Description

This a set of Arabic book reviews containing over 63,000 reviews. This is the largest sentiment analysis dataset in Arabic to-date. The dataset was downloaded from www.goodreads.com during the month of March 2013. The package contains the cleaned up reviews, together with a utility class in Python that provides an easy interface to loading the standard training and tests. More information is available in the reference below and the README file.

Download

• LABR v2.0 [11.6 MB] or browse and download the code and data from GitHub.
• Includes standard splits of the data into training, validation, and testing, as well as scripts to reproduce the basic experiments described in [2].
• Contains splits into three sentiment polarities: positive, negative, and neutral instead of just two classes as in version 1.
  
  
• LABR v1.0 [8.5 MB] or browse and download the code and data from GitHub.
• Includes standard splits of the data into training, validation, and testing, as well as scripts to reproduce the basic experiments described in [1].

References

1. Mohamed Aly and Amir Atiya. LABR: Large Scale Arabic Book Reviews Dataset, Meetings of the Association of Computational Linguistics (ACL), Sofia, Bulgaria, August 2013. [pdf]
2. Mahmoud Nabil, Mohamed Aly, and Amir Atiya. LABR 2.0: Large Scale Arabic Sentiment Analysis Benchmark. arXiv e-print (arXiv:1411.6718), 2014. [pdf]