Ontology Summit 2011: "Grand Challenges" synthesis

Co-champions: RamSriram, Ernie Lucier, AldenDima

Overview

The Grand Challenges Track had the following goals:

• Invite speakers from a variety of domains

• For each domain presented:
  • Identify a grand challenge problem
  • Summarize current state of the art on ontology use
  • Determine gaps that hinder real-world ontology applications
  • Enumerate the actions that need to be taken to overcome these gaps
  • Brainstorm ideas for a grand challenge problem

Speakers and Domains Represented

Our invited speakers represented three domains (Health Care, Social networks and Homeland Security):

• ChristopherWelty (IBM) - Grand challenge for Watson-like Systems

• RameshJain (UC, Irvine) - Social Life Networks - Ontology-based Recognition

• EliotSiegel (UMD, School of Medicine) - The Dr. Watson Project: Clinical Perspective

• ChristopherChute (Mayo Clinic) - Relationships among Biomedical Ontologies and Classifications

• NabilAdam (DHS) - Ontology Applications in Homeland Security

• ChristopherFrangione (X Prize) - Revolution through Competition: Designing Effective Incentive Prizes

Health Care

Grand Challenge:

Much of the Health Care domain presentations focused on IBM'S Watson which has captured much attention. It was built by IBM to advance QA (question answering) techniques. In a heavily publicized competition it won against two previous Jeopardy champions eventually but lost to Congressman Rush Holt. Its [[DeepQA]] components use the "Hypothesize-Score-Rank" strategy where the scoring (Evaluating) employs non-linguistic, ontology and background algorithms.

Challenges were identified in two settings:

• 1) during medical school; and
• 2) after medical school.

Helping the medical student

• Medical knowledge from various sources, such as Harrison's Principles of Internal Medicine, Merck Manual Medical Library, Washington Manual of Medical Therapeutics, NLM's Clinical Question Repository, NEJM's (New England Journal of Medicine) CPC (Clinicopathological Conferences) cases and quiz material, can be utilized in their digital form.
• The above can be augmented with the computer-based simulations of human physiology and disease processes.
• Machine learning techniques can mine knowledge from Electronic Medical Records (EMRs)(secondary use of data) and provide "experiential learning."

Helping the physician

• Continue providing access to medical knowledge
• Automated chart review of EMRs
• Access multiple databases in a unified manner
• Aid in diagnosis and treatment, though encoding expert knowledge bases, including drug-drug interactions.
• Address questions of teaching Dr. Watson bedside manners.

The above challenges need advances in the following areas.

• Mining and molding knowledge
• Machine Learning
• Systems Medicine (integrating genomics and clinical systems)
• Image analysis and interpretation
• Management of medical vocabularies

Social Networking

Social Life Networks (SLNs), which are instances of Net-centric societies, combine two evolving paradigms: computer-mediated social networks and Internet of things. Jain calls this "connecting people with resources," since an SLN can be viewed as a network of people and sensor objects, such as mobile phones and associated senors. In SLNs a considerable amount of data -- image, text, other sensors -- passes through the network and should be converted into higher abstractions that can be used in appropriate reasoning.

The discussion of Social Networking focused on three themes:

• The emerging Internet of things
• Systems for Situational Analysis and Recommendation
• The need for context-based interpretation of images and text

Several trends were identified as driving Social Networking:

• Their expanding role in communication
• Micro-blogs are becoming major source of News
• The emergence of the Internet of Things
• More than 75% of the world population owns mobile phones.

Challenges to be addressed in SLNs:

• Security of information
• Making sense of data/information
• Multimodal sensor integration
• Dealing with system complexity

Ontologies could help in developing strategies to address above challenges.

Homeland Security

Homeland Security was the final domain to be presented. The DHS infrastructure and the needs for modeling, simulation and analysis were discussed. In this domain, the development, validation, acceptance, update, and integration of ontologies were seen as the key challenges. Other potential domains were identified: Sustainability, Emergency Response Management, and Financial Services. These domains are important but we were not able to cover them in our teleconference.

Other Challenges

Several other domains were identified during the summit. These include the following domains:

• Sustainability
• Natural disasters
• Financial (including help with taxes)

Incentive for Progress

The final portion of our discussion focused on developing incentive systems for moving forward in grand challenge areas. The X Prize was presented as an example of using prizes to motivate progress in key areas. The various attributes, models and development phases of prizes were presented. The $10M Ansari X Prize was offered as a successful example of using prizes. In the resulting competition, 26 teams from 7 nations spent over $100M to attempt to be the first privately-financed team to fly a spaceship capable of carrying three people to 100 kilometers altitude twice in two weeks. The next step for the community is to pick one domain and develop the appropriate X prize.

Reference

• Ontology Summit 2011: "Grand Challenges - I" Panel Session - ConferenceCall_2011_03_24

• Track-5 Summary Report presentation at the OntologySummit2011_Symposium

--

maintained by the Track-5 champions ... please do not edit