IV.
PROPOSED APPROACH
The proposed approach to support semantic search in digital library includes semantic query processing and document annotation
A. Semantic Query Processing
The overall query process of ontology based information retrieval is illustrated in Figure 1. In order to facilitate the semantic search in the domain of digital library, an ontology based information retrieval framework is proposed.
SELECT ?supervisor ?name WHERE {:Studentl :superviseBy ?supervisor. ?supervisor rdfs:label ?name}
The result for the query will return the following tuples which shows that studentl, who is Arifah Alhadi is supervised by Supervisor 1 who is Prof Madya Dr Shahrul Azman Mohd Noah and PM Dr Shahrul Azman where it refers to the same person:
Once the list of documents is formed, a semantic similarity value between the query and each document is computed by the system. The ranking algorithm of the system combines two factors which are the concept weight specified in the user's query and its relevance to a document. All concept names (or instance names) returned by the inference engine form a query vector. In VSM, a document vector Xj in the extended Term Document Matrix (TDM) calculated is ranked according to the similarity between it and the query vector q. Then a vector of index term weights is calculated. These weights,พ are calculated by most often used tf-idf scheme as shown in the equation (1) below where t represent time, f represent frequency and id represent id of a document:
tfiix idfi
freq is the number of occurrences of term i t in document j d , N IS number of documents in collection, and i ท is the document frequency for term i t in the whole document collection. The similarity which is presented as sim, measure between a document d and the query q is computed as shown in equation (2) below: