Wikipedia_Talk_Labels

This data set includes over 100k labeled discussion comments from English Wikipedia. Each comment was labeled by multiple annotators via Crowdflower on whether it contains a personal attack. We also include some demographic data for each crowd-worker. See our wiki for documentation of the schema of each file and our research paper for documentation on the data collection and modeling methodology. For a quick demo of how to use the data for model building and analysis, check out this ipython notebook. This is originally a multi-label classification problem. However, for the purpose of this task, we have combined the labels into a single column. Create a Multiclass Target Column: Encode the attack labels (recipient_attack,third_party_attack,attack, etc.) as a single class. # Combine labels into a single column def encode_multiclass(row): return ','.join(row.index[row == 1]) Here, each combination of the labels results in a unique class. For example, a comment with the labels recipient_attack and third_party_attack will have a class of recipient_attack,third_party_attack. And so on. Drop Original Labels after creating the class column. We created the final dataset by merging the three datasets attack_annotated_comments.tsv, attack_annotations.tsv, and attack_worker_demographics.tsv. This is the order we followed to merge the datasets: 1. Merge attack_annotated_comments.tsv with attack_annotations.tsv on 'rev_id' 2. Merge the resultant dataset with attack_worker_demographics.tsv on 'worker_id' paper_url = "https://arxiv.org/pdf/1610.08914" original_data_url = "https://figshare.com/articles/dataset/Wikipedia_Talk_Labels_Personal_Attacks/4054689"