Wildlife crime, recognized as a key challenge under the United Nations Sustainable Development Goals, threatens biodiversity, ecological balance and conservation efforts worldwide. Yet unlike conventional crime domains, its data trail is scattered across unstructured web sources, making systematic analysis difficult for researchers, policymakers and law enforcement agencies. A new open access study in EPJ Data Science addresses this gap with WILD-CRIME, a semantically enriched wildlife crime dataset built by an automated, knowledge-centric framework.
The framework collates information from a variety of sources, filters it for wildlife crime, performs semantic enrichment and consolidates the material into a web-scale knowledge graph that supports analysis. According to the authors, the absence of standardized, semantically structured datasets has so far hampered evidence-driven interventions in this field, and the new resource is designed to close that gap.
A central technical contribution is a lightweight, expert-based classification model that labels articles as belonging to the wildlife-crime class. The model improves F1-score by 4.2% against the best-performing baseline, RoBERTa. The researchers note that relying on a single model proves insufficient for fully disambiguating nuanced cases, such as articles that are crime-related but not wildlife-specific, or vice versa, which motivates their expert-based approach.
To demonstrate the dataset’s practical value, the team validated it on downstream intelligence tasks. These include the identification of wildlife crime hot-spots, the analysis of species target patterns and the study of frequently used transit routes. The framework is complemented by API-driven query and analysis interfaces designed specifically for law enforcement agencies, allowing operational users to interrogate the consolidated knowledge graph directly.
The work, published on 25 September 2026 by Poonam Goyal, Sunil Yadav, Shekhar Rana, Anushree Karve, Prawaal Sharma and Navneet Goyal with colleagues, is released as open access. The published version is being shared early as a citable, peer-reviewed accepted manuscript carrying a permanent DOI, subject to further edits before the final Version of Record. By turning dispersed, unstructured reporting into semantically structured, queryable intelligence, WILD-CRIME offers a data-driven foundation for analyzing and intervening in wildlife crime at scale.
Subject of Research: A semantically enriched web-scale knowledge graph and automated framework for wildlife crime intelligence and analytics
Article Title: WILD-CRIME: a web-scale, semantically enriched knowledge graph for wildlife crime intelligence and analytics
Article References: Goyal, P., Yadav, S., Rana, S., Karve, A., Sharma, P., & Goyal, N. (2026). WILD-CRIME: a web-scale, semantically enriched knowledge graph for wildlife crime intelligence and analytics. EPJ Data Science. https://doi.org/10.1140/epjds/s13688-026-00703-9
Image Credits: AI Generated
DOI: 10.1140/epjds/s13688-026-00703-9
Keywords: wildlife crime, knowledge graph, semantic enrichment, data science, machine learning, RoBERTa, text classification, law enforcement, biodiversity, conservation, open access, EPJ Data Science
News Source: Margaret Porter. (October 28, 2026). New web-scale knowledge graph maps wildlife crime from scattered web sources. Scienmag.
