We performed a thorough replicate study of the top systems performing in the yearly SemEval Twitter Sentiment Analysis task. We highlight some differences between the results obtained by the top systems and the ones we are able to compute. We also propose SentiME, an ensemble system composed of 5 state-of-the-art sentiment classifiers. SentiME first trains the different classifiers using the Bootstrap Aggregating Algorithm. The classification results are then aggregated using a linear function that averages the classification distributions of the different classifiers. SentiME has also been tested over the SemEval2015 test set, properly trained with the SemEval2015 train test, outperforming the best ranked system of the challenge.
A replication study of the top performing systems in SemEval Twitter sentiment analysis
ISWC 2016, 15th International Semantic Web Conference, Resources Track, October 17-21, 2016, Kobe, Japan
© Springer. Personal use of this material is permitted. The definitive version of this paper was published in ISWC 2016, 15th International Semantic Web Conference, Resources Track, October 17-21, 2016, Kobe, Japan and is available at : http://dx.doi.org/10.1007/978-3-319-46547-0_22
PERMALINK : https://www.eurecom.fr/publication/5009