PAR: Political Actor Representation Learning with Social Context and Expert Knowledge

Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 12022 - 12036 December 7-11, 2022 ©2022 Association for Computational Linguistics PAR: Political Actor Representation Learning with Social Context and Expert Knowledge Shangbin Feng♣♠ Zhaoxuan Tan♣ Zilong Chen♣♡ Ningnan Wang♣ Peisheng Yu♣♢ Qinghua Zheng♣ Xiaojun Chang⃝ Minnan Luo♣∗ Xi’an Jiaotong University♣, University of Washington♠, Tsinghua University♡ University of California San Diego♢, University of Technology Sydney⃝ contact: shangbin@cs.washington.edu Abstract Modeling the ideological perspectives of po- litical actors is an essential task in compu- tational political science with applications in many downstream tasks. Existing approaches are generally limited to textual data and voting records, while they neglect the rich social con- text and valuable expert knowledge for holistic ideological analysis. In this paper, we propose PAR, a Political Actor Representation learning framework that jointly leverages social context and expert knowledge. Specifically, we retrieve and extract factual statements about legislators to leverage social context information. We then construct a heterogeneous information network to incorporate social context and use relational graph neural networks to learn legislator rep- resentations. Finally, we train PAR with three objectives to align representation learning with expert knowledge, model ideological stance consistency, and simulate the echo chamber phenomenon. Extensive experiments demon- strate that PAR is better at augmenting political text understanding and successfully advances the state-of-the-art in political perspective de- tection and roll call vote prediction. Further analysis proves that PAR learns representations that reflect the political reality and provide new insights into political behavior. 1 Introduction Modeling the perspectives of political actors has ap- plications in various downstream tasks such as roll call vote prediction (Mou et al., 2021) and political perspective detection (Feng et al., 2021). Exist- ing approaches generally focus on voting records or textual information of political actors to induce their stances. Ideal point model (Clinton et al., 2004) is one of the most widely used approach for vote-based analysis, while later works enhance the ideal point model (Kraft et al., 2016; Gu et al., 2014; Gerrish and Blei, 2011) and yield promising ∗Corresponding author. New York State Republican Party Rep. Alexandria Ocasio-Cortez (D-NY) Rep. Alexandria Ocasio-Cortez (D-NY) Rep. Jim Joran (R-OH) Rep. Jim Joran (R-OH) Social Context Sen. Susan Collins (R-ME) Sen. Susan Collins (R-ME) Sen. Sherrod Brown (D-OH) Sen. Sherrod Brown (D-OH) Expert Knowledge AFL-CIO Scores Rep. Alexandria Ocasio- Cortez (D-NY) 95% Rep. Jim Jordan (R-OH) 11% Sen. Susan Collins (R-ME) 75% Sen. Sherrod Brown (D-OH) 92% Rep. Elise Stefanik (R-NY) 37% Heritage Action Scores Rep. Alexandria Ocasio- Cortez (D-NY) 8% Rep. Jim Jordan (R-OH) 96% Sen. Susan Collins (R-ME) 47% Sen. Sherrod Brown (D-OH) 0% Rep. Elise Stefanik (R-NY) 88% Rep. Elise Stefanik (R-NY) Rep. Elise Stefanik (R-NY) Ohil State Ohil State Maine State Maine State Democratic Party Democratic Party Figure 1: Social context information and political expert knowledge that helps to model political actors. results on the task of roll call vote prediction. For text-based methods, text analysis techniques are combined with textual information in social media posts (Li …

Content truncated to 500 words