Closed nasifimtiazohi closed 6 years ago
It was initially not chosen because the tool does a binary classification (negative vs non-negative)
However, contacting the authors, they now provided a larger Oracle. If we train the tool on this, we will get a ternary classification. Currently, the training is ongoing. It is taking some time.
generated SentiCR results, it performs similarly bad. However it performs poorer in the first agreed dataset and improves in the final one unlike the other tools. Senti4SD and Sentistrength are still the two best tools in the paper.
writing with results https://github.com/DeveloperLiberationFront/AffectAnalysisToolEvaluation/commit/3c8f9401c2ac5caf7355cf5e30f39754cb4aa917
R2