Develop a method for fine-grained classification of disagreement and editor-editor conflict networks on Wikipedia
- Data basis: Article content with meta-data on each word, consisting of all deletes and reintroductions over time by different editors + additional metadata as needed about edits, editors, talk pages, etc. Example articles will be provided that contain interesting disagreement structures between editors.
- Goal: Classify disagreement on certain words or sections on the range between mere corrections to full-blown opinion clashes (or develop your own dimensions), try to develop roles of editors in conflict. Best possible outcome: an automated method that “tells the story” of e.g. an edit war as accurately as possible.
- Use existing literature on controversy and Wikipedias talk pages and other resources to gain theoretical insight; to build a ground truth, hand-label some conflict structures in Wikipedia articles
- Automatic labeling/categorization: e.g., by filtering stopwords, POS tagging, Word2Vec, or using edit comments and the temporal sequence of edits.
- Team: should consist of at least 50% members with medium to good programming skills