DARPA Agent Markup Language November 2001 Enabling agent

DARPA Agent Markup Language November 2001 Enabling “agent” communication at a Web-wide scale. Murray Burke Information Exploitation Office

Military Systems are Moving to the Web Army Joint Global Command & Control System Air Force for Army Lessons Learnedn Global Command Support System n JOPES 2000 n Warrior Knowledge Network n Advanced Course Of Action ACTD n Web Mapping Testbed n CINC 21 n Joint Center for Lessons Learned n Center Navy n Expeditionary Sensor Grid n Task Force Web n Navy Lessons Learned System n Navy Warfare On-Line Library n Air Marines n Joint Battlespace Infosphere n Theater Battle Management Core System n Broad. Sword n Air Mobility Command GDSS Intelligence Community n Intelink – Ground Combat System n Joint Intelligence Virtual Architec n Distance Learning Network n Logistics Bases Inventory Visibility n Horus n Crypto Analysis Web n Items Applications On-Line n Master Header Information On-Line n NIMA Geospatial Engine DAML Review_Nov 2001 p 2

Web Limitations Average WWW searches examine only about 25% of potentially relevant sites World Wide Web Intelink-S Intelink Web searches return a lot of unwanted information Doubles in size every six months Problems grow worse as Web continues to grow Information on web is not suitable for software agents DAML Review_Nov 2001 p 3

The Evolution of the Web World Wide Web HTML XML Formatted for humans to read via web browsers l Formatted for machine readability but with very limited semantics l Semantic Web Machine readability with very rich semantics to support agents for: DAML n n Intelligence Analysis and Production Military Planning and Operations Software C 4 ISR Agents Sensor Fusion DAML is building the language and tools to realize the bold vision of the Semantic Web DAML Review_Nov 2001 p 4

Web Migration to New Technology 100% HTML “In 30 years e-commerce will have become second nature. Lifelike, intelligent virtual assistants will be performing most routine transactions and simple negotiations electronically on our behalf. More technological change will have taken place in that period than during the entire twentieth century, and the curve will continue to steepen exponentially into the foreseeable future. ” Ray Kurzweil DAML 50% XML “Fifty percent of the content on the Web will be in XML format by the end of 2003” ………. Gartner Group 0% 2000 2005 2010 DAML Review_Nov 2001 p 5

The Bold Vision – Distributed Agents on the Semantic Web Determine Force Mix Get Weather Info Provide Intelligence Update Coordinate Transportation Order Supplies Select Targets “Make preparations to capture Bagram air base. ” DAML Review_Nov 2001 p 6

Query Today WWW Hotbot What is Al Qaeda? The answer may be somewhere in this list of URLs DAML Review_Nov 2001 p 7

Semantic Query What is Al Qaeda? A terrorist organization Would you like additional information on? q Membership q Locations q Structure q Finances q Tactics q Other terrorist organizations DAML Review_Nov 2001 p 8

What’s Hard? Scale Scope and Lack of Consistency in Global C 2 I Networks n Diversity: Ontologies and Ontology Mapping n Markup Simplicity: l Automatically Derived Markup l Constrained Logic n Enabling Software Agents to Locate and Use Appropriate Services: l Services description language l Infrastructure: Yellow Pages, Brokers, Mediaters n Trust: Logical Proofs DAML Review_Nov 2001 p 9

Precedents 1995. . . . . 1998 Dozens of application design languages Booch Odell Rumbaugh Rational Rose Paradigm Plus 2000. . . . . 2003 Dozens of knowledge representation lang Loom KREME KEE RDF KIF OMG DARPA / W 3 C UM L DAM L One modeling language One Web KR language DAML Review_Nov 2001 p 10

DAML Foundation Language to describe information Tools for users and developers Transition activities DAML Semantic Web l l WWW NIPRNET INTELINK-S INTELINK DAML Review_Nov 2001 p 11

