Ontology Summit 2011 Communiqué: Making the Case for Ontology

[Michael F Uschold <uschold@xxxxxxxxx>]

Here is a snapshot of the draft so far.   John Bateman took the original contributions from the track chairs and folded into a first draft.  The editing team had a look at it and came up with suggestions for improvement.

 

Michael 

Ontology Summit 2011 Communiqué:
Making the Case for Ontology

Version: -1.0.0 (pre; rough version: JohnB)

The Ontology Summit 2011 is the 6th in a series of annual events, by and for the Ontology Community. This year’s summit was co-organized by Ontolog, NIST, NCOR, NCBO, IAOA and NCO_NITRD.

Summary

Each annual Ontology Summit initiative makes a statement appropriate to that year’s theme as part of our general advocacy designed to bring ontology science and engineering into the mainstream. The theme this year is "Making the Case for Ontology”. The rapid growth of both ontology-related technology and information-rich application challenges makes it more important than ever that ontology developers and providers communicate effectively with decision-makers concerning how, where and why ontology can be effectively deployed. This communiqué summarizes the consensus of those who were engaged in the year's summit discourse on making the case for ontology and of those who endorse below.

NB: blue text denotes that the content is not part of the draft, but comments or suggestions that may contribute to the final communique.

Key Points to make in the communique

1. Ontologies are ready for prime time
1. (many examples in industry today, e.g. major manufacturers doing things. Amdocs, …Siri?)
2. many vendors and tools
3. many indicators of activity (conferences, jobs, etc)
4. emerging standards,
2. ontologies give flexibility (in various ways).
3. ontologies a boon for interoperability/integration
4. ontologies help people (even before computers) understand each other...
5. representing knowledge explicitly is a new transformative technology  needed to deal with current scale and diversity of Internet of services and things.
6. (good) ontologies “save lives” in critical applications... (?)
7. have specific set of things that are very good for investigating now with semantic technology, prerequisites for what makes sense.

Other important points important to elaborate.

1. ontology not just taxonomy and classification, also important to have axioms and reasoning.
2. use of ontology has a social dimension
3. good ontologies important compared to bad ones
4. make case that simple lightweight ontologies can be fine when meaning is shared for terms.  Otherwise, you need more axioms in the ontology to clarify meaning.
5. There are two main ways of using ontologies for information integration:
1. the meaning of terms is shared already, and ontologies (mainly taxonomies) are used to reason on the relationships among terms
2. the meaning terms is not shared already, and (deep) ontologies are used to understand and compare such meanings

Editing Team Tasks

1. need at least one case study spelled out in some detail. Two others very different summarized more concisely.
2. research use of ontologies in Siri before we commit to this example.
3. make sure we have key takeaways and a clear call to action,
4. Value metrics: there are few/no examples of actual metrics, needs to be fixed.
5. grand challenge: not clear how this can be converted into tips for readers in making the case.
   IDEA: have it after Mills’ summary of what things to look for for semantic technology. ROle is to inspire people.

Writing Tasks:

1. [JohnS] Write opening paragraphs, hook people in.  
2. [Mills] Pull together summary material showing what are good areas to use for ontology projects.

Ideas for opening hook

Ali Hashemi, Todd Schneider
Every person, organization or system in interacting with the world access (at the very least) an internal ontology – the things presumed to exist in the world and how they behave. Indeed, these very assumptions pervade and underpin our deliberations, inform our decisions and ultimately guide our actions. In large socio-technical systems, such as companies or organizations, each person, each technological artefact and system carries with it their own view of the relevant world. Reconciling and streamlining these fragments\, means understanding and reconciling the employed, often tacit ontologies.

