TuneR

Abstract

A rule-based entity matching task requires the definition of an effective set of rules, which is a time-consuming and error-prone process. The typical approach adopted for its resolution is a trial and error method, where the rules are incrementally added and modified until satisfactory results are obtained. This approach requires significant human intervention, since a typical dataset needs the definition of a large number of rules and possible interconnections that cannot be manually managed. In this paper, we propose TuneR, a software library supporting developers (i.e., coders, scientists, and domain experts) in tuning sets of matching rules. It aims to reduce human intervention by offering a tool for the optimization of rule sets based on user-defined criteria (such as effectiveness, interpretability, etc.). Our goal is to integrate the framework in the Magellan ecosystem, thus completing the functionalities required by the developers for performing Entity Matching tasks.

Proposed threshold-based and rule-based approaches to detecting duplicates in bibliographic database

“…The rule-based approach is a trial-and-error-based method that requires human intervention in the form of adding and modifying rules to ensure that satisfactory results are obtained. Research performed by Paganelli et al [28] developed rules in the Magellan ecosystem by developing a software library called TuneR, designed for ease of use, specifically for application developers. Rule formation considers three critical pieces of information: attributes, similarity functions, and threshold values.…”

“…The rule used in duplication detection is a rule for filtering a set of records [26], not a rule for performing the reasoning or inferencing process. The rule is built using several components, including attributes, the similarity function, operators, and threshold values [28], such that the rule is formed using the predicate p: (a, f, op, thr), where a ∈ A, f ∈ F, op ∈ O, and thr ∈ R are the threshold values. A dataset D comprises a set of records, which is d1, ..., dN, while a record has an attribute value that can be denoted as A={a1, ..., am}.…”

Towards Explainable Automatic Knowledge Graph Construction with Human-in-the-Loop

“…Neural LP [129] [61], Abstraction [107] Human-in-the-loop [43] SystemER [75], TuneR [69] Table 1. Overview of explainable knowledge graph construction methods.…”

“…For entity and relation extraction, explanations often refer to contextual cues such as triggers [49,52] and sentences [91]. Explanations for entity resolution tend to use entity matching rules [69,75] and (ranked) attributes of the entity pair [8,26]. Finally, link prediction methods use the topology and reasoning capabilities of the KG.…”

“…There are very few papers considering human inputs or oversight, which are critical in trustworthy AI frameworks and guidance [23]. Human input in isolated cases [43,69,75] often involves providing or revising rules for tasks like entity resolution. Furthermore, most approaches have not been comprehensively evaluated.…”