dan su department of computer science brigham young university

Post on 31-Dec-2015

18 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Towards Semantic Web: An Attribute-Driven Algorithm to Identifying an Ontology Associated with a Given Web Page. Dan Su Department of Computer Science Brigham Young University. Motivation. Semantic Web Enrich current web manually will be laborious, tedious and error-prone Ontology - PowerPoint PPT Presentation

TRANSCRIPT

Towards Semantic Web: An Attribute-Towards Semantic Web: An Attribute-Driven Algorithm to Identifying an Driven Algorithm to Identifying an

Ontology Associated with a Given Web Ontology Associated with a Given Web PagePage

Dan Su

Department of Computer Science

Brigham Young University

MotivationMotivation

Semantic WebEnrich current web manually will be laborio

us, tedious and error-proneOntology

Capture the semantics of information from various sources and output a concise description

The number of different ontologies is increasing

given a web page, how could you identify which ontology in the ontology library will be associated with it?

Ontology LibraryOntology Library

It is necessary to classify ontology in an ontology library system in order to facilitate searching, managing and re-using ontology

Text CategorizationText Categorization

Assign a Boolean value to each pair <dj, ci>

Є D * CTraditional approachesAdvantage of traditional approachesDisadvantage of traditional approaches

Why not a direct matching?Why not a direct matching?

Direct MatchingReasons for discarding a direct matching:

1. The increasing large number of ontologies

2. Ignore weights of different attributes

Thesis StatementThesis Statement

focus on identifying the ontology associated with the given web page from an ontology library based on the attributes similarity computation

discuss the feasibility of an ontology-based machine learning approach

Overview of MethodOverview of Method

Assumption of Ontology Assumption of Ontology LibraryLibraryUnique identification: a unique URI or a

unique nameLife-time: the valid period for current

ontology versionDomain concept and the associated

attributes

Automatic Construction of Automatic Construction of Training corpusTraining corpus <html> <p> Honda MH-2 1994</p> ……… </html> ----------------------------------- <html> <ontology car.ontology> </ontology> <p><attri make> Honda </attri> <attri model>MH-2</

attri><attri year> 1994</attri> </html>

Attributes Similarity Attributes Similarity ComputationComputation Attribute Recognizer

<html>

<p> Honda price $9000, model…</p> ……… </html>

make

TFIDF-like computationTFIDF-like computation

Oj= (wj1, wj2, ………wjn) wk=afjk * idfjk

EvaluationEvaluation

Precision and RecallCompare our results with the results using

Pure TFIDF algorithm.

ContributionContribution

Automatically identify the ontology associated with a give web page

Advance the transformation from current web to a semantic web

Can be extended to the text categorization field

top related