hasadna/kikar-hamedina

DataScience Research Task - Sentiment Analysis on Facebook Comments

Opened this issue · 0 comments

yotammanor commented

Materials:

  1. An Excel with ~1600 Comments:
    Each comment was written as a response to a Facebook Status published by an Israeli MK, During 2015-2016, randomly selected. Each comment has some 80 extracted features, and 9 manually tagged/classified features regarding the sentiment in the comment. Link here.

  2. An Excel with Feature Description:
    As mentioned above, each Comment was tagged for 9 different Attributes/features/dependent variables. The Codebook describes the feature and classifcation guidelines.
    Link here.

  3. A link to ~5.3M Comments, unclassified. a txt file, one comment per row. Link here.

  4. A More detailed description on the data collection and sampling process, and a discussion on some of its features can be found here (Chapter 2 and onwards).

Goals:

  • Build interesting and reliable predictive models.
  • Any result will be interesting, but a focus on good classification of comment sentiment will be the most useful for current efforts.