On Semantics, Semantic Web and Services Science - Amit Sheth

Interdisciplinary research - learning from the success of Steve Jobs

2011-12-21T06:27:00.000-08:00

This is a companion piece to the previous ones that shares some thoughts about the choices students seeking to go for graduate studies and research careers face.

I recently heard +Fareed Zakaria's interview on CNN GPS today w/ Steve Jobs biographer. It once again brought out the key to Jobs' success: confluence of arts/music/design w/ technology. My own students who came in with non-CS/IT background (Mathematics, Statistics, Management, Cognitive Science) ended up doing significantly better than their colleagues whose background was in CS (my area is CS). Meena Nagarajan, a BITS Management student, got invited to give a keynote at an international workshop even before she got her PhD (something totally unheard of), got prestigious NSF CI-Fellowship, and hired as a researcher at prestigious IBM-Almaden (and a Stanford post-doc settled for post-doc in that dept). Satya Sahoo, a BS in Statistics, competed with 300-400 PhD applicants (including those from so called top 10-20 schools) and got hired as a tenure track asst professor in School of Medicine, and his Stats background, CS degree and interdisciplinary research in bio-medicine has something to do with his success. Even changing a subarea within field could be useful. Ramakanth Kavuluru, who did his dissertation research in security and post-doc with me working on application of Semantic Web to Biomedicine and Health Care applications got hired as a tenure track faculty in School of Public Health at University of Kentucky (all faculty recruiting is extremely competitive). This experience has led me to advice my straight tech/CS students to cultivate their hidden talent in non-CS and nontechnical subjects. Combining hard science skills with soft science expertise is usually very rewarding and open up lot more exciting career opportunities. Leaders like Steve Jobs succeeded because they identified unique need and importance of combining technology (a hard science) with design (a soft science), and creating something fairly distinct and new.

So, as you consider your options, do not be limited by your current degree-- also consider your interests. Let me give you an example: If you are a Computer Science student with some exposure to biology, you can consider a program like Biomedical Sciences PhD at a Wright State University. Don't just blindly run after so called highly ranked universities. This is a highly selective program (very few are admitted), but has 4 concentrations (one of the is computing focused, see attached brochure) but all graduates have excellent career choices and importantly for many, those admitted are fully funded for the duration of their studies of 4 or 5 years-- and that is a rarity these days!

Some more thoughts: http://knoesis.org/aboutus/joiningus/joiningfaq

21st century approach to seeking graduate studies

2011-12-04T20:04:00.000-08:00

Preface: I monitor some lists where undergrads discuss their quests and questions for graduate studies and applications. Here are some thoughts from someone on the other side of the equation. For those who are interested in a quick degree and a well paying job, this post is not too relevant. For those interested in building a career in which they have lot more control, those interested in research - whether thought a MS thesis or a PhD - this may offer a few useful pointers.

We are already a decade into 21st century. Looking at the posts of undergraduate seeking to go to the best school they can go to, it still seems perceived ranking -- typically of the university or department, plays a key role in deciding those 5-10 places most applicants focus on. This approach is very outdated.

Here is how I went about it (when I applied in 1980 for 1981 admissions). I had already decided to do PhD (my father was a professor and research attracted me-- already had 2-3 publications, and most importantly, I had a chance to work at ISRO which gave me good understanding of what it meant to do a PhD-esp. in those pre-Internet days). And had also decided to do it in the area of Databases. So I went to IIT-D library, found two most influential (at that time, most published authors in ACM/IEEE journals) professors in the most active area of the area (happened to be "database machines") that interested me (of course I could not fully understand most publications I stumbled upon). They happened to be at Ohio State and U Wisconsin, so applied there with the intent of working with either Prof. Dewitt (if I got a chance at Wisconsin) or Prof. Hsiao (Ohio State). I think my two internships at ISRO and letter by then Director of ISRO, Dr. Yash Pal, gave me a competitive edge when compared withmy peers with somewhat higher GPAs.

