Personalized Digital Library using Dynamic User Profile and CYC Ontology

1
Personalized Digital Library using Dynamic User
Profile and CYC Ontology

• Khulthida Chumsaktrakul

2
Agenda

• Related Topics
• Project Overview
• Architecture
• Contributions and Implementation
• Future Work

3
Related Topics

• Digital library
• Personalization and Recommendation Issues
• Ontology and Knowledge Base

4
What is digital library?

• Tool for changing the way information is used
• Collection of digital objects
• Build upon underlying streams of content

5
Greenstone Digital Library

• Software for constructing and presenting
  collections of documents

6
Greenstone Digital Library

• Provide collections of text, multimedia
• Support multi language

• Easy to use and maintain
• Allow flexible searching and browsing
• Provide dynamic mechanisms

7
How Greenstone works?
8
(No Transcript)
9
Personalization Recommendation

• Gather user-information during interaction with
  the user. Systems take input from users and will
  then make personalised recommendations of
  products or services.
• The aim is to improve the users experience of a
  service.
• One type of personalisation that is growing in
  use is recommender systems.

10
Personalization Recommendation (cont.)

• Content-based personalization - recommends new
  items that correlate with a users profile.
• Collaborative personalization - recommends new
  items based on a correlation of other users
  ratings.
• The hybrid approach retains the advantages of
  both content-based and collaborative approach

11
Personalization Recommendation (cont.)

• Removing titles from this list by giving ratings
  helps to generate new recommendaions

12
Personalization Recommendation (cont.)

• Reduce the amount of explicit user input and to
  operate based on usage patterns

13
Personalization Recommendation (cont.)

• One difficult problem faced by recommender
  systems is the cold-start problem where
  recommendations are required for new users whom
  little or no information has yet been acquired
  for
• In digital libraries, users have their profiles.
  We can recommend since users will eventually
  search and browse items, sometimes put some items
  in their lists like Bookshelf

14
Metadata

• data about data used to provide documentation to
  characterize data
• answers who, what, when, where, why, and how
  about every facet of the data that are being
  documented

15
Ontology

• The Classification Taxonomy of a digital library
  is a simple Ontology 
• Ontology is an conceptualization of a domain into
  a human-understandable but machine-readable format

16
CYC Ontology

• the world's largest and most complete general
  knowledge base and commonsense reasoning engine
• 100,000 basic concepts
• 1,000,000 assertions in microtheories
• Run on both Windows and Linux

17
Microtheories Inheritance
UniversalVocabularyMt
genlMt
HumanActivitiesMt
genlMt
UnitedStatesSocialLifeMt
genlMt
genlMt
MainstreamAmericanCultureMt
WorldMythologyMt
18
Ontologies knowledge bases

• Difficult and costly to create
• ResearchCYC , Commercial CYC, OpenCYC
• There are many easy-to-use tools to create an
  Ontology in a particular domain
• However, use CYC as
• a preliminary experiment

19
CYC

• The content of the Knowledge Base comprises
• A vast taxonomy of concepts and relations
• A rich formal representation of their
  interconnections
• The main advantage of Cyc for representing
  knowledge is its use of a formal language in
  which inferential connections between concepts
  and statements are encoded in a machine
  accessible way.

20
Example
21
Example (cont.)

• Sewing Needle usually leads CYC to Textile domain

22
Example (cont.)

• If user X and user Y have many overlapping
  documents in their virtual bookshelf
• We can assume that something interests X may be
  of interest to Y under certain constraint.
• If-then Rules

23
(No Transcript)
24
Project Overview

• Deliver information where and when it is needed
• Users frustrate with the increased information
• Users expect more sophisticated services
• Recommends their interested books
• Book reserved for them while other users can
  still read

25
Components

• Web interface
• HTML
• PHP
• MySQL
• Greenstone
• CYC Ontology
• (run on Apache)

Greenstone
Users Profile
CYC
Apache
WEB Interfaces
26
Architecture
Users Profile
WEB Interfaces
keywords
CYC
27
Architecture
Users Profile
WEB Interfaces
keywords
CYC
28
Contributions and Implementation
29
(No Transcript)
30
(No Transcript)
31
CYC DL
32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
Other tools Bookshelf
37
Retrieve documents
38
Other tools Profile
39
Other tools Add to BookShelf
40
(No Transcript)
41
Future Work

• Documents added to bookshelf automatically
• Just one click, dont need to fill in anything
• Recommend items by collaborative filtering
• Keep track of searching behavior
• Get usage.txt in Greenstone
• Use occupation to infer in CYC
• Connect digital library to CYC by using API
• Apply for ResearchCyc License

42
Questions and Answers?