Open Issues on Semantic Web Daniel W. Gillman US Bureau of Labor Statistics.
Download
Report
Transcript Open Issues on Semantic Web Daniel W. Gillman US Bureau of Labor Statistics.
Open Issues on Semantic Web
Daniel W. Gillman
US Bureau of Labor Statistics
The BLS Mission
The Bureau of Labor Statistics (BLS) is the
principal fact-finding agency for the Federal
Government in the broad field of labor
economics and statistics. The BLS collects,
processes, analyzes, and disseminates
essential statistical data to the public,
Congress, Federal agencies, State and local
governments, business, and labor.
Outline
Semantic Web – Description
Scenario
Problems
Semantic Web Technologies
Semantic Web and Metadata Management
Analysis
Identify problems / use scenario
Discovery, Judgment, Meaning
Not Semantic Web criticism / Stimulus for debate
METIS
2010-03-12
4
Semantic Web - Description
Berners-Lee -- 1999
I have a dream for the Web [in which
computers] become capable of analyzing
all the data on the Web – the content,
links, and transactions between people and
computers. A ‘Semantic Web’, which should
make this possible, has yet to emerge, but
when it does, the day-to-day mechanisms
of trade, bureaucracy and our daily lives
will be handled by machines talking to
machines. The ‘intelligent agents’ people
have touted for ages will finally materialize.
METIS
2010-03-12
5
Semantic Web - Description
Web pages, readable
B y computer
Instead, now, humans
Determine height of Mt Everest
Reserve table at favorite restaurant
Find best prices for tires for the car
Semantic Web will demand more
METIS
2010-03-12
6
Semantic Web - Description
Two new IT artifacts
Web Services
Ontologies
Service
Set of events with a defined interface
Web Service
Software designed to support
interoperable machine-to-machine
interaction over a network
METIS
2010-03-12
7
Semantic Web - Description
Ontology
Set of concepts, the relations among
them, and a computational description
Purpose is to be able to reason, i.e.,
make inferences
Knowledge representation languages
Bridge between web service and
ontology
METIS
2010-03-12
8
Scenario
“America’s Safest Cities”
by Zack O’Malley Greenburg
26 October 2009
Forbes Magazine
Rank cities by “livability”
Workplace fatalities
Traffic fatalities
Violent crimes
Natural disaster risk
METIS
2010-03-12
9
Scenario
Base comparison on MSA
Metropolitan statistical area
Rank MSAs based on
Numerical ranking for each measure
Sum of rankings
Questions
Can we find such data?
If so, where?
METIS
2010-03-12
10
Scenario
Finding data -- Discovery
Workplace fatalities
– Bureau of Labor Statistics
– Data based on MSA
– Data given as number, not rate
Traffic fatalities
– National Highway Traffic Safety
Administration
– Data based on city, not MSA
– Based on rates
METIS
2010-03-12
11
Scenario
Violent crime
– Federal Bureau of Investigation
– Based on MSA
– Given as rate
Natural disaster risk
– SustainLane.Com
– Not federal site, based on government data
– Data based on city, but only a few
– No data, no rates, just a rank
METIS
2010-03-12
12
Scenario
Using data – Judgment
Unit of analysis = MSA
Questions
How can we combine this data?
Can we harmonize the differences?
City as proxy for MSA?
Decisions are
Qualitative
Require human judgment
METIS
2010-03-12
13
Scenario
How do we know
MSA vs. city
Number vs. rate
Rank vs. rate?
Understanding – Meaning
Requires
Links from data sets to metadata
Good metadata model for data semantics
METIS is good at this
METIS
2010-03-12
14
Problems
Meaning
Easy – needs agency metadata
Link meanings to data
– Straightforward
– Mechanical, once metadata is captured
Discovery
Harder –
–
–
–
–
Difficult search
Takes a lot of work
Numerous comparisons
Not easy to know when to stop
METIS
2010-03-12
15
Problems
Judgment
Very hard –
– Difficult to see how to automate
– Case by case basis
If proxy OK?
Need population for MSA
Again, where?
– Discovery (Census Bureau)
– Judgment (Appropriate?)
– Meaning (Data elements correct?)
METIS
2010-03-12
16
Semantic Web Technologies
Web services
Any action in Semantic Web
Several kinds
Operation required? Web service called
Examples based on scenario
Read data from a data set
Display data dictionary of data set
Calculate rates, ranks, and overall rank
METIS
2010-03-12
17
Semantic Web Technologies
Ontologies
Concept systems
– Set of concepts
– Relations among them
Computational description
– How one makes inferences
– Logical system
Means for organizing knowledge
– Concepts organized for some purpose
METIS
2010-03-12
18
Semantic Web Technologies
Ontologies
Logics
– Predicate calculus
– Description logic
– First order logic
– Others
Low to high formality
METIS
2010-03-12
19
Semantic Web Technologies
Knowledge representation languages
Bridge between ontology and web service
Service uses KRL to make inferences
Typical languages
RDF – Resource Description Framework
– Based on “triples”
• Subject – verb – object
– Triples can be linked
• Object of one is subject of another
– Creates Directed Graph structure
METIS
2010-03-12
20
Semantic Web Technologies
Typical languages – cont’d
OWL – Web Ontology Language
– Comes in 3 main types
• OWL – lite
•
•
» More powerful than RDF, easiest, a DL
OWL – DL
» More powerful than OWL – lite, a DL also
OWL – full
» Equivalent to RDF-Schema, almost FOL
» Most powerful OWL, hard to implement
METIS
2010-03-12
21
Semantic Web Technologies
Typical languages – cont’d
RDF and OWL – W3C specifications
Common Logic – ISO/IEC 24707
– Very powerful
– Full FOL, including some extensions
However – Using KR ≠> Ontology
KR languages – Difficult to implement
– Work to build non-trivial ontology is huge
•
•
•
Subject matter experts
Terminology experts
KR and logic experts
METIS
2010-03-12
22
Semantic Web and
Metadata Management
Metadata play central role in SW
Linked Data – newer aspect of SW
Berners-Lee given credit again
Laid out 4 criteria
– Use URIs to identify things.
– Use HTTP URIs for dereferencing
– Provide useful metadata when URI
dereferenced.
– Include links to other, related URIs
METIS
2010-03-12
23
Semantic Web and
Metadata Management
2 main reactions:
1) No difference with traditional metadata
management
2) Begs the question
– How does one FIND the right URI (URL)?
Answer – Ontologies! – See above!
Successful ontology
Consistent
Complete
Useful
METIS
2010-03-12
24
Semantic Web and
Metadata Management
Consistent & Compete ≠> Useful
Discovery doesn’t need new methods
Registries are designed for this
SDMX
ISO/IEC 11179
Library card catalog
METIS
2010-03-12
25
Semantic Web and
Metadata Management
Judgment
SW offers no help
Meaning
Metadata management already solves
METIS members are experts
METIS
2010-03-12
26
Conclusion
Verdict
SW not offering much new
SW descriptions
Make hard problems seem easy
Make easy problems seem hard
– Often the “sexy” stuff
METIS
2010-03-12
27
Contact Information
Daniel Gillman
[email protected]