Much has changed: information is plentiful if you know where to look at, technology and

research areas evolve at much faster pace, there are too many exciting things happening even

in a subarea (eg, World Wide Web) of a discipline (Computer Science), most of the exciting

things are happening at the intersection of disciplines, there are more disciplines and interdisciplinary programs to choose from, a top high-tech company 10 years ago is no longer the most exciting place to be, there is a lot better insight into quality and quantity, you as a student have a lot more say in defining your own graduate study program …. and I could go on. Here is an excellent piece by Thomas Friedman: time has come to invent your own job, not apply for one.

So what does the above mean in your investigation and decision making process? First is that one cannot just focus on a discipline as a whole, university as a whole, department as a whole, and one cannot focus on the skill or technology. Broad ranking such as those in US News are practically useless except to make your family feel good if you got an admission in a program they rank highly. But for improving your own outcome at the end of your graduate studies, you would need to focus more on personalities—your own interests and strengths (eg, do you like hardcore science/math/stat/algorithms, or do you like softer things – modeling, knowledge, semantics, human computer interactions?). When you take a subarea of a field such as computer games or social media, you will find potential to work in either type of work. And you will need to ask who can guide you and provide an environment to grow as an independent researcher, innovator, architect or designer. When you think of a guide (advisor), think about (and investigate) the guide both as a technical expert as well as a person. Here are some of the things to do: ask if this person has reinvented (and continue to reinvent) during his/her career (eg., is the person doing the same thing s/he did during his/her PhD or more than any 5 year period?), how much impact his/her work has (not just number of publications), what type of funding (is there a healthy dose of highly competitive NIH/NSF funds?), how well s/he guides and sponsors (does students do variety of things, do they have freedom, can they travel anywhere if they have a paper? can they buy anything they need for research? is the faculty's group international or are all students from the same national/ethnic background (if latter, I would be cautious). Be sure to contact current advisees, and pay particular attention to how successful are his/her advises (perhaps the most important factor of all!). Understand that the outcomes vary widely. While salary is not the only or the main way to describe this, it is the easiest. A typical MS graduate gets $65K-90K in the first year. But my last two MS advisees who graduated in 2011 are making $110K and $120K in the first year working in companies in exactly the area they wanted to work in (Semantic Web and Cloud Computing, respectively). If you ask them, more than the money (which tells how much the company values them) they will point out the quality of people they work with, the ability to work in the area they wanted to work in, the ability to work on innovative products and services, and the ability to architect or design instead of just implement what a supervise defines. The same wide gap exists in the outcome for PhDs (it typically take more than 5 year to break even on income bases vis-a-vis a MS, but there are many other reasons one may want to do a PhD).

So how do you do finer investigations about choosing places to apply and faculty to work with?

Look at the web sites of the professor, his/her Google scholar/citation index (go to http://scholar.google.com or just Google his/her name), lab/center’s web site including its social media presence (eg, here is one of Kno.e.sis' social media presence-- notice the level and variety of activity on such sites), senior students’ web pages, funding information (especially in last few years)
Do not focus on the field as a whole (eg Computer Sc)—look at one or two of the subareas that attract you most. Example: Suppose your interest is in World Wide Web: then look at influence in World Wide Web for University (it will usually cover that university’s department’s outcome in that area) and who are the most influential authors in WWW. But say your interest is in Data Mining or Human-Computer Interaction. You will see an entirely different cast of organizations and researchers! And if you look up last five year versus “all year” statistics, you will see huge changes! What if your selection was based on department or professor who was well known but has shown little activity and energy in recent years? If you choose any one of the traditional “top twenty Computer Sc departments”, it is quite possible that it is not even top 50 in your subfield of interest. And you will miss out on gems! I can give you an example close to my heart – Wright State University (WSU) is at 4^th position among universities (8th including all organizations) in North America based on 5-yr impact in WWW and it is one of the smallest organization in top 50 (Kno.e.sis where practically all its work in WWW is done was established only in 2007). So while you will not many find WSU among top in many subfields (it is ranked 60^th in CS research spending now—not bad at all – in fact quite good for its size and might make it top 30 on per capita basis; on the other hand in Semantic Web and in several Web 3.0 topics it is probably the largest academic group in the USA with funding from Google, Microsoft Research, IBM Research, IBM Research and so on!). Now knowing such data could give you what people in industry call unfair (unique) advantage- something that you know before others find out!

