Dynamic Integration of Distributed Semantic Services:   
Infrastructure for Process Queries and Question Answering 
 
Paul Thompson 
45 Lyme Road, Suite 200 
Dartmouth College 
Hanover, New Hampshire 03755 
Paul.Thompson@dartmouth.edu 
 
 
 
 
1  The IR2P Prototype 
 
The DARPA IXO mission is to develop “systems 
for real-time sensing, exploitation, and decision making 
in a rich tactical environment”.  The mission includes 
the development of individual technologies for sensors, 
sensor exploitation and command/control as well as the 
technology of information integration. Our research 
focuses on how to integrate distributed services in a 
dynamic networked environment to support IXO 
applications. This dynamic networked environment 
should include the following capabilities (DARPA, 
2002): 
 
• Information users should have scalable dynamically 
changing subscription services to heterogeneous 
information services; 
• Information providers should have scalable 
publishing services for their dynamically changing 
information products; 
• Scalable intelligent middleware to dynamically 
broker, compose, and manage the intermediate 
services necessary to connect information users to 
the right information products at the right time. 
 
While some of these capabilities can be realized 
with existing technologies, several challenging 
problems, particularly in the areas of scalability, 
semantic interoperability and dynamic extensibility, 
may need 5 or 10 years basic research efforts to 
adequately address. The Information Integration 
Research Prototype (I2RP) we describe is a vehicle for 
exploring which new paradigms and frameworks are 
most promising for future investment while calibrating 
what existing technologies can do today (Jiang et al., 
2002). This demonstration illustrates the I2RP 
architecture and the underlying technology approaches.  
In six months, we implemented an extensible prototype 
system with basic capability as a proof-of-concept to 
show some fundamental new ideas for implementing 
next generation dynamic information integration 
systems.   
 
2  Current Status 
 
IR2P is an information integration research 
prototype built with commercial-off-shelf technologies. 
With the prototype system, the declaration-composition-
production process of semantic integration has been 
investigated, with specific emphasis on semantic 
interoperability issues (Sirin et al., 2002). A target-
tracking scenario was developed to test the system and 
all components in the I2RP were successfully integrated 
“on the fly” to support this mission. Meanwhile we 
proposed and analyzed some metrics to quantify the 
semantic integration process such as semantic depth, 
markup complexity and information fluidity. From 
different aspects, we built some models to analyze the 
relationship between semantic depth and markup 
complexity, and between semantic interoperability and 
information fluidity.  
 
3  Process Queries and Natural Language 
Queries 
 
The work on IR2P has evolved into a more general 
engine for tracking, fusing, and querying processes.  We 
are presently developing a generic process querying 
capability that can be used to retrieve information about 
objects as diverse as physical objects on a battlefield or 
worms propagating through the Internet.  Later we plan 
to build a natural language querying capability as 
proposed by Lulich and Thompson (2002). 
 
References 
 
DARPA.  2002.  http://dtsn.darpa.mil/ixo/ 
 
Guofei Jiang, George Cybenko, Wayne Chung, Paul 
Thompson, Glenn Nofsinger Annarita Giani, Yong 
Sheng, Diego Hernando and Han Li, Jim Hendler, 
Evren Sirin, Bijan Parsia , Jennifer Golbeck, Kenneth 
Whitebread and Martin Hoffman. 2002.  IXO 
                                                               Edmonton, May-June 2003
                                                            Demonstrations , pp. 27-28
                                                         Proceedings of HLT-NAACL 2003
Seedling Project Technical Report Dynamic 
Integration of Distributed Semantic Services Thayer 
School of Engineering, Dartmouth College. 
 
Steven Lulich and Paul Thompson.  2002.  Question 
Answering in the Infosphere:  Semantic 
Interoperability and Lexicon Development  Language 
Evaluation Resources Conference Workshop on 
Question Answering Strategies, Las Palmas de Gran 
Canaria, Spain. 
 
Evren Sirin, James Hendler, and Bijan Parsia, 2002.  
Semi-automatic Composition of Web Services using 
Semantic Descriptions. Accepted to Web Services: 
Modeling, Architecture and Infrastructure workshop 
in conjunction with ICEIS2003. 
