Abstract:Knowledge Graph Completion (KGC) has emerged as a promising solution to address the issue of incompleteness within Knowledge Graphs (KGs). Traditional KGC research primarily centers on triple classification and link prediction. Nevertheless, we contend that these tasks do not align well with real-world scenarios and merely serve as surrogate benchmarks. In this paper, we investigate three crucial processes relevant to real-world construction scenarios: (a) the verification process, which arises from the necessity and limitations of human verifiers; (b) the mining process, which identifies the most promising candidates for verification; and (c) the training process, which harnesses verified data for subsequent utilization; in order to achieve a transition toward more realistic challenges. By integrating these three processes, we introduce the Progressive Knowledge Graph Completion (PKGC) task, which simulates the gradual completion of KGs in real-world scenarios. Furthermore, to expedite PKGC processing, we propose two acceleration modules: Optimized Top-$k$ algorithm and Semantic Validity Filter. These modules significantly enhance the efficiency of the mining procedure. Our experiments demonstrate that performance in link prediction does not accurately reflect performance in PKGC. A more in-depth analysis reveals the key factors influencing the results and provides potential directions for future research.

What problem does this paper attempt to address?

The paper primarily focuses on addressing the issue of incompleteness in Knowledge Graphs (KGs) and proposes a Knowledge Graph Completion (KGC) method that is more aligned with real-world application scenarios. Specifically, traditional KGC research has mainly concentrated on tasks such as triple classification and link prediction, but these tasks do not fully meet the needs of real-world scenarios. Therefore, the paper introduces three key processes to simulate the construction of knowledge graphs in the real world: 1. **Verification Process**: Considering the necessity and limitations of manual verification, this process emphasizes the need for human involvement in validating the knowledge generated by the model in practical applications. 2. **Mining Process**: This process is used to identify the most likely correct candidate facts for subsequent manual verification, aiming to improve efficiency and accuracy. 3. **Training Process**: Utilizing the verified data to further optimize the model, thereby achieving iterative improvements. Based on these three processes, the authors introduce the "Progressive Knowledge Graph Completion" (PKGC) task, which simulates the gradual perfection process of knowledge graphs in real-world scenarios. Moreover, to accelerate the PKGC process, the paper also proposes two acceleration modules: an optimized Top-k algorithm and a Semantic Validity Filter (SVF). These modules significantly enhance the efficiency of the mining process. In summary, the paper attempts to address the issues present in traditional KGC tasks by proposing a method that is closer to real-world application scenarios and validates the effectiveness of the proposed method through experiments.

Progressive Knowledge Graph Completion

Mixed Geometry Message and Trainable Convolutional Attention Network for Knowledge Graph Completion

A Re-evaluation of Knowledge Graph Completion Methods

Knowledge Graph Completion: A Review

Step out of KG: Knowledge Graph Completion via Knowledgeable Retrieval and Reading Comprehension

A comprehensive overview of knowledge graph completion

Multi-Level Interaction Based Knowledge Graph Completion

Multi-level Shared Knowledge Guided Learning for Knowledge Graph Completion

Knowledge Graph Completion with Information Adaptation and Refinement

Simple knowledge graph completion model based on PU learning and prompt learning

Generative Knowledge Graph Construction: A Review

Semantic-Enhanced Knowledge Graph Completion

KERMIT: Knowledge Graph Completion of Enhanced Relation Modeling with Inverse Transformation

Relational Message Passing for Fully Inductive Knowledge Graph Completion

Explainable Sparse Knowledge Graph Completion Via High-order Graph Reasoning Network

Retrieval, Reasoning, Re-ranking: A Context-Enriched Framework for Knowledge Graph Completion

Exploring & Exploiting High-Order Graph Structure for Sparse Knowledge Graph Completion

A Pre-training Framework for Knowledge Graph Completion

Do Pre-trained Models Benefit Knowledge Graph Completion? A Reliable Evaluation and a Reasonable Approach.

KGExplainer: Towards Exploring Connected Subgraph Explanations for Knowledge Graph Completion

Large Models and Multimodal: A Survey of Cutting-Edge Approaches to Knowledge Graph Completion