Fig 1.
Overview of annotation procedure.
Table 1.
Examples of annotated comments.
Table 2.
Statistics of SSD dataset.
Table 3.
LIWC Feature Statistics for Social Support Analysis. All values (except Word Count) represent the average percentage of words in each category relative to total words per comment.
Table 4.
Emotion scores for social support subtasks.
Table 5.
Sentiment analysis scores for social support subtasks.
Table 6.
Parameters for deep learning models.
Table 7.
Classification results using LIWC, emotions and sentiments features.
Table 8.
Classification results using Unigrams with TF-IDF values.
Table 9.
Classification results using all features.
Table 10.
Data classification results for deep learning models.
Table 11.
Classwise scores for the best-performing models for each subtask, including sample sizes from Table 2.
Fig 2.
Confusion matrix for the best-performing model in subtask 1 (SVM (linear) + all features).
Fig 3.
Confusion matrix for the best-performing model in subtask 2 (CNN + Glove).
Fig 4.
Confusion matrix for the best-performing model in subtask 3 (soft voting + TF-IDF).