MITRE at SemEval-2016 Task 6: Transfer Learning for Stance Detection
This work addresses stance detection for social media analysis, but it is incremental as it applies existing transfer learning methods to a specific dataset.
The paper tackled stance detection in tweets by developing a system that achieved the top score in a supervised task with an average F1 score of 67.8, using transfer learning to enhance performance with limited labeled data.
We describe MITRE's submission to the SemEval-2016 Task 6, Detecting Stance in Tweets. This effort achieved the top score in Task A on supervised stance detection, producing an average F1 score of 67.8 when assessing whether a tweet author was in favor or against a topic. We employed a recurrent neural network initialized with features learned via distant supervision on two large unlabeled datasets. We trained embeddings of words and phrases with the word2vec skip-gram method, then used those features to learn sentence representations via a hashtag prediction auxiliary task. These sentence vectors were then fine-tuned for stance detection on several hundred labeled examples. The result was a high performing system that used transfer learning to maximize the value of the available training data.