Growing complexity and a need for smarter use of resources and solutions that cut across silos, means that it has become ever important to make explicit these implicit ontologies.
Concurrently, advances in computing, networking technologies and the Internet means that it is possible to fruitfully leverage computational ontology to address an increasing array of socio-technical problems. Moreover, in recent years, we have witnessed the increased maturation and transition of ontology from academia to industry and government. The time is ripe to know what you know and share it with others.
--
Michael Uschold:
Start of with seomthign that is surprising, the the reader may not have known.  I'm going to make up some 'facts' below that wodul be good if it were true:

1. 50 Fortune 500 companiese have hired ontologists in  house
2. Company X saves $10/year after a $3m one time investment in ontologies
3. The nubmer of vendors has increased from 50 to 500 in the past 3 years.

Mills may have some of these kinds of figures at hand.  Ideall we want to pick one ontology application that is exemplary is every way. Suggested criteria for the perfect one:

1. There was a calculated ROI or other well documented benefits
2. Someone on our team persononally knows someone on the project, so we have more confidence than reading hype form a company pitch.
3. The use of the ontology has to have been critical to the success of the project, not just "one was used, and it helped a bit".  
4. The way that the ontology was used has to be well documented and easy to explain.
5. The company should be a household name, e.g. fortune 500.
6. It is easy for everyone to do their own research on the project from publicly available documentation.
7. The company has at least one full time ontologist on staff, possibley.

We may not find one that matches all of these, but we should aim for the best one.

I woud like to propose the Sallie Mae example that Dave McCoomb talked about.  There was a very substantial ROI, but I do not know details of how or whether it was calculated.  I know the people who worked on the project, we can get details on exactly how the ontology was used.  It was critical to the success of the project.  It is not too hard to explain where the value came from.  A key benefit is flexibility.  They went out and hired a full time ontologist after the project was over, that person continues to talk about ontology projects at SemTech events.   There is some material on this that is publicly available from talks at SemTech.

Can anyone think of a better example?   SIRI has been proposed.  I would love for this to be our example, but I am doubtful that it meets most of these criteria. They don't talk much about the technology, so we have no idea if they used an ontology or how.  Tom Gruber had a successful company in the travel area. It looked like he could have been using ontologies. I asked him and he said not really, other than having a simple taxonomy to do geo containment reasoning.  I have no information about SIRI, but I would guess the story there would be the same.  The key technology there was voice and NLP and maybe machine learning.  The key to success was the business model, not just the technology.
--

Arturo Sanchez:
Paragraph 1:
The Data Interoperability Problem (DIP) is ubiquitous; instances of it appear in any branch of society in the following pattern: (a) entity X possesses data that entity Y wants to peruse; (b) although X’s data is immediately available, it cannot be used by Y because the software processes that will consume them do not “understand” them directly. An example that has been receiving ample national attention in recent years is that for which entities X and Y are healthcare organizations. Consider the simple—and yet so common, and sometimes so tragic—case of a patient who goes to an emergency room (ER) of hospital X, and cannot immediately share her/his personal health records with ER physicians, because they are under the custody of healthcare organization Y; and X and Y do not directly “talk to each other.”

Paragraph 2:
If you were now consulting with a healthcare provider (HCP) in some clinic/hospital, most often than not, s/he would not be able to access your electronic health records (EHR). Not because you do not authorize it, or because it is physically/technologically unfeasible; but because the various software systems where your EHR are stored do not interoperate with the HCP’s. This is an instance of the Data Interoperability Problem (DIP) applied to EHR.

Introduction

While the field of ontology, in the information science sense, has blossomed since the late 1980s, the use of ontologies in commercial applications is only now beginning to be exploited and recognized by the mainstream technical community. Many in the ontology community are asked for good examples where using an ontology brings clear benefits to addressing a commercial need - indeed the quest continues for the "killer app" for ontology. Particular challenges in making such a case effectively are raised by finding the appropriate language for communicating with very diverse stakeholders and by the need to be able to draw on existing experiences and benchmarks for problem solving when using ontologies. To support this process, this year’s Summit has focused on ways of classifying the uses that have been made of ontologies and ontology-based technology to date and on providing value metrics by which results can be quantified and communicated in terms relevant for stakeholders. In addition, experience in applying ontologies is being gathered in a growing list of case studies where benefits of applying ontology are documented. The Summit also discussed strategies for employing the information gained for effective communication with decision-makers: i.e., how best to make the point that ontologies may be what is required to get the job done, whatever that job may be for individual stakeholders. Basic awareness and trust on the part of stakeholders that ontologies are worthy of  consideration as practical solutions for semantics-rich applications also need to be addressed and strengthened and for this the Summit addressed potential “Grand Challenges” that might serve as high-profile focus demonstrations that ontologies work.