Just looking at even subfield of WWW might not be good enough-- your results for infrastructure aspects of WWW would be very different than Web 3.0 (social, sensor, and semantic Web) topics!

Even the above strategy will not be full proof. Publication impact is just one of many factors of importance and success- how about industry relationships/collaborations, ability to offer vision and define new subfields? In the brave new world, grass root or field work, development of new tools and systems that is used by others, technology transfer and commercialization, writing influential articles in non-academic venues or significant impact through social media, etc. can be extremely important (for example, MIT Media Labs makes huge waves through non-academic papers).

Now that you have some more ammunition about investigating your choices, it is equally important to know how will you sell yourself (but not oversell) once you find your top choices. Unfortunately, this post is getting too long already, so let me point to some links at the end of this FAQ for graduate applicants. And here is the link I intend to add to that FAQ: The 9 things that matter more than GPA. When I evaluate prospective PhD students, I look for passion, desire to be at the top (ambition), work ethics (ability to work towards that ambition), communications skills, experience with or understanding of what research and innovation means (a taste of it demonstrated with research centric projects, publications and/or internships). Some of these come through the way prospective student communicates with me, personal statement (that says why you want to work with me or my group) and letters. And typically this involves one or more Skype or Google hangout calls. Of course, scrutiny for a MS applicant is lot less stringent, but then most MS students are not funded initially (many get funded while they are doing research and/or thesis component of their studies-- eg the above mentioned MS graduates did not have funding during the time they were doing core courses, but then they joined a research project and got funded. And don't you think the final outcome was worth it?).

Well-- good luck, my best wishes for success in this increasingly complex and interest times!

p.s.: [Jan 24, 2011] : The education system is changing very rapidly, see for example: Udacity and the future of online universities. The whole purpose of going to a university is changing rapidly, since you do not have to go to a traditional university to take standard courses. So you would increasingly go to a graduate program for activities that are more personalized, more interactive, and focused on learning that occurs beyond the course work.

Semantics Scales Up: Beyond Search in Web 3.0

2011-09-17T17:45:00.000-07:00

Note: The Internet Computing article is open access (no-cost access).
Following is pre-print material.

Abstract: Concern for scalability- both in computational terms and in terms of human effort needed to develop semantic models and background knowledge- have hampered adoption of semantic techniques and Semantic Web. This concern is now misplaced as we have seen extensive progress in last decade on standards, methods and technologies for developing semantic models or ontologies, semantic annotations, and techniques for semantic integration, analysis and reasoning. This is complemented by plenty of recent success stories- not all well publicized- that use semantics in broad based applications like Web search, as well as in growing number of vertical domains. As the future of computing expands beyond cyberspace to cyber-physical-social computing, with extensive growth in social and sensor data, semantics will play even larger and more pervasive role in exploiting larger amounts of increasingly heterogeneous and multimodal data.

Keywords: scaling semantics, semantic search, computing for human experience, semantics in Web3.0, semantics empowered physical-virtual systems, semantics empowered cyber-physical systems

Semantics can enhance a broad variety of information processing — search, integration, analysis, pattern extraction and mining, discovery, situational awareness, and question-answering. Consider search: a search system that could distinguish between “Merry Christmas” as a greeting and one of the 60 or so songs with “Merry Christmas” in the song title as cataloged in MusicBrainz (a community-created music encyclopedia, would have a powerful semantic search capability. Practical solutions utilizing semantics involve using a conceptual or domain model to organize information (MusicBrainz in our example), creating metadata (or annotations) with respect to the model (indicating as an attribute/label/facet of “Merry Christmas” whether it’s a greeting or a song), and then utilizing the metadata and model for enhanced computation.

