A Probabilistic-Logical Framework for Ontology Matching
| A Probabilistic-Logical Framework for Ontology Matching | |
|---|---|
A Probabilistic-Logical Framework for Ontology Matching
| |
| Bibliographical Metadata | |
| Subject: | Ontology Matching |
| Year: | 2010 |
| Authors: | Mathias Niepert, Christian Meilicke, Heiner Stuckenschmidt |
| Venue | AAAI |
| Content Metadata | |
| Problem: | Link Discovery |
| Approach: | Probabilistic-logical framework for ontology matching based on Markov logiP |
| Implementation: | ml-match |
| Evaluation: | Using thresholds on the a-priori similarity measure |
Abstract
Ontology matching is the problem of determining correspondences between concepts, properties, and individuals of different heterogeneous ontologies. With this paper we present a novel probabilistic-logical framework for ontology matching based on Markov logic. We define the syntax and semantics and provide a formalization of the ontology matching problem within the framework. The approach has several advantages over existing methods such as ease of experimentation, incoherence mitigation during the alignment process, and the incorporation of apriori confidence values. We show empirically that the approach is efficient and more accurate than existing matchers on an established ontology alignment benchmark dataset.Property "Has abstract" (as page type) with input value "Ontology matching is the problem of determining correspondences between concepts, properties, and individuals of different heterogeneous ontologies. With this paper we present a novel probabilistic-logical framework for ontology matching based on Markov logic. We define the syntax and semantics and provide a formalization of the ontology matching problem within the framework. The approach has several advantages over existing methods such as ease of experimentation, incoherence mitigation during the alignment process, and the incorporation of apriori confidence values. We show empirically that the approach is efficient and more accurate than existing matchers on an established ontology alignment benchmark dataset." contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.
Conclusion
We presented a Markov logic based framework for ontology matching capturing a wide range of matching strategies. Since these strategies are expressed with a unified syntax and semantics we can isolate variations and empirically evaluate their effects. Even though we focused only on a small subset of possible alignment strategies the results are already quite promising. We have also successfully learned weights for soft formulae within the framework. In cases where training data is not available, weights set manually by experts still result in improved alignment quality. Research related to determining appropriate weights based on structural properties of ontologies is a topic of future work.Property "Has conclusion" (as page type) with input value "We presented a Markov logic based framework for ontology matching capturing a wide range of matching strategies. Since these strategies are expressed with a unified syntax and semantics we can isolate variations and empirically evaluate their effects. Even though we focused only on a small subset of possible alignment strategies the results are already quite promising. We have also successfully learned weights for soft formulae within the framework. In cases</br>where training data is not available, weights set manually by experts still result in improved alignment quality. Research related to determining appropriate weights based on structural properties of ontologies is a topic of future work." contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.
Future work
The framework is not only useful for aligning concepts and properties but can also include instance matching. For this purpose, one would only need to add a hidden predicate modeling instance correspondences. The resulting matching approach would immediately benefit from probabilistic joint inference, taking into account the interdependencies between terminological and instance correspondences.Property "Has future work" (as page type) with input value "The framework is not only useful for aligning concepts and properties but can also include instance matching. For this purpose, one would only need to add a hidden predicate modeling instance correspondences. The resulting matching</br>approach would immediately benefit from probabilistic joint inference, taking into account the interdependencies between terminological and instance correspondences." contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.
Approach
Positive Aspects: The approach has several advantages over existing methods such as ease of experimentation, incoherence mitigation during the alignment process, and the incorporation of a-priori confidence values. In cases where training data is not available, weights set manually by experts still result in improved alignment quality. The framework is not only useful for aligning concepts and properties but can also include instance matching. For this purpose, one would only need to add a hidden predicate modeling instance correspondences. The resulting matching approach would immediately benefit from probabilistic joint inference, taking into account the interdependencies between terminological and instance correspondences.Property "Has PositiveAspects" (as page type) with input value "The approach has several advantages over</br>existing methods such as ease of experimentation, incoherence</br>mitigation during the alignment process, and</br>the incorporation of a-priori confidence values.</br>In cases</br>where training data is not available, weights set manually</br>by experts still result in improved alignment quality.</br>The framework is not only useful for aligning concepts and</br>properties but can also include instance matching. For this</br>purpose, one would only need to add a hidden predicate</br>modeling instance correspondences. The resulting matching</br>approach would immediately benefit from probabilistic</br>joint inference, taking into account the interdependencies</br>between terminological and instance correspondences." contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.
Negative Aspects: No data available now.
Limitations: No data available now.
Challenges: No data available now.
Proposes Algorithm: No data available now.
Methodology: No data available now.
Requirements: training data
Limitations: No data available now.
Implementations
Download-page: http://code.google.com/p/ml-match/
Access API: No data available now.
Information Representation: No data available now.
Data Catalogue: {{{Catalogue}}}Property "Has DataCatalouge" (as page type) with input value "{{{Catalogue}}}" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.
Runs on OS: No data available now.
Vendor: No data available now.
Uses Framework: TheBeast
Has Documentation URL: https://code.google.com/archive/p/ml-match/wikis/MLExample.wiki
Programming Language: No data available now.
Version: No data available now.
Platform: No data available now.
Toolbox: No data available now.
GUI: No
Research Problem
Subproblem of: Ontology Alignment
RelatedProblem: Concept similarity
Motivation: matching ontologies enables the knowledge and data expressed in the matched ontologies to interoperate.
Evaluation
Experiment Setup: All experiments were conducted on a desktop PC with AMD Athlon Dual Core Processor 5400B with 2.6GHz and 1GB RAM.
Evaluation Method : Average F1-values over the 21 OAEI reference alignments for manual weights vs. learned weights vs. formulation without stability constraints; thresholds range from 0.6 to 0.95.
Hypothesis: No data available now.
Dimensions: Accuracy
Benchmark used: Ontofarm dataset
Results: Using stability constraints improves alignment quality with both learned and manually set weights.