To address these concerns the Summit organized contributions along 5 tracks with many points of inter-connection. The principal intent of this collection is to start providing members of the ontology community and ontology developers in general with a toolkit that can assist them in making the case for ontology with diverse parties with correspondingly diverse requirements and goals. The main conclusions reached in the 5 theme areas are presented in this communiqué along with pointers to further details and places to obtain information in the accompanying materials. The themes of the Summit are now described in turn; they are organized as follows:

1. Strategy: how to best articulate a case for ontological applications with decision-makers and stakeholders?

2. Ontology Deployment Framework: how to describe ontology applications and their benefits?

3. Case Studies: what experiences are there in the community on successful ontology application?

4. Value Metrics: how to measure the return on investment in an ontology application (cost, capability, performance, etc.)?

5. Grand Challenge: how to raise awareness of the potential impact of ontologies and their application?

Strategy

Ontology is always an enabling component of some larger project or programme. The benefits of an ontology appear in the successful integration and interoperability of the whole rather than any specific part. Recognition of this may lead ontologists to keep the use of ontology under the hood, rather than something that is argued for up front. However, it should equally be possible to explain how an ontology contributes to the overall success, even if the overall success cannot be claimed for the ontology alone. The ontology deployment framework below provides ways of characterising just what role the ontology technology is playing in complete solutions, while the value metrics below give ways of measuring and documenting those benefits.

In commercial and public administration environments the route to showing how ontologies support better decision making through better informed decisions. An ontology will either be used to automate some part of the decision making process, or be used to improve the quality of information used in decision making. It is usually relatively easy to show that better decisions will lead to better business outcomes, or alternatively that poor decisions based on poor information lead to poor business outcomes. It can be quite a valuable exercise to look at recent poor outcomes and reconstruct how these resulted from poor decisions that were poorly informed. This can be a good way to find the justification for projects in which ontology will have a role, and as a way to assess whether potential projects are likely to deliver benefits.

Two types of business case are the incremental and the disruptive.

• An incremental business case is one where something that is being done already is done more efficiently. An example would be the replacement of the manual rekeying of information from one system into another by an automated interface that used the mapping of the semantics of one system to the other.

• A disruptive business case would be one where something that was not previously practical is enabled through the use of ontology. This will typically result from reducing the cost and improving the timeliness of providing information. An example of this is the Siri product that supports voice or plain text questions, returns options, and is able to transact on them on the user’s confirmation. Not only does it provide the natural language processing to identify the query, but it is context and location aware, and it integrates quite a number of web based resources to produce the answer in a reasonable time, and optionally, act on them.

There are a number of strategic areas where there are opportunities for an ontological approach, including: high volume data, particularly Linked Open Data, high value data, where the quality is critical to success, integrating data crossing boundaries of e.g. legislative systems, industries, disciplines. There is also at least one area where an ontological approach has a strategic advantage, Master Data Management. Master Data contains the ontology of the enterprise in terms of its organization, products, geography, and assets – both physical and financial, their classifications, and the business rules that govern them. Relational approaches struggle with Master Data because of the high level of properties and inter-relationships that exist between them, thereby providing an opportunity for semantic technologies.

