ieee knowledge media networking kmn’02 keynote address, crl, kyoto japan, july 11, 2002 concept...
TRANSCRIPT
IEEE Knowledge Media Networking KMN’02IEEE Knowledge Media Networking KMN’02Keynote Address, CRL, Kyoto Japan, July 11, 2002Keynote Address, CRL, Kyoto Japan, July 11, 2002
Concept Switching in the Interspace:Concept Switching in the Interspace:Networking Infrastructure for Networking Infrastructure for
Community KnowledgeCommunity Knowledge
Bruce SchatzCANIS Laboratory
Graduate School of Library and Information ScienceUniversity of Illinois at Urbana-Champaign
Graduate School of Informatics, Kyoto University
[email protected], www.canis.uiuc.edu
from Objects to Concepts from Syntax to Semantics Infrastructure is Interaction with Abstraction
Internet is packet transmission across computers
Interspace is concept navigation across repositories
CONCEPT SPACESCONCEPT SPACES
Technology
Engineering
Electrical
FORMAL
INFORMAL
(manual)
(automatic)
IEEE
communities
groups
individuals
LEVELS OF INDEXESLEVELS OF INDEXES
THE DISTRIBUTED WORLDTHE DISTRIBUTED WORLD
Community Repositories in the Interspace Peer to Peer Networking Infrastructure Every Person performs Every Role
USER requestLIBRARIAN referenceINDEXER classifyPUBLISHER qualityAUTHOR generate
Meta DataMeta Data
How to Represent theCommunity Knowledge
Automatic and InteractiveRepresentation Techniques
for Capturing the Fundamental Structure
Meta MapsMeta Maps
How to Locate theCommunity Knowledge
Automatic and InteractiveLocation Techniques
for Capturing the Fundamental Landscape
SCALABLE SEMANTICSSCALABLE SEMANTICS
Automatic indexing Domain-Independent indexing Statistical clustering
Compute Context of
concepts within documents documents within repositories
1992 1993 1995 1996 1998
COMPUTING CONCEPTSCOMPUTING CONCEPTS
‘92: 4,000 (molecular biology)
‘93: 40,000 (molecular biology)
‘95: 400,000 (electrical engineering)
‘96: 4,000,000 (engineering)
‘98: 40,000,000 (medicine)
SIMULATING A NEW WORLDSIMULATING A NEW WORLD Obtain discipline-scale collection
MEDLINE from NLM, 10M bibliographic abstracts human classification: Medical Subject Headings
Partition discipline into Community Repositories 4 core terms per abstract for MeSH classification 32K nodes with core terms (classification tree)
Community is all abstracts classified by core term 40M abstracts containing 280M concepts concept spaces took 2 days on NCSA Origin 2000
Simulating World of Medical Communities 10K repositories with > 1K abstracts (1K w/ > 10K)
Semantic IndexingSemantic Indexing Extracting Concepts (AI)
Canonical noun phrases Generic statistical parser
Computing Context (IR) Co-occurrence frequency, in collection Useful interactively, not strict ordering
System Side InfrastructureSystem Side Infrastructure
Classification Technologies for Multimedia Documents
Phrases (multi-word nouns) Concepts (generic phrases) Types (identified concepts) Clusters (grouped types) Structures (semantic
universals)
INTERSPACE NAVIGATIONINTERSPACE NAVIGATION
Semantic Indexes for Community Repositories
Navigating Abstractions within Repository concept space & category map
Interactive browsing by Community experts
*www.canis.uiuc.edu/interspace-prototype
Navigation in MEDSPACENavigation in MEDSPACE
For a patient with Rheumatoid Arthritis Find a drug that reduces the pain (analgesic) but does not cause stomach (gastrointestinal) bleeding
Choose DomainChoose Domain
User Side InfrastructureUser Side Infrastructure
Navigation Technologies forSearch Interfaces
Exact Match (noun phrases) Relationship List (concept suggestions) Cluster Comparison (groups to groups) Spreading Activation (group intersections) Artificial Landscapes (semantic distances)
SWITCHINGSWITCHING
In the Interspace…
each Community maintains its own repository
Switching is navigating Across repositories
