Abstract
This paper analyzed the “people’s” perceptions of threats to Popular Mobilization Units (PMUs) — an umbrella organization of former Shiite militias criticized for instigating sectarian conflict in Iraq — and its transformation by applying quantitative text analysis of Twitter data.
The PMUs, which emerged during the chaos sparked by the fall of Mosul at the hands of IS in 2014, were perceived as “heroes” for saving Iraq from a national crisis as it liberated a number of IS-controlled areas successively. Once the PMUs were exposed for their violations against Sunnis, however, people began to recognize them as a “threat,” one which provoked sectarian conflicts. To clarify this transformation in people’s ambivalent perceptions of the PMUs, we collected all tweets in Arabic related to the PMUs posted on Twitter and analyzed them using a machine learning method called Latent Semantic Scaling.
The results revealed that people’s perception of threats to the PMUs is higher when it is discussed within the context of Iran, which provides full support for the PMUs, compared with other topics. In addition, people’s perceptions of the threat tended to increase almost twice as much when the PMUs’ political influence expanded, rather than when its military influence increased.
