Upload
killian-levacher
View
483
Download
1
Tags:
Embed Size (px)
Citation preview
Towards a Framework for
Open-Corpus Content Preparation
supporting
Adaptive Hypermedia Systems
Killian Levacher
Outline
Increasing importance of adaptive systems on the Web
AHS impediments to full mainstream adoption
Novel content preparation framework solution
Framework benefits
Novel challenges introduced
Roadmap ahead
Manually Authored by Small groups of Users
Lack of Diversity and Up to Date Content
Pre-existing Documents
Authored in particular Formats
Content Availability Impedes AHS Full
Mainstream Adoption
Mainly due to low availability of suitable content in terms of volume, style, diversity, meta-data, granularity…
Wealth of Information on the Web
Content not directly re-usable by AHS
• Usually built for single purpose usages
• Limited amount of meta-data
• Very heterogeneous (Style…)
• Different languages
• Very coarse grained
• Contains noisy information
Automated content preparation service
Wide variety of up to date content
Re-purposing of existing content
No generic structure to comply with
Open Corpus Content Preparation
Avail AHS with the wealth of open-corpus information
Bridge the gap between open-corpus content & AHS specific information requirements
Fully decouple content from core adaptive system
Service that prepares open-corpus content for AHS usage
Content Analysis Services
A Priori
A Slice• is a semantically independent piece of content extracted
from a pre-existing document• Is retrieved in a chosen format• represents a AH subjective perspective of a document
Benefits of this Framework
Open Corpus processing re-purposing content
Automated content Slicing vs Manual Authorship
• Possible solution to content authorship scalability
Removal of content format dependency for AH
• Content preparation approach solves Interoperability issues
Pipelined approach enables new content annotators to be plugged in seamlessly
Concept of subjective slices of existing content
New Challenges Introduced
Structural Segmenter fulfilling large domain and
processing speed requirements
Semantic Annotator will be a critical component
• How much semantic meta-data can we really aim for? Will it necessarily be domain dependant?
• Has a direct influence on Slice Precision
Roadmap Ahead
Selection or composition of a framework specific structural segmenter
Initial comparative evaluation of semantic analyzer to evaluate the quality and volume of meta-data expected
Implementation of framework within a PersonalizedMulti-Lingual Customer Care System
Summary
Content provision impedes the full mainstream adoption of AHS
Content provision should be fully loosely coupled with adaptive systems
Novel open-corpus preparation framework
Solution provides content scalability, interoperability, volume, diversity
New challenges ahead
Initial proof of concept planned within PMCC system