There are multiple systems for sentiment analysis with heterogeneous designs and varying levels of performance. In this paper, we propose a model of computational specifications of polarity identification systems. The model makes easier the comparison of the different techniques used. We seek to create awareness among researchers of the necessary information for the elaboration of specifications that allow the replication of systems. Additionally, we discuss the difficulties of evaluating and conducting consistent comparisons among systems, and going beyond the result that can be obtained for a specific task on a particular data set. We are convinced that a complete and clear framework that encompasses all the modules of the systems that participate in competitions is crucial to enrich the knowledge for improving the state-of-the-art in sentiment analysis.
Comparative analysis of the computational characteristics in modern sentiment analysis systems for Spanish
0 pre-pub reviews
0 post-pub reviews