SemEval2016-Twitter_Sentiment_Evaluation

The code used for my SemEval 2016 submission.

Contents

The repository describes my participation as team "TwiSE" in the SemEval 2016 challenge. Specifically, we participated in Task 4, namely "Sentiment Analysis in Twitter" for which I implemented sentiment classification systems for subtasks A, B, C and D. The approach is described at this paper

Performance

The approach yielded satisfactory results in each of the subtasks, and especially in subtasks C and D. In the official leaderboard the results were ranked

• 9/35 for Subtask A: Message Polarity Classification (Positive-Neutral-Negative)
• 8/19 for Subtask B: Tweet classification according to a two-point scale
• 1/11 for Subtask C: Tweet classification according to a five-point scale, and
• 2/14 for Subtask D: Tweet quantification according to a two-point scale

References

If you use the code, please cite the following paper:

@article{balikas2016semeval,
  author    = {Georgios Balikas and Massih-Reza Amini},
  title     = {TwiSE at SemEval-2016 Task 4: Twitter Sentiment Classification},
  journal   = {CoRR},
  volume    = {abs/1606.04351},
  year      = {2016},
  url       = {http://arxiv.org/abs/1606.04351},
}