Skip to content

Latest commit

 

History

History
5 lines (3 loc) · 454 Bytes

README.md

File metadata and controls

5 lines (3 loc) · 454 Bytes

Tweet-Semantics

Three distinct sentiment analysis classifier are used that are capable of labellign tweets as either positive, neutral or negative. The tweet dataset and general project was heavily inspired by the semeval competition.

The classifiers were tested against evaluated according to the macro-averaged F1-score, which meant that the inbalance in the tweets labels (negatives were nearly half of positives) was quite an important obstacle.