Abstract

Arabic blogs have emerged in the past few years as new mean of communication on the Internet. Recently, they started gaining a lot of attention from the Arabic population and the media. They represent a voice for the individuals to broadcast openly their thoughts, views and opinions about Arabic and international affairs.

In this study, we emphasize the nature and the characteristics of the Arabic blogs. This is done by building the Arabic blogs dataset and performing various experiments to describe language usage and topics in blogs. Machine learning methods are used to highlight certain properties of blogs. The experiment results showed that a good distinction can be made between the Arabic language and at least one of its dialects. From the experiments, it became clear that certain topics can be sufficiently identified in blogs. However, the identification of other topics can be challenging.