PENGELOMPOKAN SURAT ELEKTRONIK SECARA SEMANTIK, HIRARKI DAN LANGSUNG PADA BASIS DATA ARSIP EMAIL
| Main Authors: | , M. Zudha Ghofur, , Drs. Edi Winarko, M.Sc.,Ph.D |
|---|---|
| Format: | Thesis NonPeerReviewed |
| Terbitan: |
[Yogyakarta] : Universitas Gadjah Mada
, 2014
|
| Subjects: | |
| Online Access: |
https://repository.ugm.ac.id/133139/ http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=73694 |
Daftar Isi:
- To manage the stored email, it currently takes a manually created rules to filter/ categorize as desired. Meanwhile, to rediscover the email that has been saved, it can be used search feature. Search results in the email client is currently just a flat list of email. To facilitate users easily and quickly search for relevant email, needed a way to classify emails into a relevant label/category. Email clustering not only by words but also in the form of semantic/phrases, and can also be presented in a hierarchical and online clustering. The research was conducted start from data collection of electronic mail are stored as email database archive, then do preprocessing includes stopping and stemming. The next stage is feature extraction, here used key phrases which are extracted from a collection of email as a feature of the email. Having obtained the email feature, then do the clustering by using orthogonal clustering, which used Singular Value Decomposition. At this stage it will be analyzed, whether clustering with SVD produces accurate grouping and persistent or not. The next stage is to determine the label and hierarchy, the label specified key words/phrases that have the greatest value in the group vector, and the hierarchical done by comparing each pair of groups, to see whether the merger can be treated as a group or a parent-child relationship. The last stage of this research will be testing each stage. Seeing the results of stage-by-stage testing research, it can be seen that the orthogonal clustering system on the email archive database capable of clustering semantically electronic mail, and also can be made of a hierarchy that allows users to browse the relevant email.