Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Anomaly Detection and Classification in Knowledge Graphs

Asara Senaratne, Peter Christen, Pouya Omran, Graham Williams

TL;DR

SEKA advances unsupervised anomaly detection in knowledge graphs by combining Corroborative Path based feature generation with one-class $ν$-SVM classification, complemented by ENTGENE for entity typing and TAXO for a comprehensive anomaly taxonomy. The approach detects both anomalous triples and anomalous entities using only intrinsic KG information, without external supervision, and demonstrates improved precision, recall, and downstream KG completion performance across four real-world KGs. The TAXO taxonomy provides a structured framework to classify and reason about anomalies, aiding domain experts in remediation and guiding future algorithmic development. Overall, SEKA offers a scalable, interpretable, and effective toolkit for enhancing KG quality and utility, with planned extensions to dynamic and temporal graphs.

Abstract

Anomalies such as redundant, inconsistent, contradictory, and deficient values in a Knowledge Graph (KG) are unavoidable, as these graphs are often curated manually, or extracted using machine learning and natural language processing techniques. Therefore, anomaly detection is a task that can enhance the quality of KGs. In this paper, we propose SEKA (SEeking Knowledge graph Anomalies), an unsupervised approach for the detection of abnormal triples and entities in KGs. SEKA can help improve the correctness of a KG whilst retaining its coverage. We propose an adaption of the Path Rank Algorithm (PRA), named the Corroborative Path Rank Algorithm (CPRA), which is an efficient adaptation of PRA that is customized to detect anomalies in KGs. Furthermore, we also present TAXO (TAXOnomy of anomaly types in KGs), a taxonomy of possible anomaly types that can occur in a KG. This taxonomy provides a classification of the anomalies discovered by SEKA with an extensive discussion of possible data quality issues in a KG. We evaluate both approaches using the four real-world KGs YAGO-1, KBpedia, Wikidata, and DSKG to demonstrate the ability of SEKA and TAXO to outperform the baselines.

Anomaly Detection and Classification in Knowledge Graphs

TL;DR

SEKA advances unsupervised anomaly detection in knowledge graphs by combining Corroborative Path based feature generation with one-class -SVM classification, complemented by ENTGENE for entity typing and TAXO for a comprehensive anomaly taxonomy. The approach detects both anomalous triples and anomalous entities using only intrinsic KG information, without external supervision, and demonstrates improved precision, recall, and downstream KG completion performance across four real-world KGs. The TAXO taxonomy provides a structured framework to classify and reason about anomalies, aiding domain experts in remediation and guiding future algorithmic development. Overall, SEKA offers a scalable, interpretable, and effective toolkit for enhancing KG quality and utility, with planned extensions to dynamic and temporal graphs.

Abstract

Anomalies such as redundant, inconsistent, contradictory, and deficient values in a Knowledge Graph (KG) are unavoidable, as these graphs are often curated manually, or extracted using machine learning and natural language processing techniques. Therefore, anomaly detection is a task that can enhance the quality of KGs. In this paper, we propose SEKA (SEeking Knowledge graph Anomalies), an unsupervised approach for the detection of abnormal triples and entities in KGs. SEKA can help improve the correctness of a KG whilst retaining its coverage. We propose an adaption of the Path Rank Algorithm (PRA), named the Corroborative Path Rank Algorithm (CPRA), which is an efficient adaptation of PRA that is customized to detect anomalies in KGs. Furthermore, we also present TAXO (TAXOnomy of anomaly types in KGs), a taxonomy of possible anomaly types that can occur in a KG. This taxonomy provides a classification of the anomalies discovered by SEKA with an extensive discussion of possible data quality issues in a KG. We evaluate both approaches using the four real-world KGs YAGO-1, KBpedia, Wikidata, and DSKG to demonstrate the ability of SEKA and TAXO to outperform the baselines.

Paper Structure

This paper contains 24 sections, 1 figure, 11 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries
  4. SEKA: Seeking Knowledge Graph Anomalies
  5. Feature Generation
  6. Learning a One-Class SVM
  7. Entity Type Generation for Anomaly Detection
  8. TAXO: Taxonomy of Knowledge Graph Anomalies
  9. Missingness
  10. Incorrectness
  11. Inconsistency
  12. Ambiguity
  13. Contradictions
  14. Unusual
  15. Redundant Triples
  16. ...and 9 more sections

Figures (1)

  • Figure 1: Overview of SEKA, the anomaly detection process to identify anomalous triples and entities in a KG. The abnormal triples and entities are marked in red in the KG on the left-hand side of the figure (taken from senaratne2023seka).

Theorems & Definitions (3)

  • Definition 1: Path
  • Definition 2: Neighborhood
  • Definition 3: Knowledge Graph