SemEval-2017 Task 5: Fine-Grained Sentiment Analysis on Financial Microblogs and News

Abstract

This paper discusses the “Fine-Grained \nSentiment Analysis on Financial Microblogs \nand News” task as part of \nSemEval-2017, specifically under the \n“Detecting sentiment, humour, and truth” \ntheme. This task contains two tracks, where \nthe first one concerns Microblog messages \nand the second one covers News Statements \nand Headlines. The main goal behind both \ntracks was to predict the sentiment score for \neach of the mentioned companies/stocks. \nThe sentiment scores for each text instance \nadopted floating point values in the range \nof -1 (very negative/bearish) to 1 (very \npositive/bullish), with 0 designating neutral \nsentiment. This task attracted a total of 32 \nparticipants, with 25 participating in Track \n1 and 29 in Track 2.