@@ -7,8 +7,8 @@ This dataset includes reviews (ratings, text, helpfulness votes), product metada
This dataset is probably preferable for sentiment analysis type tasks.
## Link to dataset <br/>
aggressively deduplicated data (18gb)
## Link to dataset
[aggressively deduplicated data (18gb)](http://snap.stanford.edu/data/amazon/productGraph/aggressive_dedup.json.gz)
No duplicates whatsoever (82.83 million reviews). file removes duplicates more aggressively, removing duplicates even if they are written by different users. This accounts for users with multiple accounts or plagiarized reviews.