Developing the Language AI Community l Formal Logics l Knowledge Representation l Reasoning Local Environments DARPA WWW Environment Web Standards DAML US – EU DAML Joint Committee European OIL Community DAML + OIL W 3 C l HTML l XML l RDF Web. Ont WG DAML Review_Nov 2001 p 12

Example Ontology These ontologies accessed at remote locations DAML Review_Nov 2001 p 13

DAML Example (Markup) <? xml version='1. 0' encoding='ISO-8859 -1'? > <!DOCTYPE rdf: RDF [ <!ENTITY rdf 'http: //www. w 3. org/1999/02/22 -rdf-syntaxns#'> <!ENTITY countries-ont 'http: //www. daml. org/2001/09/countries/iso-3166 -ont#'> <!ENTITY cities-ont 'http: //www. daml. ri. cmu. edu/ont/City. daml#'>] <rdf: RDF xmlns: rdf ="&rdf; " xmlns: rdfs="http: //www. w 3. org/2000/01/rdf-schema#" xmlns: daml="http: //www. daml. org/2001/03/daml+oil#" <daml: Ontology rdf: about=""> <daml: version. Info>$Id: map-ont. daml, v 1. 2 2001/06/16 13: 44: 29 mdean Exp $</daml: version. Info> <rdfs: comment>Map Overlay Ontology</rdfs: comment> </daml: Ontology> <rdfs: Class rdf: ID="Terrorist. Event"> <rdfs: sub. Class. Of> <daml: Restriction> <daml: on. Property rdf: resource="#name"/> <daml: to. Class rdf: resource="http: //www. w 3. org/2000/10/XMLSchema#string"/> <daml: max. Cardinality>1</daml: max. Cardinality> </daml: Restriction> </rdfs: sub. Class. Of> <daml: Restriction> <daml: on. Property rdf: resource="#location"/> <daml: to. Class rdf: resource="#Location"/> <daml: max. Cardinality>1</daml: max. Cardinality> </daml: Restriction> </rdfs: sub. Class. Of> <daml: Restriction> <daml: on. Property rdf: resource="#date"/> <daml: to. Class rdf: resource="http: //www. w 3. org/2000/10/XMLSchema#date"/> <daml: max. Cardinality>1</daml: max. Cardinality> </daml: Restriction> </rdfs: sub. Class. Of> </rdfs: Class> rdfs: Class rdf: ID="Terrorist"> <rdfs: sub. Class. Of> <daml: Restriction> <daml: on. Property rdf: resource="#name"/> <daml: to. Class rdf: resource="http: //www. w 3. org/2000/10/XMLSchema#string"/>< /daml: Restriction> </rdfs: sub. Class. Of> <daml: Restriction> <daml: on. Property rdf: resource="#citizen. Of"/> <daml: to. Class rdf: resource="&countries-ont; Country"/> </daml: Restriction> </rdfs: sub. Class. Of> <daml: Restriction> <daml: on. Property rdf: resource="#resides. In"/> <daml: to. Class rdf: resource="&countries-ont; Country"/> </daml: Restriction> </rdfs: sub. Class. Of> </rdfs: Class> DAML Review_Nov 2001 p 14

Agent-Based Inferencing <daml: Class rdf: ID=“Bin Laden"> <rdfs: sub. Class. Of rdf: resource="#terrorist"/> </daml: Class> Implies n If x is Bin Laden, he must be a terrorist n If x is a terrorist, then he may or may not be Bin Laden n If x is not a terrorist, then he is not Bin Laden n If x is not Bin Laden, he may or may not be a terrorist DAML Review_Nov 2001 p 15

DAML Services Ontology provides Service pr s rt What the service does described. By Service Profile po es en p su ts Resources Service Grounding How to access it Service Model How it works DAML Review_Nov 2001 p 16

Developing the Tools Operational Prototype Tools APIs Edit Visualize/ View Report Generator Validation/ Analysis Determine Force Mix Browser Get Weather Info Provide. Intelligence Update Coordinate Transportation Inference Order Supplies Select Targets Parser Import Search/Query Transform Developer and User Tools Semantic Web DAML Review_Nov 2001 p 17

