INAGQA: German Financial Question-Answering System Evaluation Results

Evaluation data for the INAGQA question-answering system, developed by Jamal Al Qundus and last updated in May 2026. The dataset likely contains performance metrics from experiments comparing INAGQA against systems like BERT-KGQA on German financial queries. Results include an F1 score of 0.91 validated on 2,100 expert-annotated questions.

Use Cases

• Benchmarking German-language QA systems based on the reported F1 scores and comparison data.
• Evaluating headword-centric parsing techniques for financial domain queries based on the system's described innovations.
• Analyzing the impact of domain-optimized shallow parsing on relation-linking error reduction mentioned in the description.
• Studying temporal and quantitative variant handling in QA systems based on the reported 98% accuracy for such cases.

Strengths

• Performance validated on 2,100 expert-annotated German financial questions.
• Reports specific, comparative results including a 0.91 F1 score for INAGQA and scores for baseline systems.
• Describes a 35% reduction in relation-linking errors for compound nouns through domain-optimized parsing.

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

Likely contains experimental results and evaluation metrics from the INAGQA system research.

Freshness

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