INAGQA: German Financial Question-Answering System Outputs

2,100 expert-annotated German financial questions were used to validate the INAGQA question-answering system. The dataset likely contains the system's outputs, including parsed queries and linked knowledge graph entities. Authored by Jamal Al Qundus and last updated on 2026-05-04, it is shared under a CC-BY-4.0 license.

Use Cases

• Benchmarking German-language QA systems based on the reported 0.91 F1 score.
• Analyzing headword-centric parsing performance for compound nouns like 'Eigenkapitalrendite'.
• Evaluating temporal and quantitative query handling based on the described 98% accuracy for variants.
• Testing knowledge integration and response times in financial analyst case studies.

Strengths

• Validation on 2,100 expert-annotated German financial queries.
• Reported system performance metrics include a 0.91 F1 score and 2.1s average response time.
• Describes a 35% reduction in relation-linking errors for compound nouns.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small at 5.5 KB, indicating limited scope.

Provenance

Source

figshare

Collection Method

Outputs from the INAGQA QA system, validated on expert-annotated questions.

Freshness

Last updated 2026-05-04 17:51:19; freshness should be verified.