LANGUAGE MODEL FOR DIGITAL RECOURSE OBJECTS RETRIEVAL

Language model has been successfully applied for use in information retrieval to retrieve structure and unstructured information. Typically, language model involves three basic models namely: N-gram language models, smoothing model and estimation model. Language model has been approved outperforms of other retrieval model such as vector space model and probabilistic model. The problem arises when language model uses to retrieve digital Resource Objects which use metadata to describe their content. Digital Resource Objects have special three characteristics: lack in metadata content (short document), short query, and heterogeneity metadata content. This paper presents a performance comparison among information retrieval models (Vector Space Model and Probabilistic Model) using a Digital Resource Objects (CHiC2013 collection). Further, an overview for language model approaches to determine which models are suitable for digital Resource Objects, despite being a traditional review, a comprehensive comparative analysis is conducted among different approaches of Language model