Abstract:
Recent advances in information and communication technology make huge amount of heterogeneous
information available for us. But semantically integration of information and provide
machine understandable meaning to information is still a great challenge in current web technology.
In overcoming the challenges, ontology matching which is introduced by semantic web
technology plays a vital role. In this thesis, we propose a new method of ontology matching
using parallelization and distribution technique. To apply parallelism, we develop a partitioning
algorithm by using property-by-class and subclass of relationship, which partitions the ontology
into smaller clusters. Then the clusters from different ontologies are matched based on
terminological and structural similarity with semantic verification. All these tasks of matching
are handled in a parallel way and all the tasks are distributed over the computational resources.
Thus, we significantly reduce the time complexity and space complexity of large scale matching
task. Our proposed method reduces misaligned pairs while increasing correct aligned concepts.
Validity of our claims have been substantiated through different experiments on small and large
ontologies.
