girona)
Type: Conference paper
Abstract
This paper describes our participation at the third edition of the work- shop on Sentiment Analysis focused on Spanish tweets, tass 2014. This year’s eval- uation campaign includes four challenges: (1) global sentiment analysis, (2) topic classification, (3) aspect-extraction and (4) aspect-based sentiment analysis. Tasks 1 and 2 are addressed from a machine learning approach, using several linguistic resources and other information extracted from the training corpus to feed to a su- pervised classifier. With respect to task 3, we develop a naive approach, collecting a set of representations to identify the predefined aspects requested by the organisers. Finally, task 4 uses heuristics to identify the scope of each aspect, to then classify their sentiment via a supervised classifier. The experimental results are promising and will serve us as the starting point to develop more complex techniques.