Even in the Web’s early days, simple schemas and metadata were used for faceted or attribute-based search of Web-based documents and data. An example is the InfoHarness system at Bellcore, commercialized in 1995. It supported extracting metadata from heterogeneous data and provided Mozilla-browser-based faceted search. Tom Gruber introduced the concept of ontologies in an information systems context in the early 1990s. This term has since become increasingly used for conceptual or domain models that also capture shared vocabulary and agreement, often with relevant factual knowledge. Several efforts in the mid-to-late 1990s such as SIMS, Observer, InfoSleuth, and InfoQuilt, demonstrated ontology-based Web data integration and querying.

In the late 1990s, I realized that it was possible to design conceptual models or ontologies — not too different from schema.org descriptions today — for many domains of practical interest for Web search (politics, business, finance, sports, entertainment, and so on). Taalee’s MediaAnywhere, Voquette’s SCORE and Semagix’s Freedom (products/services from Taalee which I founded in1999 and its follow-on mergers) could then extract, integrate, and repurpose high-quality datasets to populate these ontologies with factual information and background knowledge. For example, these systems extracted the official site for Major League Baseball or a database with equivalent content to populate part of a baseball ontology, and used sources similar to MusicBrainz to populate the music component of an entertainment ontology.

These ontologies, represented in a Resource Description Framework (RDF)-like language, supported a semantic and faceted Web search engine, MediaAnywhere, developed at Taalee, (see http://slidesha.re/sw-ib, http://bit.ly/sw-p, and http://bit.ly/sw-ic). Although the system scaled to a few hundred websites and enterprise semantic applications, it was ahead of its time; the concept had to wait for technology and market acceptability to catch up. Would this approach have scaled to the Web? I believe yes, but I couldn’t have convincingly argued or demonstrated this — until now, when Yahoo, Bing, and Google are all exploiting minimal ontologies, metadata provided by content developers, and large subject- (domain-) specific object and knowledge bases. Before discussing this, however, let’s first review why some subscribed to the perception during the past 10 years that semantic solutions can’t scale.

Ontologies and Web Search

The Web has continued to see explosive growth in the number of documents and amount of data accessible through it. We can view earlier directory-based approaches by Yahoo and DMOZ as one type of semantic approach, given that they used human-created taxonomies to manually catalog information; these approaches soon failed to scale. Search became the primary way for people to find the information they needed. Its success led many (beginning with Googlers like Larry Page and Peter Norvig) to believe that all you need is enough data and you can adequately, or even exclusively, extract semantics in a bottom up manner — or that, given the Web’s broad coverage, ontologies, models, and background knowledge simply aren’t relevant nor would they scale.

Some in the semantics camp, including myself, felt that only a limited form of implicit semantics is embedded in most data on the Web, and that semantics can and will scale. I’ve argued that we can readily apply background knowledge in developing a semantic solution (as we did with MediaAnywhere). The reader can find a link to Norvig’s post and my views on this matter in a 2005 blog at http://bit.ly/s-search.

Early, albeit unnecessary, emphasis on an AI focus to defining a Semantic Web approach also hurt, given the lack of success in scaling AI solutions during the 1980s and ’90s. If you were to interpret ontologies narrowly as models crafted in formal logic languages, or looked only at those ontologies exquisitely crafted with care by experts, the idea that they don’t scale might be valid. Fortunately, we can develop ontologies and associate background knowledge in a variety of ways, especially when we’re dealing with a Web search and browsing application or an information processing task not requiring complete consistency in the knowledge base. Examples include using (a) ad hoc specifications such as Schema.org and grounding concepts in Linked Open Data cloud, (b) using domain specific community maintained resources (e.g., MusicBrainz for Music, IMDB for movies), and (c) dynamically generated domains models by tools such as Doozer++.

Semantics at the Web scale is gaining acceptance, and our ability to support it is increasing. In the case of Web search, several significant factors are helping semantics to improve it. First, each of the major Web search engines is creating domain-specific structured knowledge. Search engines can and are exploiting semantics via at least three methods.