Language & Tools Transition W 3 C Web. Ont Standard DAML Prototype Tools Feedback Loop DAML Language Specification Commercial Tools Near-term C 2 I Applications Longer-term Military & Commercial Applications DAML Review_Nov 2001 p 18

DAML Language Strategy Ontology Logic Formal Semantic Underpinnings W 3 C Standardization Proof • Model-Based • Axiom-Based Trust DAML Review_Nov 2001 p 19

Transition Process DAML Focal Point for Military Early Adopters • 60 Attendees at First Conference • 100 Expected at 2 nd Conference on 13 -14 Nov Do. D Projects n n DAML Semantic Web for Military Users Conferences (Identify Requirements) n n n Intelink Management Office: Horus Navy: Expeditionary Sensor Grid Air Force: Foreign Clearance Guide Army: Center for Lessons Learned Army: Warrior Knowledge Network TRANSCOM: Global Transportation Network Air Force: Joint Battlespace Infosphere Navy: Task Force Web DAML Review_Nov 2001 p 20

Horus IMO paid half in FY 00, FY 01 and FY 02 ~ $1. 5 M each year DAML Review_Nov 2001 p 21

Intelink Transition n Intelink has 150 K users and expects 400 K users by end of 2002 l A controlled testbed, effectively a microcosm of the WWW l A quick military transition of emerging DAML technologies n Quotes from Intelink Management Office (IMO) Conference l “Terrific New Alliance With DARPA” l “IMO Has Linked to DARPA’s DAML Program u Tim Berners-Lee, Ralph Swick etc (The “A” team)” l “Large Opportunity For Web Technologies u Many Models On Internet u Many We Can Develop With DARPA” DAML Review_Nov 2001 p 22

Getting the Word Out DAML LAB @ DARPA TIC www. daml. org www. daml. net 18 DAML Research Teams PKI n n n Mail Lists Web Server Software Repository Ontology Library Hot. DAML Newsletters Scientific American Article Internet Lots of Interest From the Rest of the World (Almost Two Million Hits) DAML Review_Nov 2001 p 23

DAML in 2002 DAML Tools DAML Language DAML Research Team Releases in May and September Horus DAML Integrated Demonstration & Experiment Feedback ESG Other Metrics DAML Review_Nov 2001 p 24

DAML DEMONSTRATION AND EXPERIMENT CONCEPT OF OPERATIONS Joint Battlespace Infosphere (JBI) DAML Rapid Knowledge Formation (translation) Weather Web Site Air Force Lessons Learned System Semantic Web Planner DAML Air Force Lessons Learned Markup DAML CIA Fact Book Markup Actual DAML ontologies & content Agents Ontology Editing Tool(s) Military or Do. D User Web Browser Navy Expeditionary Sensor DAML Grid DAML Air Force Plan Markup Internet for demo But could be SIPRNET for real application Weather Ontology Creates / Accesses DAML ontologies & content DAML Foreign Clearance Guide Markup Unifying Ontology (GEOLOC) . NET. ASP Server & DAML Query & Inference Engine Acts as proxies Complex querying for senior military and inferencing users to do about plan information background tasks Provides support to military user via his/her Web Browser DAML AT 2000 Legacy Data Bases Transition Programs DAML Review_Nov 2001 p 25

Metrics Now 1 Node 1 Ontology 6 Months 3 Data Base 0 Agents 3 -4 Nodes 4 Ontologies 3 -5 Data Bases 12 Months 1 Agents 4 -5 Nodes 6 -8 Ontologies 5 -7 Data Bases 2002 3 -5 Agents Experime nt Vision Millions of Nodes n. Millions of Nodes Millions of Ontologies n. Millions of Ontologies Millions of DBs n. Millions of Data Bases of DBs Millions of. Agents of Agents n. Millions of Agents Active on the Semanic Web DAML Review_Nov 2001 p 26