I'm doing my project on Semantic Web mining using ontologies, but even after rigorous googling, I'm not able to comprehend what is the use of ontology in searching. We have to create ontology by ourselves then for what purpose on the web? Why are we accessing the Web? And how can we create an ontology related to each and every topic in this world covering each and every aspect?
Simple (keyword) search works based on string comparison. If you are doing anything other than some linguistic processing (stemming, lemmatization, etc.), thats already semantic search. The most common way to add semantics is to use synonyms. More sophisticated is the taxonomy based similarity, where the relationship is either broader or narrower concept. One can introduce other type of relationships, like has-type or located-at, etc. There are online databases (ontologies) with thousands of entities and relationships among them. The most known one is the DBPedia.
Now when you are indexing documents with a semantic search engine, using a list of entity labels (people names, places, organizations, etc.) the engine can recognize these in the document being indexed. Once the entity is associated with the document, any related entity can be associated as well (like entity Paris was found and France is also added as metadata of the document in the index). This way if you search for Paris, results will include the document above, but searching for France too. Please note, in this example France was not mentioned in the document, it is coming from the DBPedia semantic service.
Obviously this was an oversimplified example.
answered 12 Jan '13, 14:10