Since there are many different audiences involved, it is also essential to appropriately tailor the case for ontology and ontology-based technologies according to the audience and the diverse roles taken up ny members of that audience within an overall decision process. Categories of roles for stakeholders include the following groups of people we need to 'Make our Case' to:

• Policy Makers / Strategic Decision maker ... convincing them that this is the strategic direction to go

• Budget Holders / Business Decision Maker ... to get the work funded

• Technology Decision Makers (CIOs, Architects, etc.) ... convincing them that this is the approach (at an architecture level)

• Technology Implementers (engineers and developers) ... convincing them that this is the approach (at the implementation level)  

• Users/consumers of the technology ... create the trust and the buzz, as they will be driving the needs

• Educators ... to produce the people with the right skills to actually do it (when the market is there: cf. OntologSummit2010)

[This was originally in the value metrics track, but it sounds to me as overlapping with the Strategy track: combine these as suggested here?]

The strategy is to move from the general to the specific: From the big picture problem down to the specific expectations of employing ontologies. The scope or context of ‘the big picture’ depends on both the audience or stakeholders involved and the problem(s) to be solved. The problem(s) and its scope can be drastically different between a president of an organization, a chief information officer, a department manager or IT staff and so it is crucial to adopt appropriate granularities and abstractions concerning just what problems are being addressed with what kinds of benefits.

The strategy is divided into four steps.

1. identify the business or operational problem to be solved.

2. classify the type of solution or application needed to solve the problem(s) identified. The starting point is the Ontology Deployment Framework given below. A particular solution may involve aspects from several of the types identified there.

3. identify the stakeholders for the business problem identified in step one. This may depend on the scope of the problem being addressed. Those who make policy or set strategic directions would more likely be involved in problems that span the entire business or organization. At some point those that hold the money or make the budget decisions will need to be convinced.

4. select the value models on the basis of the previous three steps.

5. identify the strategy to be used. Is the problem to be solved with an incremental approach or a disruptive change? The metrics used for each strategy will probably be somewhat different in that a disruptive change would most likely require more extensive metrics in order to justify the disruption.

Ontology Usage Framework

The objective of ontology usage framework is to provide a common terminology for describing applications of ontologies and the benefits that ontologies deliver within these applications. In addition to classifying ontology applications, this provides the basis for benchmarks and the ability to compare different applications of ontologies. Characterizing at the outset just what an ontology is to be used for is a critical initial step for motivating appropriate technologies, selecting relevant case studies documenting previous experience, and specifying in advance what the relevant metrics for evaluation are going to be. These all need to be seen from the perspective of those who expect the use of ontologies to be for their benefit - i.e., those who need to see the cost benefits of their investment.

The first demand is to identify what problems ontology-based technology are to be employed for; the Summit identified the following classes of applications:

• Integration: In this class of applications, the primary functionality is the matching and mapping of concepts, while the primary architecture is within a set of multiple systems. Ontologies are typically used at runtime by application developers (who are in the best position to write translators among the systems). Examples include information integration, database integration, software interoperability, and software access.
• Decision support: In this class of applications, automated inference is the primary functionality.  For example, deduction from Axioms via Theorem Provers, deduction from Axioms via Special-Purpose Reasoners, and validation of integrity constraints
• Semantic augmentation: Applications in this class focus on using the ontologies as the basis for specification within a single system. As a result, the ontologies are used primarily at runtime by knowledge workers and application developers. For example,  ontology-based algorithms, ontology embedded into software, model-driven design, ontology-based standards, and adding semantic meaning to existing systems.
• Knowledge management: In this class of applications, knowledge workers use ontologies for classification within a single system. This includes knowledge discovery, categorization of information resources, and knowledge capture.

Each of these classes of applications need to be additionally cross-classified according to several functional dimensions in order to pinpoint relevant evaluation metrics and case studies. The Summit identified the following dimensions:

• functionality: how is the ontology being used?

• architecture: single system or across systems?

• use time: design time or run-time?

• ontology users: who is using the ontologies (ontology author, data author, application developer, application user, knowledge worker)?

• ontology content: what ontologies are being used?

