Cargando…

Implicit User Interest Profile

User interest profile presents items that the users are interested in. Typically those items can be listed or grouped. Listing is good but it does not possess interests at different abstraction levels - the higher-level interests are more general, while the lower-level ones are more specific. Furthe...

Descripción completa

Detalles Bibliográficos
Autores principales: Chan, K, Kim, R
Lenguaje:eng
Publicado: 2002
Materias:
Acceso en línea:http://cds.cern.ch/record/579221
_version_ 1780899428830281728
author Chan, K
Kim, R
author_facet Chan, K
Kim, R
author_sort Chan, K
collection CERN
description User interest profile presents items that the users are interested in. Typically those items can be listed or grouped. Listing is good but it does not possess interests at different abstraction levels - the higher-level interests are more general, while the lower-level ones are more specific. Furthermore, more general interests, in some sense, correspond to longer-term interests, while more specific interests correspond to shorter-term interests. This hierarchical user interest profile has obvious advantages: specifying user's specific interests and general interests and representing their relationships. Current user interest profile structures mostly do not use implicit method, nor use an appropriate clustering algorithm especially for conceptually hierarchical structures. This research studies building a hierarchical user interest profile (HUIP) and the hierarchical divisive algorithm (HDC). Several users visit hundreds of web pages and each page is recorded in each users profile. These web pages are used to calculate HUIP for each user. Using the web pages from the users, the relations between words are calculated and clusters of words are made hierarchically. We use AEMI function to calculate the similarity of words, which means how much the words are related, and several techniques to calculate threshold to divide clusters to build more accurate and detail profiles.
id cern-579221
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2002
record_format invenio
spelling cern-5792212019-09-30T06:29:59Zhttp://cds.cern.ch/record/579221engChan, KKim, RImplicit User Interest ProfileComputing and ComputersUser interest profile presents items that the users are interested in. Typically those items can be listed or grouped. Listing is good but it does not possess interests at different abstraction levels - the higher-level interests are more general, while the lower-level ones are more specific. Furthermore, more general interests, in some sense, correspond to longer-term interests, while more specific interests correspond to shorter-term interests. This hierarchical user interest profile has obvious advantages: specifying user's specific interests and general interests and representing their relationships. Current user interest profile structures mostly do not use implicit method, nor use an appropriate clustering algorithm especially for conceptually hierarchical structures. This research studies building a hierarchical user interest profile (HUIP) and the hierarchical divisive algorithm (HDC). Several users visit hundreds of web pages and each page is recorded in each users profile. These web pages are used to calculate HUIP for each user. Using the web pages from the users, the relations between words are calculated and clusters of words are made hierarchically. We use AEMI function to calculate the similarity of words, which means how much the words are related, and several techniques to calculate threshold to divide clusters to build more accurate and detail profiles.EXT-2002-065oai:cds.cern.ch:5792212002-09-04
spellingShingle Computing and Computers
Chan, K
Kim, R
Implicit User Interest Profile
title Implicit User Interest Profile
title_full Implicit User Interest Profile
title_fullStr Implicit User Interest Profile
title_full_unstemmed Implicit User Interest Profile
title_short Implicit User Interest Profile
title_sort implicit user interest profile
topic Computing and Computers
url http://cds.cern.ch/record/579221
work_keys_str_mv AT chank implicituserinterestprofile
AT kimr implicituserinterestprofile