Structured Dirichlet Smoothing Model for Digital Resource Objects
Wafa’ Za’ alAlma’ aitah1, Abdullah Zawawi Talib2, Mohd Azam Osman3
1Wafa’ Za’ alAlma’aitah*, Department of Computer Sciences, University Sains Malaysia, Penang, Malaysia.
2Abdullah Zawawi Talib, School of Computer Sciences, University Sains Malaysia, Penang, Malaysia.
3Mohd Azam Osman, School of Computer Sciences, University Sains Malaysia, Penang, Malaysia.
Manuscript received on September 07, 2019. | Revised Manuscript received on September 22, 2019. | Manuscript published on October 30, 2019. | PP: 3427-3430 | Volume-9 Issue-1, October 2019 | Retrieval Number: A2647109119/2019©BEIESP | DOI: 10.35940/ijeat.A2647.109119
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Typically, the results of digital resource object retrieval are made up of whole documents. The problem is that each document contains a large number of metadata units. These units are located in a single document describing different topics. Therefore, retrieving the entire document means retrieving all these units which may be mostly irrelevant to the user‘s query. The Dirichlet smoothing model usually calculates the probability of having the query in each document and then displays the related documents basedon the query. It is therefore best to retrieve the nearest and relevant metadata units for the query regardless of which document they belong to. To achieve this, astructured Dirichlet smoothing model is proposed in this paper thatcalculates the likelihood between the query and the metadata units instead of between the query and the whole document. The experiments which were conducted on the cultural heritage CHiC2013 collectionhave shown a statistically significant improvement over the traditional Dirichlet smoothing model.
Keywords: Digital resource objects, Dirichlet smoothing model, Information retrieval.