From the viewpoint of the application, particular attention needs to be paid in addition to the expected benefits of employing ontology-based technology, what does the stakeholder expect to be the return on investment? This needs to be characterized according to a precise understanding of just what problem is being solved expressed in the users’ own terms. Here also the impact of the ontologies on the functional/nonfunctional requirements of existing systems need to be carefully assessed. Ontologies may be allowing those systems to provide new functionality that was not previously available due to their semantic augmentation of the application, or they may equally be improving existing functionality in terms of quality, performance, maintenance, cost, portability, reliability, scalability, robustness, usability, and extensibility.

Finally, the enterprise context must be established; this defines various environments for the deployment of ontology-technology, each with differing requirements and evaluation demands. For example, an enterprise will have at least one and maybe several core processes, such as the provision of goods or services to customers. These may be supported by information systems. A key part of this will include Master Data Management so that all the systems use the same identification for key things like products and customers: these may be subject to ontological modelling. There will also be a number of things needed to support the core processes, such as factories and vehicles, each of which has a lifecycle that needs to be integrated. This requires information to be passed between lifecycle stages and between the information systems that support the different processes in those different stages, presenting different opportunities for semantic augmentation by ontologies. When enterprises procure goods in particular, but also services, they do not just capture price information, part of the product will be information about that product and its use. This information needs to be integrated within the systems that are involved with the operation and use of those products. And, to understand how well an enterprise is performing its processes, the data arising from the performance of its processes must be analysed, which also usually involves collecting and integrating the data from multiple systems. Being clear about just where ontology-technology is to be employed with an enterprise is likely to support a better shared understanding of what the benefits, impacts and relevant metrics for measuring success of the ontology deployment are going to be.

Case Examples

Ontologies solve a broad range of real-world problems. Summit sessions presented twenty case examples of mainstream applications where knowledge technologies, semantics, and ontology are being applied and delivering benefits.  Here is a list of the case examples documented in the supporting materials:

• Apple SIRI — Virtual assistant as a next UI paradigm for consumer markets

• BeInformed — Smart knowledge-driven citizen-centric services

• Cambridge Semantics — Do it yourself data exploration and analytics

• Connotate — Do-it-yourself semantic agents to discover, aggregate, analyze & report information

• DHS — (a) DHS infrastructure taxonomy; (b) Complex event modeling, simulation and analysis (CEMSA)

• EDM Council — Standardization of terms and definitions for financial services and a pilot test of the semantic resource as applied to mortgage-backed securities

• Franz — AMDOCS -- Ontologies and CRM for Telecoms

• IBM — Dr. Watson Project

• Innovative Query — (a) Content intelligence and smart applications; & (b) Semantic BI for blogging

• Mayo Clinic — Relationships among biomedical ontologies and classifications

• Model-driven Development — Architectures and ontologies for business value

• Recognos Financial — Better access with semantic search, navigation, query & question answering

• Revelytix — DoD knowledge-centric information webs & process interoperability

• Sallie Mae — Integration of multiple systems from multiple companies

• Sandpiper — Semantic technology in rental product marketing

• Semantic Arts — Applying semantics to enterprise systems - Proctor and Gamble case study

• Top Quadrant — Valuing the harvest from using ontologies (includes multiple examples)

• Trigent Software — (a) Ontology and rules provide rapid natural language understanding; (b) Ontology and rules provide mass customization of vehicles

• Visual Knowledge — Policy-driven compliance, risk, and change management pilot

• zAgile — Model-driven framework for process deployment with  extreme traceability

When using case examples to make the case for ontology, place emphasis first and foremost on communicating the value to the business. Our template for summarizing the value delivered iincludes the following topics:

• Challenge — What is the problem or opportunity being addressed? What is its significance?

• Solution —  What is the solution? What role do knowledge technologies, semantics, and ontology play in it?

• Screen shot  and key features —  Give a flavor of the use of ontology in this case.

• Benefits — What measures were used to demonstrate the value delivered?

• Resource — What commercial or non-profit entities provided technology products and services used in the case example?

