Bottom-up Anytime Discovery of Generalised Multimodal Graph Patterns for Knowledge Graphs

TL;DR

An anytime algorithm for the bottom-up discovery of generalized multimodal graph patterns in knowledge graphs that are converted to SPARQL queries and presented in an interactive facet browser together with metadata and provenance information, enabling scholars to explore, analyse, and share queries.

Abstract

Vast amounts of heterogeneous knowledge are becoming publicly available in the form of knowledge graphs, often linking multiple sources of data that have never been together before, and thereby enabling scholars to answer many new research questions. It is often not known beforehand, however, which questions the data might have the answers to, potentially leaving many interesting and novel insights to remain undiscovered. To support scholars during this scientific workflow, we introduce an anytime algorithm for the bottom-up discovery of generalized multimodal graph patterns in knowledge graphs. Each pattern is a conjunction of binary statements with (data-) type variables, constants, and/or value patterns. Upon discovery, the patterns are converted to SPARQL queries and presented in an interactive facet browser together with metadata and provenance information, enabling scholars to explore, analyse, and share queries. We evaluate our method from a user perspective, with the help of domain experts in the humanities.

Figures (5)

• Figure 1: An example of a subgraph in the civil registry domain (left) and a possible graph pattern (right). Circles and squares represent entities and classes, respectively, and attribute values are within quotation marks. The coloured area indicates the structural component of the pattern, whereas the distribution conveys the non-structural component.
• Figure 2: Updating the domain of a pattern $\phi$ after adding clause $p(b,c)$. Domains are depicted as sets with integer-encoded resources, whereas the maps between resources represent assertions in the graph. Since resource $6$ is not connected to any of the resources in the domain of $c$, adding $p(b,c)$ thus reduces the domain of $b$ (by removing resource 6), which, in turn, reduces that of $a$ (by removing resource 3).
• Figure 3: A screenshot of the facet browser showing a graph pattern encoded as SPARQL query and visualized as a graph.
• Figure 4: Responses on a 5-point Likert scale (Kendall's $W = 0.14$) about the perceived novelty, validity, utility, and interpretability of the presented graph patterns.
• Figure 5: Responses on a 5-point Likert scale (Kendall's $W = 0.11$) about the perceived helpfulness, intuitiveness, understandability, and pleasantness of the pattern browser.