Tag Recommendation Dataset - Elo7 Products

Elo7 dataset - Files and formats *.test*: test sets*.train.*: train sets wordmap: Contain all words in the dataset's vocabulary and their corresponding integer identifiers. Format: each line contains a pair *.ids: product identifiers*.ground_truth: relevant tags for each object (one product per line, aligned with .ids file) titledesc.csv: title and descriptions of products*.feats (available after the review process of our article): tag quality attributes, in RankLib format, that is: RelevanceLabel qid:number attribute1:value1 attribute2:value ... #TagID Example: 0 qid:1 1:2 2:10 #10 1 qid:1 1:3 2:10 #42 In the above example, there are two candidate tags associated with product 1 (qid:1). The first (ID=10) is non relevant (label=0) while the second (ID=42) is relevant (label=1)