Sistem Penelusuran Katalog Perpustakaan Menggunakan Metode Rocchio Relevance Feedback

摘要

In information retrieval, the number of relevant documents that are refound will be affected by the number of keywords used for the searching. In the Web, the search is generally done using a short query, just less than three words (Spink et al. 2000). It is quite difficult because of the short queries provide little information retrieval system for the needs of the user\u27s searching. As a result, although the number of documents that are refound many but only a few which are considered relevant by the user. Rocchiorelevancefeedbackmethodis thequerymodificationmethodthat maximizesthe differencebetweenthe average ofthe relevant documentsto theirrelevant. Rocchiorelevancefeedbackmethodconsistsoffour stages:tokenizing, filtering, stemming, and termweighting. In thetokenizingphase, a termthat the user enteredseparated intosingle words. In thefilteringstage, importanttermswill be takenandwhichare notdescriptivetermswill be discarded. At this stage ofstemming, wordchanged torootby removingparticle word. At this stage oftermweighting, term weightwillbe calculatedtodetermine the availability ofthat term.