Are Missing Links Predictable? An Inferential Benchmark for Knowledge Graph Completion

TL;DR

InferWiki is a new knowledge graph completion benchmark that emphasizes inferential reasoning, uses rule-guided data splitting, and includes diverse inference patterns to better evaluate models' inferential capabilities.

Contribution

The paper introduces InferWiki, a novel KGC dataset with improved inferential challenge, rule-guided splits, and comprehensive evaluation patterns, advancing the benchmarking of reasoning in knowledge graphs.

Findings

InferWiki improves inferential ability assessment.

Models show significant performance gaps across patterns.

The dataset highlights the need for advanced reasoning methods.

Abstract

We present InferWiki, a Knowledge Graph Completion (KGC) dataset that improves upon existing benchmarks in inferential ability, assumptions, and patterns. First, each testing sample is predictable with supportive data in the training set. To ensure it, we propose to utilize rule-guided train/test generation, instead of conventional random split. Second, InferWiki initiates the evaluation following the open-world assumption and improves the inferential difficulty of the closed-world assumption, by providing manually annotated negative and unknown triples. Third, we include various inference patterns (e.g., reasoning path length and types) for comprehensive evaluation. In experiments, we curate two settings of InferWiki varying in sizes and structures, and apply the construction process on CoDEx as comparative datasets. The results and empirical analyses demonstrate the necessity and high-quality of InferWiki. Nevertheless, the performance gap among various inferential assumptions and patterns presents the difficulty and inspires future research direction. Our datasets can be found in https://github.com/TaoMiner/inferwiki