SemEval-2017 Task 5: Fine-Grained Sentiment Analysis on Financial Microblogs and News focuses on analyzing the sentiments and opinions on financial data that can affect market dynamics. There were two subtasks:
- Subtask 1: Calculate polarity score from microblog posts collected from Twitter and Stocktwits.
- Subtask 2: Calculate polarity score from financial news headlines collected from various financial news sources like Yahoo Finance.
The RiTUAL-UH team secured the second position with a weighted cosine similarity score of 0.7437 in Subtask2 and the eighth position with a weighted cosine similarity score of 0.7234 in Subtask1.
Organizers decided to use an alternate scoring system where RiTUAL-UH ranked the first in both subtasks.
The detailed results are here.