Pain and Research

Today, I got up from bed very late. I had a terrible teeth pain at the night. After taking my breakfast , I was writing something ontology matching techniques for my thesis. I want to share with it who will read my blog. Hope that they will pray for me for my thesis presentation.

Ontology matching techniques can be classified in the following :

• Element Level Matching
  
• Corpus-based Matching
  
• Knowledge base Matching
• Semantic Matching.

In Element Level Matching System, we start the process by comparing two strings. To compare strings, there are several methods are exiting.

1. Prefix,
2. suffix
3. Edit distance
4. N-gram

More interested readers , I can suggest that you should read the book of "Ontology Matching by Pavel"

In Corpus-based Matching, a large number of corpus are included. tokens are most important for this kind of matching . we can find matching using:

• LSI(Latent Semantic Indexing)
• Cluster Code Difference
• Formal Concept Analysis
• Common Instance Comparision

In knowledge-base matching, external resources are included. i.e. WordNet, Thesauri , Taxonomies, etc.

In semantic matching , match acts as operator and takes two graph and produces mapping but it depends on knowledge techniques as well . I think semantic matching techniques can not be accomplish without help of knowledge base. However, I can suggest to read a paper for semantic matching

"Semantic Matching" By Pavel Shvaiko and Fausto Giunchiglia

I think that matching is one of the hardest task. You can not achieve 100 % matching results by automatic matcher. I am not pessimistic person , I am optimistic . I am sure that we can overcome all this problem. It will be major break through for heterogeneity problem of data integration.