Social-RAG
3 views
Skip to first unread message
L N
Feb 20, 2026, 1:15:38 PM (7 days ago) Feb 20
to computation...@googlegroups.com
Dear colleagues,
I would like to share a recent paper from our group introducing Social-RAG, a Retrieval-Augmented Generation (RAG) architecture designed for AI-assisted qualitative analysis of large digital corpora (e.g., Telegram messages).
Instead of generating answers from parametric memory alone, the system first retrieves relevant passages from the dataset and then produces responses grounded in those excerpts. This enables semantic search, traceability of evidence, and verification of claims while preserving interpretive analysis rather than replacing it.
The paper details the architecture, methodological choices, and implications for computational social science research.
[Preprint] Article: https://doi.org/10.31235/osf.io/wmc2q_v1
Best,
I would like to share a recent paper from our group introducing Social-RAG, a Retrieval-Augmented Generation (RAG) architecture designed for AI-assisted qualitative analysis of large digital corpora (e.g., Telegram messages).
Instead of generating answers from parametric memory alone, the system first retrieves relevant passages from the dataset and then produces responses grounded in those excerpts. This enables semantic search, traceability of evidence, and verification of claims while preserving interpretive analysis rather than replacing it.
The paper details the architecture, methodological choices, and implications for computational social science research.
[Preprint] Article: https://doi.org/10.31235/osf.io/wmc2q_v1
Best,
Leonardo F. Nascimento
Associate Professor at Federal University of Bahia
Digital Humanities Laboratory - LABHDUFBA
CV Lattes: http://lattes.cnpq.br/7141811368487014
Website: http://leofn.com
LABHDUFBA: http://www.labhd.ufba.br/
Orcid: https://orcid.org/0000-0003-2929-1115
Github: https://github.com/leofn
Twitter: https://twitter.com/leofn3
Reply all
Reply to author
Forward
0 new messages