Ontology-based entity matching in attributed graphs

Abstract

Keys for graphs incorporate the topology and value constraints needed to uniquely identify entities in a graph. They have been studied to support object identification, knowledge fusion, and social network reconciliation. Existing key constraints identify entities as the matches of a graph pattern by subgraph isomorphism, which enforce label equality on node types. These constraints can be too restrictive to characterize structures and node labels that are syntactically different but semantically equivalent. We propose a new class of key constraints, Ontological Graph Keys (OGKs) that extend conventional graph keys by ontological subgraph matching between entity labels and an external ontology. We show that the implication and validation problems for OGKs are each NP-complete. To reduce the entity matching cost, we also provide an algorithm to compute a minimal cover for OGKs. We then study the entity matching problem with OGKs, and a practical variant with a budget on the matching cost. We develop efficient algorithms to perform entity matching based on a (budgeted) Chase procedure. Using real-world graphs, we experimentally verify the efficiency and accuracy of OGK-based entity matching.

Relation-Aware Entity Matching Using Sentence-BERT

“…Some studies are weighted match correction rules (WMRRs) based on similarity matching to capture more errors [24] and regular expression-based data repair [25]. More recent approaches also compute the minimum coverage of ontology graph keys (OGKs) to do EM [26], and several ideas use classical algorithmic methods to help EM [27][28][29].…”

Attributed Graph Matching via Seeds Guiding

“…[21] applied quantum walk to attributed graph matching. To reduce the cost of time to match the large graph [22], and [23] proposed two high-efficiency algorithms, respectively. For some specific application areas, Almasi et al in Ref.…”

Generating Referring Expressions from RDF Knowledge Graphs for Data Linking

“…The linkage rules introduced above can be used either logically to deduce identity links, or by linking tools where simple similarity measures and aggregation functions can be introduced. Since available existing linking tools like [24,34,23] do not consider such intricate graph patterns (i.e., not just paths of properties), we have developed a simple bottom-up approach explained in Figure 2, where normalizations or classical similarity measures can be declared and applied to datatype properties.…”

“…We have used all REs of depth 1 whose statistics were delineated in Table 1, with strict string equality. The quality of linking results is reported in terms of precision, recall and F-measure, and is compared to the results of linking with keys (Ks), keys and conditional keys (Ks + CKs) reported in [31], ontological graph keys (OGK) reported in [23], and the random baseline (RBL).…”

“…A conditional key is a valid key for a specified part of a class's instances [31]. Ontological graph keys defined in [23] are a variant of keys defined by a graph pattern extended by ontological pattern matching. Table 3 shows that RE-miner outperforms the other approaches in terms of recall and F-measure on all classes except book.…”