Possible DAML Use
CIA World Fact
BookProvide a DAML representation of the CIA
World Fact Book.
- Temporal data (many values are tagged by year).
- Uncertainty (some values are tagged by "est.").
- The ability to add semantic structure to a widely referenced data
source. The basic data structure is fairly simple, but we may be able to
add value by extending the ontology to allow reasoning based on
geographic region, industry categories, etc.
- The CIA World Fact Book itself is in the public domain. No
Intellectual Property issues.
resumes of a company's employees and job applicants in DAML.
- Temporal ranges for employee-of relations. Using DAML for historical
- Synonymy, as company names change over time.
- What existing employees may have worked with an interview candidate
in the past? (based on matching company/dates and optionally city)
- What employees have worked with both knowledge representation and
JournalImagine if the Wall Street Journal
were published every day using DAML as well as text.
- The ability to mix and relate structured and unstructured content.
Ideally, the DAML facts would contain references to the underlying text
from the article.
- Reification (DAML assertions are "as reported by the Wall Street
- Temporal data (articles are dated).
- Various corpora (including the Wall Street Journal) are available
from the Linguistic Data Consortium at
- Notify me with normal priority whenever a story appears about my
company is mentioned. Notify me with high priority whenever such a story
portrays my company in a bad light.
- Notify me whenever one of my competitors is mentioned (e.g. based on
Standard Industrial Classification code).
- We could potentially use another newspaper source, but the narrower
domain focus and recognition of the WSJ are attractive.
- This was demonstrated by BBN, using New York Times articles from the
MUC-7 conference, at the DAML Kickoff Meeting in August 2000.
Center for Army Lessons
LearnedUse DAML to encode the information
stored by the Center for Army Lessons Learned.
- CALL officials have already expressed interest in DAML, and attended
the DAML Kickoff meeting.
WorthVarious personal financial information
is now available via the WWW (bank accounts, stock portfolios, mutual
funds, retirement funds, credit card balances, etc.). It would be
possible to write an agent that polled these sources daily to track one's
"InterNet Worth" (assets - liabilities based on
Internet-accessible data sources), but this would currently require a
considerable amount of code customized for each site. How much easier
would it be to develop such an agent if all of this information were
published using DAML?
- Increased development efficiency (reduction in source lines of code
required to build agents) afforded by DAML.
- Security and privacy issues.
- Encoding efficiency afforded by DAML. Agents don't read ads!
- Most major financial institutions now make account information
available via the Internet (in ad hoc, proprietary formats).
- What's my instantaneous net worth?
AuthorizationThe W3C DAML proposal is
available at http://XXX. ...
- Logical inference
DirectoryThe Mozilla Open Directory (dmoz)
effort is building a standard taxonomy of WWW sites similar to that used
by Yahoo, Netscape, etc. Many of the major portals have already adopted
- The ability to handle a very large, though simple, ontology.
is a new DARPA ITO program under Jean Scholtz that seeks to provide
cached versions of WWW pages that have been identified as particularly
relevant by staff at operational military commands. It would be desirable
to include DAML content as part of this material.
- Mike Sullivan and other BBN staff involved in a Web-in-a-Box
jumpstart effort are currently surveying operational sites to identify
high priority unclassified WWW sources.
- The Web-in-a-Box BAA00-41 was just cancelled.
SEC EDGARThe U.S.
Security and Exchange Commission's EDGAR database of filings for
publically-traded companies have long been available on the Internet. The
submission are semi-structured SGML with lots of large text blocks. The
submissions contain a wealth of timely information about financial
performance, corporate officers, subsidiary relationships, competition,
- Digital signing of DAML assertions (the EDGAR submissions are already
signed, although these signatures wouldn't match for transformed data).
- ontology translation. An EDGAR ontology is likely to be similar to,
but perhaps different from, a Wall Street Journal ontology. This could
present an opportunity for ontology translation (and/or sharing) in a
restricted yet compelling domain.
- The EDGAR data is in the public domain.
Mike Dean, in
Sun, 13 Aug 2000 07:21:47 -0700