Hierarchical Structured Model for Fine-to-coarse Manifesto Text Analysis
This provides a tool for political scientists and analysts to efficiently process multilingual manifesto data, though it is incremental as it builds on existing methods for text analysis.
The paper tackles the problem of automatically analyzing political party manifestos at both fine-grained issue positions and coarse-grained left-right spectrum positioning, proposing a two-stage hierarchical multi-task structured deep model with post-hoc calibration that outperforms state-of-the-art approaches across manifestos from twelve countries in ten languages.
Election manifestos document the intentions, motives, and views of political parties. They are often used for analysing a party's fine-grained position on a particular issue, as well as for coarse-grained positioning of a party on the left--right spectrum. In this paper we propose a two-stage model for automatically performing both levels of analysis over manifestos. In the first step we employ a hierarchical multi-task structured deep model to predict fine- and coarse-grained positions, and in the second step we perform post-hoc calibration of coarse-grained positions using probabilistic soft logic. We empirically show that the proposed model outperforms state-of-art approaches at both granularities using manifestos from twelve countries, written in ten different languages.