That’s your evidence?: Classifying stance in online political debate

From AcaWiki
Jump to: navigation, search

Citation: Marilyn A. Walker, Pranav Anand, Rob Abbott, Jean E. Fox Tree, Craig Martelly, and Joseph King That’s your evidence?: Classifying stance in online political debate. Decision Support Sciences (RSS)
DOI (original publisher): 10.1016/j.dss.2012.05.032
Semantic Scholar (metadata): 10.1016/j.dss.2012.05.032
Sci-Hub (fulltext): 10.1016/j.dss.2012.05.032
Internet Archive Scholar (search for fulltext): That’s your evidence?: Classifying stance in online political debate
Download: http://users.soe.ucsc.edu/~maw/papers/wassa article.pdf
Tagged: Computer Science (RSS) online argumentation (RSS), NLP (RSS), stance (RSS), disagreement (RSS), Mechanical Turk (RSS), natural language processing (RSS), discourse analysis (RSS), cue words (RSS)

Summary

This is a journal submission version of the Cats rule and dogs drool!: Classifying stance in online debate. It covers 14, rather than 12 topics; has more extensive figures; and further examples. It uses a third machine learning algorithm (SVM using libLinear) and experiments with MinCut. Examples of posts that are difficult to side are particularly helpful.

The authors further examined 386 incorrectly classified posts and indicate that context was needed nearly 30% of the time (6% were pure disagreement without content, 12% needed more context, 12% contained quotations as context but the algorithm took these as verbatim statements of the post). 33% were "qualitatively labeled as 'hard'").

Selected references

Theoretical and Practical Relevance

Significant review of related work, embedded in the paper.