j-hartmann
commited on
Commit
·
312b829
1
Parent(s):
a51ab62
Update README.md
Browse files
README.md
CHANGED
@@ -46,7 +46,7 @@ Thanks to Samuel Domdey and chrsiebert for their support in making this model av
|
|
46 |
|
47 |
## Appendix 📚
|
48 |
|
49 |
-
Please find an overview of the datasets used for training below. All datasets contain English text. The table summarizes which emotions are available in each of the datasets.
|
50 |
|
51 |
|Name|anger|disgust|fear|joy|neutral|sadness|surprise|
|
52 |
|---|---|---|---|---|---|---|---|
|
@@ -57,4 +57,6 @@ Please find an overview of the datasets used for training below. All datasets co
|
|
57 |
|MELD, Poria et al. (2019)|Yes|Yes|Yes|Yes|Yes|Yes|Yes|
|
58 |
|SemEval-2018, EI-reg (Mohammad et al. 2018) |Yes|-|Yes|Yes|-|Yes|-|
|
59 |
|
60 |
-
The datasets represent a diverse collection of text types. Specifically, they contain emotion labels for texts from Twitter, Reddit, student self-reports, and utterances from TV dialogues. As MELD (Multimodal EmotionLines Dataset) extends the popular EmotionLines dataset, EmotionLines itself is not included here.
|
|
|
|
|
|
46 |
|
47 |
## Appendix 📚
|
48 |
|
49 |
+
Please find an overview of the datasets used for training below. All datasets contain English text. The table summarizes which emotions are available in each of the datasets.
|
50 |
|
51 |
|Name|anger|disgust|fear|joy|neutral|sadness|surprise|
|
52 |
|---|---|---|---|---|---|---|---|
|
|
|
57 |
|MELD, Poria et al. (2019)|Yes|Yes|Yes|Yes|Yes|Yes|Yes|
|
58 |
|SemEval-2018, EI-reg (Mohammad et al. 2018) |Yes|-|Yes|Yes|-|Yes|-|
|
59 |
|
60 |
+
The datasets represent a diverse collection of text types. Specifically, they contain emotion labels for texts from Twitter, Reddit, student self-reports, and utterances from TV dialogues. As MELD (Multimodal EmotionLines Dataset) extends the popular EmotionLines dataset, EmotionLines itself is not included here.
|
61 |
+
|
62 |
+
The model is trained on a balanced subset from the datasets listed above (2,811 observations per emotion, i.e., nearly 20k observations in total). The evaluation accuracy on a holdout test set is 66% (and significantly above the random-chance baseline of 1/7 = 14%).
|