KQA Pro: A Dataset with Explicit Compositional Programs for Complex Question Answering over Knowledge Base

TL;DR

KQA Pro is a large, diverse dataset for complex knowledge base question answering that includes explicit reasoning programs, revealing current models' limitations and guiding future research in compositional reasoning.

Contribution

The paper introduces KQA Pro, a new dataset with explicit reasoning programs and diverse questions, advancing the evaluation and development of complex KBQA systems.

Findings

State-of-the-art KBQA models perform poorly on KQA Pro

KQA Pro includes explicit reasoning programs and SPARQL queries

The dataset challenges current models and highlights research gaps

Abstract

Complex question answering over knowledge base (Complex KBQA) is challenging because it requires various compositional reasoning capabilities, such as multi-hop inference, attribute comparison, set operation. Existing benchmarks have some shortcomings that limit the development of Complex KBQA: 1) they only provide QA pairs without explicit reasoning processes; 2) questions are poor in diversity or scale. To this end, we introduce KQA Pro, a dataset for Complex KBQA including ~120K diverse natural language questions. We introduce a compositional and interpretable programming language KoPL to represent the reasoning process of complex questions. For each question, we provide the corresponding KoPL program and SPARQL query, so that KQA Pro serves for both KBQA and semantic parsing tasks. Experimental results show that SOTA KBQA methods cannot achieve promising results on KQA Pro as on current datasets, which suggests that KQA Pro is challenging and Complex KBQA requires further research efforts. We also treat KQA Pro as a diagnostic dataset for testing multiple reasoning skills, conduct a thorough evaluation of existing models and discuss further directions for Complex KBQA. Our codes and datasets can be obtained from https://github.com/shijx12/KQAPro_Baselines.