CogMap: A Cognitive Support Approach to Property and Instance Alignment
The iterative user interaction approach for data integration proposed by Falconer and Noy can be generalized to consider interactions between integration tools (generators) that generate potential schema mappings and users or analysis tools (analyzers) th
- PDF / 1,045,761 Bytes
- 17 Pages / 439.37 x 666.142 pts Page_size
- 30 Downloads / 237 Views
Abstract. The iterative user interaction approach for data integration proposed by Falconer and Noy can be generalized to consider interactions between integration tools (generators) that generate potential schema mappings and users or analysis tools (analyzers) that select the best mapping. Each such selection then provides high-confidence guidance for the next iteration of the integration tool. We have implemented this generalized approach in CogMap, a matching system for both property and instance alignments between heterogeneous data. The generator in CogMap uses the instance alignment from the previous iteration to create high-quality property alignments and presents these alignments and their consequences to the analyzer. Our experiments show that multiple iterations as well as the interplay between instance and property alignment serve to improve the final alignments.
1
Introduction
In recent years, companies have spent more and more effort in building knowledge graphs based on light-weight ontologies, which incorporate data from multiple heterogeneous sources (which we will henceforth call “information stores”). A key challenge of these efforts is determining the best alignment of the schema of a new store to the ontology of the knowledge graph, while minimizing the “manual” analytical effort required of a human knowledge engineer. Most of the current ontology alignment systems, such as those evaluated recently in the annual ontology alignment evaluation initiative [1], have several limitations. Most of these alignment algorithms solve one integration problem (deriving a mapping between two ontologies) using a fully-automated, “one-shot” approach. Thus, they are often not able to improve by iterating over previous alignments. Partly for this reason, the results of fully automated algorithms are often error prone [32] and cannot be reliably used for high-quality data integration. Currently much information to be integrated is obtained from non-ontological sources such as relational databases or XML documents. Classical ontology alignment systems are often not able to process this data [1]. To address this need, systems like OntoDB [20] and standards like D2RQ [3] have emerged. However, c Springer International Publishing Switzerland 2015 M. Arenas et al. (Eds.): ISWC 2015, Part I, LNCS 9366, pp. 269–285, 2015. DOI: 10.1007/978-3-319-25007-6 16
270
J. N¨ oßner et al.
these solutions do not include semi-automated alignment algorithms which take instance information into account. Our approach, implemented in the CogMap system, follows a cognitivelyinspired, iterative approach. With multiple iterations the system is able to improve over time, since it builds on the results of previous iterations (or, in the case of the first iteration, seed queries given by the user). At each iteration, the results are augmented with new information that has been verified by a user or automated verification capability. CogMap uses instance information to perform property alignment. While most state-of-the-art schema alignment alg
Data Loading...
