EUVIMEDEUVIMED entdecken
Crossref

PEARL: Provenance-aware Evidence-grounded Adaptive Retrieval Layer

Shreyas Nagoor Shreyas Nagoor, Bhaskar Anand Bhaskar Anand, Ayush Satpathy Ayush Satpathy, Anitha R Anitha R, S. Kuzhalvaimozhi S. Kuzhalvaimozhi

International Journal of Creative and Open Research in Engineering and Management · 21.05.2026 · 02 (05) · 1-5

DOI: 10.55041/ijcope.v2i5.643 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Creative and Open Research in Engineering and Management, online erschienen am 21.05.2026. Verfasst von Shreyas Nagoor Shreyas Nagoor und 4 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Access to information embedded within large-scale, multilingual government and policy documents remains a criti-cal challenge in low-resource computational environments. This paper presents a Provenance-Aware Multilingual Retrieval-Augmented Generation framework designed for real-time ques-tion answering over long, heterogeneous document collections. The proposed system integrates automated language identifica-tion and bidirectional machine translation within a semantic retrieval pipeline, enabling context-preserving access to docu-ments spanning multiple languages and domains, including legal and policy terminology. By leveraging Sentence Transformer embeddings, FAISS-based vector indexing, and a provenance-grounded generative component, the framework delivers ac-curate, evidence-backed responses without reliance on GPU acceleration. Comprehensive evaluations confirm that the system achieves high extraction fidelity, retrieval relevance, and genera-tive accuracy, with an average end-to-end response latency of 2.1 seconds on standard CPU hardware. Results further demonstrate that the framework consistently outperforms baseline retrieval and generation systems in both multilingual robustness and usability, establishing it as a scalable and deployable solution for document intelligence in resource-constrained settings. Index Terms—Retrieval-Augmented Generation, Multilingual Question Answering, Semantic Retrieval, Large Language Mod-els, Document Intelligence

Zitieren

APA

Nagoor, S. N. S., Anand, B. A. B., Satpathy, A. S. A., R, A. R. A., & Kuzhalvaimozhi, S. K. S. (2026). PEARL: Provenance-aware Evidence-grounded Adaptive Retrieval Layer. International Journal of Creative and Open Research in Engineering and Management, 02(05), 1-5. https://doi.org/10.55041/ijcope.v2i5.643

Vancouver

Nagoor SNS, Anand BAB, Satpathy ASA, R ARA, Kuzhalvaimozhi SKS. PEARL: Provenance-aware Evidence-grounded Adaptive Retrieval Layer. International Journal of Creative and Open Research in Engineering and Management. 2026;02(05):1-5. doi:10.55041/ijcope.v2i5.643

BibTeX
@article{Nagoor2026,
  author = {Nagoor, Shreyas Nagoor Shreyas and Anand, Bhaskar Anand Bhaskar and Satpathy, Ayush Satpathy Ayush and R, Anitha R Anitha and Kuzhalvaimozhi, S. Kuzhalvaimozhi S.},
  title = {PEARL: Provenance-aware Evidence-grounded Adaptive Retrieval Layer},
  journal = {International Journal of Creative and Open Research in Engineering and Management},
  year = {2026},
  volume = {02},
  number = {05},
  pages = {1--5},
  doi = {10.55041/ijcope.v2i5.643}
}
Kostenlos starten →