Where and how are ontologies already being applied and delivering value?  The short answer is just about everywhere you look. In addition to the twenty reference case examples, the summit explored a broad range of consumer, social, enterprise, and systemic applications where ontologies are being used, for example:

• Provisioning and securing dynamic infrastructures where information of very different kinds is being combined from different sources on a large scale

• Using ontologies to augment knowledge worker insights and decision-making as well as application-relevant semantics and business rules in off-the-shelf commercial software

• Understanding speech, written and visual language, and intersemiotics

• Adding intelligence to the user interface to provide new capabilities and enhanced user experience

• Applying  machine-learning, data-mining and other scale-based automatic techniques for the semantic document processing and advanced analytics in fields such as law, medicine, science, defense, and intelligence  

• Enabling high-productivity, knowledge-driven collaborative work processes

• Accelerating knowledge-intensive activities such as modeling & simulation, acquisition, design, and engineering

• Powering mission-critical processes in energy, financial services, logistics, manufacturing, and transportation.

• Managing networks providing diagnostics, logistics, planning, scheduling, cyber-security, and event-driven processes

• Supporting adaptive, autonomic, & autonomous processes such as robotics, intelligent systems, and smart infrastructure,

• Assessing risk and compliance in policy-driven processes such as exceptions, fraud, case management, predictive analytics, and  emergency response,

• Building systems that know, learn, adapt & reason as people do such as e-learning, tutors, advisors, cognitive agents, and smart social games.

Where do knowledge technologies, semantics, and ontology add value?  Patterns of value delivery that emerged from these case examples have several dimensions worth noting:

• Knowledge-centric approaches and ontology-driven solutions were associated with both development and operational gains in performance and lifecycle economics.

• Knowledge technologies and knowledge-centric approaches to knowledge work accelerated the time from idea (or need) to a solution.

• Knowledge technologies enabled new methodologies and concepts of operation that resulted in significant productivity gains. Fewer people could accomplish the same work significantly faster and do-it-yourself empowerment of subject matter experts and business users resulted in huge improvements in cycle times and service levels.

• Knowledge technologies enabled new solution concepts  Knowledge-driven solutions could be documented where the design is self-documenting, the execution is self-explaining, the solution can be easily modified, simulated and updated, and total cost of ownership decreased by more than 50% compared to current IT solutions.

• Knowledge technology based solutions were more flexible and agile than IT-based solutions when it comes to making changes and evolving capabilities.

• Knowledge technologies and ontology were used to add intelligence to the user interface (UI) in order to increase relevance, helpfulness, utility, and pleasure as experienced by the user.

Value Metrics

Value models and metrics help make the case for using ontologies in information systems. Due to the diversity of functionalities and kinds of applications to be supported by ontology-based technologies, the value models and their metrics needed to make that case must be much more granular, less technically detailed and more case-specific than, for example, metrics already employed in software development. In particular, metrics need to be focused on the particular business problem(s) for which the ontologies are to be applied. In the sense of providing a toolkit for making the case for ontologies, this track sets out a value paradigm for those promoting the use of ontologies.

The value metrics identified vary depending on which stage of making the case for ontology is involved and for each particular audience as classified above. The following value models concern the first stage of making the case, i.e., identifying the business or operational problem, and derive from the case studies presented and identify results-areas for improvement of a business, organization or enterprise. They are not meant to be exhaustive but should be used as a starting point to help identify the general area of the business problem. The ordering of these value models moves from infrastructure to entire business.

• IT Efficiency – IT has become an integral part of most businesses and enterprises, therefore merits investments that benefit an entire organization though there may not be direct benefits to specific operations or business objectives. If IT efficiency is the problem space to be addressed, then what must next be identified are the particular aspects or area of IT that will be made more efficient. Such efficiency may be either organizationally or technically operational. For example, the use of ontologies may allow more agile and complex workflows, more responsive SOA implementations, more rapid prototyping of services or greater clarity in communication with (internal) customers.