use your vocabulary to search another specialty
CONCEPT SWITCHINGCONCEPT SWITCHING
“Concept” versus “Term” set of “semantically” equivalent terms
Concept switching region to region (set to set) match
term
Semantic region
Concept SpaceConcept Space
Future TechnologiesFuture Technologies Concept Switching
Spreading activation, type tagging
Dynamic Indexing On-the-fly collections, during session
Path Matching Aggregating indexes, many repositories
Semantic Analysis of MultimediaSemantic Analysis of Multimedia
Collections of Objects containing Units Text: community repository (topic proximity)
document abstracts containing noun phrases Image: aerial photograph (spatial proximity)
feature regions containing texture tiles
Units -- media-dependent (statistical parsers) Indexes -- media-independent (statistical clusters)
Media Interoperability ModelMedia Interoperability Model
text concept space & category map (geoscience) 1M phrases in 500K abstracts from Georef
and Petroleum Abstracts image concept & category maps in aerial photos
visual thesaurus maps for 200K regions in 800 images (6M tiles)
geographic map (where) v. semantic map (what) spatial gazetteer as bridge
image<=>text<=>number
Text and Number InteroperabilityText and Number Interoperability
Text and AVHRR Query:Show me information about Santa Barbara area with mild temperature and high vegetation density.
Integrated Result:Within the bounding geography location, 2 documents and 88 AVHRR records related to the integrated query are retrieved.
Image Concept SwitchingImage Concept Switching
Image Query:By browsing a texture (tile) catalog, show me information about residential and farm land areas.
Result:A set of related images are retrieved and shown in the Results Frame. The full-size image #368 is displayed with its place names and tile locations.
INFORMATION SPACEFLIGHTINFORMATION SPACEFLIGHT
Landscape as category map visualization
Valleys are semantic clusters Hills are semantic distances
Traversal across multiple levels of abstraction
INFORMATION SPACEFLIGHTINFORMATION SPACEFLIGHT
Flying through Flying through CyberspaceCyberspace
Flying through Flying through CyberspaceCyberspace
THE NET OF THE 21st CENTURYTHE NET OF THE 21st CENTURY
Beyond Objects to Concepts Beyond Search to Analysis Problem Solving via Cross-Correlating
Multimedia Information across the Net
Every community has its own special library Every community does semantic indexing
The Interspace is truetrue Cyberspace
Subject AssignmentSubject Assignment
Improved Search byIdentifying Subjects
Human Indexers classify DocumentsFrom Subject Thesaurus and Knowledge
Interactive Support for Community Curators(Subject Experts but Classification Amateurs)
Use Concept Spaces to Suggest SubjectsFrom Related Documents in the Collection
See Best Paper Nominee at ACM DL 98
Structure AssignmentStructure Assignment
Improved Search byIdentifying Structures
Human Indexers classify ClustersFrom Generic Structures beyond Subjects
Universal Structures Cross-Cultural
Interactive Support for Community Curators(Subject Experts but Classification Amateurs)
Necessary for Peer-Peer InfrastructureWhen Ordinary Persons form Communities
The Structures of Everyday LifeThe Structures of Everyday Life Bodies (individuals)
Food and Clothes Buildings (groups)
Houses and Cities
Transportation (physical interactions) Rails (trains) and Roads (cars)
Communication (logical interactions) Phones (talking) and Computers (retrieving)
Navigating Universal StructuresNavigating Universal Structures
A planet for every kid’s local environment Federating the planets into a universe Ordering all planets from kid’s Point Of View Flying through the Kids UniverseKids Universe Finding similar kids from different POVs Connecting historically through museums