Abstract
Domestic violence, a complex public health issue, demands nuanced analysis of survivors' narratives. We propose a hybrid framework integrating Microsoft GraphRAG with GPT-4o-mini to leverage graph-based reasoning and LLM capabilities for scalable, interpretable topic modeling. Analyzing transcribed interviews, GraphRAG achieved a 0.79 coherence score (outperforming LDA, BERTopic, and TopicGPT) and 97% entity accuracy (70 entities). This approach enhances topic modeling for culturally sensitive contexts, offering improved support for survivors.
| Original language | English |
|---|---|
| Pages (from-to) | 1794-1795 |
| Number of pages | 2 |
| Journal | Studies in Health Technology and Informatics |
| Volume | 329 |
| DOIs | |
| Publication status | Published - 7 Aug 2025 |
UN SDGs
This output contributes to the following UN Sustainable Development Goals (SDGs)
-
SDG 3 Good Health and Well-being -
SDG 5 Gender Equality -
SDG 16 Peace, Justice and Strong Institutions
Keywords
- Domestic violence
- GraphRAG
- Large Language Model
- Topic Modeling
ASJC Scopus subject areas
- Biomedical Engineering
- Health Informatics
- Health Information Management
Fingerprint
Dive into the research topics of 'Leveraging GraphRAG with Large Language Models to Identify Help-Seeking Information Among Domestic Violence Survivors from Qualitative Interviews'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver