Controlled Territory and Conflict Tracking (CONTACT): (Geo-)Mapping Occupied Territory from Open Source Intelligence

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Mandal, Paul K., Leo, Cole, Hurley, Connor
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915249996693504
author Mandal, Paul K.
Leo, Cole
Hurley, Connor
author_facet Mandal, Paul K.
Leo, Cole
Hurley, Connor
contents Open-source intelligence provides a stream of unstructured textual data that can inform assessments of territorial control. We present CONTACT, a framework for territorial control prediction using large language models (LLMs) and minimal supervision. We evaluate two approaches: SetFit, an embedding-based few-shot classifier, and a prompt tuning method applied to BLOOMZ-560m, a multilingual generative LLM. Our model is trained on a small hand-labeled dataset of news articles covering ISIS activity in Syria and Iraq, using prompt-conditioned extraction of control-relevant signals such as military operations, casualties, and location references. We show that the BLOOMZ-based model outperforms the SetFit baseline, and that prompt-based supervision improves generalization in low-resource settings. CONTACT demonstrates that LLMs fine-tuned using few-shot methods can reduce annotation burdens and support structured inference from open-ended OSINT streams. Our code is available at https://github.com/PaulKMandal/CONTACT/.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13730
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Controlled Territory and Conflict Tracking (CONTACT): (Geo-)Mapping Occupied Territory from Open Source Intelligence
Mandal, Paul K.
Leo, Cole
Hurley, Connor
Computation and Language
Artificial Intelligence
Machine Learning
I.2.7; I.2.6; I.2.8; H.3.1; K.4.1
Open-source intelligence provides a stream of unstructured textual data that can inform assessments of territorial control. We present CONTACT, a framework for territorial control prediction using large language models (LLMs) and minimal supervision. We evaluate two approaches: SetFit, an embedding-based few-shot classifier, and a prompt tuning method applied to BLOOMZ-560m, a multilingual generative LLM. Our model is trained on a small hand-labeled dataset of news articles covering ISIS activity in Syria and Iraq, using prompt-conditioned extraction of control-relevant signals such as military operations, casualties, and location references. We show that the BLOOMZ-based model outperforms the SetFit baseline, and that prompt-based supervision improves generalization in low-resource settings. CONTACT demonstrates that LLMs fine-tuned using few-shot methods can reduce annotation burdens and support structured inference from open-ended OSINT streams. Our code is available at https://github.com/PaulKMandal/CONTACT/.
title Controlled Territory and Conflict Tracking (CONTACT): (Geo-)Mapping Occupied Territory from Open Source Intelligence
topic Computation and Language
Artificial Intelligence
Machine Learning
I.2.7; I.2.6; I.2.8; H.3.1; K.4.1
url https://arxiv.org/abs/2504.13730