• Operational Efficiency – This refers to the operations of a part of a business or enterprise. This means that the problem to be addressed is relative to just part of the business or organization. For instance, it may relate to only one department (e.g., Human Resources, Facilities, Finance, etc.). Many businesses and organizations discover, as they grow larger, the need to separate various responsibilities in order to increase operational efficiency and productivity. So it may be that the problem to solve involves improved search or discovery. Maybe, it is the ability to have quicker and more precise responses (e.g., invoicing or billing).

• Business Agility – The ability of a business or organization to alter the products or services it provides is demonstrably crucial for many businesses to continue operating. Agility may include interoperable services the allow new products and services to be developed. Or it may involve acquiring actionable competitive business intelligence or mining a business’ own data to discover new opportunities.

• Business Efficiency – This refers problems that cut across or involve many to all parts or aspects of a business or organization. One common problem in this area is common terminology. Large organizations can evolve different sub-organizations that develop mission-specific terminology peculiar to their operations and products (i.e., their mission). These differences in terminology can become embedded, either implicitly of explicitly, in their business processes and supporting information systems and impede development of common processes or services.

• Customer Satisfaction (internal & external) – It should be kept in mind that customers can be either internal, as is the case for many IT departments, external, or both. For businesses that deal in the retail area customer relations is a common problem area. The ability to reduce the costs of managing customer relations, commonly referred to as customer relations management (CRM), or make better use of the data gathered from these activities is area of great interest to such businesses.

[were these metrics? Also reduce paras to summaries to make space for figures perhaps?]

Grand Challenges

In the Grand Challenges track experts in particular domains were asked to summarize the state of the art, determine gaps that hinder real-world ontology applications, and to suggest actions for overcoming these gaps. Using this background the idea is to come up with a grand challenge problem, which can benefit from ontologies. We believe that the grand challenge problem should be similar in scope to other challenge problems, such as the X Prize Foundation’s Ansari Prize, and will be utilizing a variety of  technologies and methodologies.  Ontology use can be one such technique.  Three primary domains were identified: 1) health care, 2) social networks, and 3) home land security.  [why not also geo and the national/international initiatives for combing info?] Summaries of grand challenge problems in these domains are discussed next.

[copy over challenge descriptions and/or pictures, summarised to fit]

Health Care (Eliot Siegel,  Christopher Chute and Christopher Welty)

Watson, developed by the team at IBM (of which Chris Welty is among), uses a multitude of techniques to answer questions posed in natural language. Recently, it beat two former Jeopardy champions on the Jeopardy quiz show, although it lost to Rush Holt, a US congressman from New Jersey.  The DeepQA methodology used in Watson is well documented (or soon to be documented). One technique that DeepQA uses to evaluate candidate answers -- generated through a parallel search technique -- makes use of distributed ontologies, as will be described in a forthcoming paper (see https://researcher.ibm.com/researcher/view_page.php?id=2121). After successfully demonstrating Watson’s capability on the Jeopardy show, IBM believes that Watson can “transform the way that healthcare professionals accomplish every day tasks.” (See IBM website). [insert?]  To achieve this IBM is collaborating with several institutions, including Siegel at University of Maryland School of Medicine.  “Dr. Watson” can be used in two settings: 1) during medical school; and 2) after medical school.

Watson helping the medical student:  Medical knowledge from various sources, such as Harrison’s Principles of Internal Medicine,  NEJM’s (New England Journal of Medicine) cases and quiz material, can be utilized in their digital form. This 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.”

Watson after medical school:  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 tasks clearly require the development of appropriate ontologies. Christopher Chute discussed the progress on one such ontology – ICD 11 –and its relationships with other ontologies/classification systems (see the ontology spectrum).  The lack of ability of logical assertions to handle probabilistic reasoning was pointed out. However, one of the participants indicated that the UMBC’s Ebiquity project extends OWL to deal with uncertainty using Bayesian networks.

Social Life Networks (Ramesh Jain)

Social Life Networks (SLNs) combines 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. This necessitates the development of ontologies which capture objects and events in SLNs. Creating and testing such ontologies – which Jain calls “Recognition Ontologies” – will aid in effective recognition and reaction in a network-centric situation awareness environment.

