CSDE Computational Demography Working Group (CDWG): Jing Xu & Yehong Deng (05/13/26)

Posted: 4/30/2026 (Local Events)

The next Computational Demography Working Group speaker will be Dr. Jing Xu and Yehong Deng (University of Washington). The talk is hybrid and will take place on May 13th in Raitt 223 from 10-11am (Pacific). Use this link to register and log onto Zoom. To receive the newsletter from CDWG, participants may choose to join our listserv here.

Title: Youth & Truth in Northern Ireland: LLM-Empowered Knowledge Graph Analysis of History Textbooks

Abstract: Post-conflict societies face a persistent challenge: how to reconcile divergent historical memories embedded in educational materials and teach the next generation. Northern Ireland provides a significant test case to examine this question, where the school system remains largely divided along group identity lines in a polarized society. Our larger project on youth and truth in NI, sponsored by the Templeton World Charity Foundation, found that history curriculum in Catholic Schools and state (Protestant) schools focus on different time periods, Option 1 (1921-1949) and Option 2 (Troubles era), during a crucial phase in students’ formative years (KS4, 14-16-year-old).

This talk presents a novel computational analysis of history textbooks focused on the two periods, Option1 and Option 2, combining LLM-powered knowledge graph (KG) with narrative network analysis. Drawing on a refined ontology, we extract subject–predicate–object triples from a corpus of history textbooks, yielding knowledge graphs of approximately 500–800 nodes and 600–1,000 edges per tradition. We analyze structural properties including centrality, reachability, community clustering, and sentiment.
Our comparative analysis reveals systematic divergences in how historical events, entities and figures are positioned across the two traditions. Option 1 textbooks center on entities and events associated with WWII and Anglo-Irish constitutional relations, while Option 2 texts focus on civil rights, internment, and power-sharing arrangements in the Troubles era. Sentiment analysis further shows that the same entities receive markedly different evaluative framings across corpora. These structural asymmetries in narrative construction may reflect and reinforce the polarized historical identities that post-conflict education efforts seek to bridge.

Learn more about Dr. Jing Xu: Affiliate Professor in Anthropology & Co-Director of the Center for Globally Beneficial AI at the University of Washington. Xu’s research integrates ethnographic, experimental and computational methods to study moral development, social learning, and AI ethics in cultural contexts. She studied at Tsinghua University (Beijing) before moving to the U.S., got her PhD in anthropology from Washington University in St. Louis, and completed postdoctoral work in psychology at the University of Washington. She is the author of two monographs, “The Good Child” (2017) and “‘Unruly’ Children” (2024). She is an Associate Editor of American Anthropologist.

Yehong Deng is a fourth-year PhD student in Sociology at the University of Washington. Her research sits at the intersection of computational social science, digital humanities, and peace education, with a focus on how historical narratives are structured and transmitted in post-conflict contexts. Her dissertation applies large language models and network analysis to study how Northern Ireland history textbooks across denominational traditions construct divergent historical narratives, drawing on knowledge graph extraction and comparative discourse analysis.

Read Full Article

Date: 05/13/2026

Time: 10 - 11 AM

Location: Raitt 223