CITATION.cff

cff-version: 1.2.0
message: "If you use this benchmark, please cite both the artifact and the accompanying paper."
title: "The Artha Personal-Finance Reasoning Benchmark"
abstract: >-
  A reproducible, model-agnostic benchmark for evaluating LLM-based
  personal-finance agents on answer quality (aggregative and multi-hop
  reasoning over a full transaction ledger) and grounding / hallucination
  resistance (verified false-premise traps), over a BLS-calibrated
  synthetic transaction population, with a four-dimension three-judge
  LLM-as-judge scoring protocol.
version: "1.0"
date-released: "2026-01-01"
license: CC-BY-4.0
type: dataset
authors:
  - family-names: "Katika"
    given-names: "Tejashwar Reddy"
    affiliation: "University of North Texas"
    ORCID:       "https://orcid.org/0009-0006-3015-5697"
keywords:
  - personal finance
  - LLM agents
  - benchmark
  - hallucination
  - grounding
  - LLM-as-judge
  - BLS calibration
# Paste the DOI minted by Hugging Face (dataset Settings -> "Generate DOI") here:
doi: "10.57967/hf/9387"
repository-code: "https://huggingface.co/datasets/Tej-Katika/artha-benchmark"
url: "https://github.com/Tej-Katika/artha"
preferred-citation:
  type: article
  title: "Artha: A Domain Ontology-Driven Agentic Framework for LLM-Based Personal Finance Reasoning"
  year: 2026
  authors:
    - family-names: "Katika"
      given-names: "Tejashwar Reddy"
  journal: "SSRN Electronic Journal"
  url: "https://ssrn.com/abstract=6885058"
  notes: "SSRN preprint."