Homeland Security (Nabil Adams)

The Department of Homeland Security (DHS) is tasked with protecting the nation’s critical infrastructure. To do so it must facilitate seamless communication and dissemination of information.  There are several initiatives at DHS which aim to accomplish this mission. One such initiative is the Complex Event Modeling, Simulation, and Analysis (CEMSA) program, which provides “DHS analysts with models, simulations, tools, and data to assess the consequences of multiple interacting complex disruptions to critical infrastructure and key resources.” A key requirement for success of such a program is the development of ontologies in a number of domains, which will aid in both simulation and information integration activities.

The X Prize (Christopher Frangione)

The X Prize Foundation (http://www.xprize.org) is a nonprofit organization which aims to address grand challenges through “incentivized competitions.”   Currently there are prizes targeted to several grand challenge problems, such as the Ansari prize for, which is a “$10 million competition for the first team to build a privately funded spaceship, able to carry 3 adults to 100 kilometers altitude, land safely, and fly the same ship into space again within two weeks.”  The development of the prize – which comes in several flavors, such as participation, exemplar, network, exposition, point solution, market simulation -- involves several steps: setting goals; identifying potential sources of funds; designing the prize; marketing; executing; and determining strategies for “life after prize.”  The challenge for us is to develop such a prize for one of the proposed grand challenge problems.  We could use the methodologies outlined in the various tracks (in this symposium) to develop a X Prize for a chosen domain (e.g., Dr. Watson, SLN, and CEMSA)

[how about a grande challenge such as:]
The World Map

National mapping agencies have for many years been concerned with combining diverse sources of information and making them interoperate: typical examples include shapes and boundaries, names, structures, elevation, transportation networks and many more. Ontologies have therefore been prominent in these activities, resulting more recently in the emergence of the GeoSemanticWeb. Now as a vision for the next generation of massively extended ‘map’ services, initiatives such as the US Center of Excellence for Geospatial Information Science
(CEGIS), supporting the US National Map, and the Infrastructure for Spatial Information in the European Community (INSPIRE) are exploring systems that combine an unprecedented range of broadly ‘geographically’-positioned data, including legacy and historical data, real-time sensor data, architectural data and much more. The grand challenge here is to take the combination and access to such data to the next level: uses are foreseen in support services for environmental decision making, accessing natural and cultural heritage, recreation opportunities, stewardship of energy and mineral resources, sound use of land and water resources, conserving and protecting fish and wildlife, simulation of catastrophic events and event response and many more. For this level of interoperability at all levels, ontologies are an essential component of providing the semantics necessary. New data models and techniques, as well as combinations of existing semantic technology, will all need to be moblised (cf. here).

Conclusions

We note that the discussion during this year’s Summit has focussed on making the case for ontology projects within a business context where there is a financial justification for the use of ontology. However, there is also a case to be made for ontology research and the determination and pursuit of research goals for a case to be made for. The Grand Challenges track may address this, but it has not generally been addressed elsewhere. Generally, the approach in this summit has been pragmatic, emphasising the need to “put food on the table”.

Acknowledgements

Grand CHallenges:

The track had six speakers, as indicated below.

1.      Christopher Welty , IBM, Topic: Grand challenge for Watson-like Systems

2.      Ramesh Jain, University of California – Irvine, Topic: Social Life Networks – Ontology-based Recognition

3.      Eliot Siege, University of Maryland School of Medicine, Baltimore, Topic: The Dr. Watson Project: Clinical Perspective

4.      Christopher Chute, Mayo Clinic, Topic: Relationships among Biomedical Ontologies and Classifications

5.      Nabil Adam, Department Homeland Security and Rutgers University, Topic: Ontology Applications in Homeland Security

6.      Christopher Frangione, X Prize Foundation, Topic:  Revolution through Competition: Designing Effective Incentive Prizes

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2011/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011  
Community Portal: http://ontolog.cim3.net/wiki/     (01)