This dataset contains 510,600 book reviews in Arabic language. The reviews were collected from GoodReads.com website during June/July 2016. This work is an extension of the early dataset of large-scale Arabic dataset, LABR, which has around 63K Arabic Book Reviews collected from GoodReads.com. The reviews are expressed mainly in Modern Standard Arabic but there are reviews in dialectal Arabic as well. The following table summarize some tatistics on the HARD Dataset.
| Property | Number | Property | Number |
|---|---|---|---|
| Number of reviews | 510,598 | Median reviews per book | 37 |
| Number of users | 76,530 | Min reviews per book | 1 |
| Avg. reviews per user | 7 | Number of books | 4993 |
| Max reviews per user | 396 | Avg. reviews per book | 102 |
| Median reviews per user | 2 | Number of tokens | 39,886,898 |
The following figure depicts the reviews distribution in BRAD Datasets; balanced (inner) and unbalanced (outer).
You may refer to our paper, cited below, for details on the dataset.
-
data/
-
bal-reviews.csv.rar: a zipped CSV file containing the balanced dataset of positive and negative reviews. Only positive (ratings 4 & 5) and negative (ratings 1 & 2) reviews are included. The dataset consists of a little bit more than 156K reviews. The format of each review record is:
rating<TAB>review_id<TAB>book_id<TAB>user_id<TAB>review -
unbal-reviews.csv.rar: the whole dataset of more than 510K reviews. This is a clean dataset that includes all reviews. The format of each review record is:
rating<TAB>review_id<TAB>book_id<TAB>user_id<TAB>reviewwhere: rating: the user rating on a scale of 1 to 5 review id: the id of the review (to access the review of a specific review) book id: the id of the hotel user id: the type of user (single, couple, ...) review: the text of the review
-
Please cite the following paper if you decise to use the dataset: Conference: Conference:
Elnagar A. and Einea O. 'BRAD 1.0: Book reviews in Arabic dataset'. 2016 IEEE/ACS 13th International Conference of
Computer Systems and Applications (AICCSA), pp. 1-8, Nov 2016. DOI: 10.1109/AICCSA.2016.7945800.