The first is creating a concept base or object base of facts (entities and relationships) one domain at a time. For example, you can use MusicBrainz, which captures comprehensive knowledge about more than 550,000 artists and their creative works, as background knowledge for the music domain. Bing’s support for specific domains such as entertainment, sports, or travel is in part powered by domain-specific models and relevant background knowledge in a way reminiscent of MediaAnywhere. This use of domain-specific knowledge will continue to expand now that Google has acquired Freebase and the adoption of linked open data (a method of publishing structured data so that it can be interlinked) increases from current tens of billions triples (facts) and tens of domains by additional one or two orders.

Second is the recent collaboration between the three major search engine players in defining schema.org, which provides schemas or conceptual models for several common domains. The third method is content developers’ increasing use and support of microdata (a simple way to embed semantic markup into HTML documents) and RDFa (for embedding rich metadata within Web documents using RDF) to improve search results. This in turn entices content developers to provide more metadata or annotations and use relevant models to add semantics. Quick on search engines’ heels, social media (such as Facebook’s Open Graph protocol), e-commerce (BestBuy’s use of the GoodRelations ontology), and a wide variety of Web businesses and services are building a growing and synergistic Semantic Web ecosystem.

Search is No Longer the King of Web Applications

Prior to the era when semantics didn’t scale, search was the king of all Web applications. But the importance of search is highly overrated. Its best days were in the past. We’re in an era with significant growth in heterogeneity (social data, mobile-device-generated data, data from sensors inside, on, and around humans, and so on) and quantity (the rate of data creation has already surpassed our ability to store it). Simply needing access to data (which a search engine can index and return as a document) no longer serves our needs. We need knowledge and insights for decision-making as well as answers to our questions. Semantics plays a pivotal role in helping us build solutions to meet these requirements. Relationships are at the heart of semantics and semantic web, and, consequently, we can transition from focusing on keywords and objects, as we did with search, to focusing on relationships and richer abstractions, including events and experiences.

Pervasive Role of Semantics in Computing's Future

Semantics is being adopted on a wide scale in various scientific and some business domains that use W3C-defined Semantic Web languages and standards. For example, in the life sciences domain, we can find nearly 300 ontologies at the BioPortal. Even more impressive is the growth of structured data on the Web, called the Web of Data and best showcased by the Linked Open Data initiative, which surpassed 25 billion triples (facts expressed as subject-predicate-object in RDF) last year and tripling year over year. BioRDF, a collection of facts and knowledge in the life sciences domain from multiple sources, exceeded 5 billion triples last year. Note that the Web of Data isn’t simply data, but is structured and reusable information that we can utilize to consistently annotate or tag data, enabling better data analysis, which would be difficult to achieve via bottom-up processing of unstructured data on the Web.

Semantics plays a central role in Web 3.0 and beyond, and is becoming the driving force behind the future of computing for several reasons.

Semantics for Integration

Semantics, in the sense of archiving shared understanding and meaning, comes from agreement. Consequently, it’s long since it had a role in integrating data in heterogeneous syntax and structure. Increasingly, it also plays a role in integrating information about the same concept or object in different modalities and media — for example, to relate a person’s images with his or her descriptive information, or to correlate information about an event on social media with corresponding sensor observations. In coming years, semantics will be crucial to integrating objects that straddle the cyber–physical-social or physical–virtual divide.

Semantics for Intelligent Processing and Reasoning

Much attention in the past has focused on data and information search and browsing, in which processing complexity is reduced because of significant human involvement in interpreting the results. As we move up the information processing value chain from search and browsing to integration, analysis, situational awareness, and question-answering, information processing’s complexity increases significantly. Looked at from another perspective, information processing is moving from keyword-based to object-based processing and on to relationship and event-centric processing. As mentioned, relationships are at the heart of semantics and, fundamentally, computations will need to focus on modeling, processing, and exploiting them (http://bit.ly/rel-at-heart). In the case of formal languages, this will involve richer forms of integrated reasoning, incorporating inductive, deductive, abductive and fuzzy reasoning. Future advanced information processing will also not be limited to silicon-based processing; rather it will increasingly involve collaboration between humans and machines, with semantics-aware sensors as intermediates.

Semantics for Knowledge-Enabled Computing

The power of human reasoning comes not only from the sophisticated computing abilities our brains support but also from background knowledge and past experiences. Similarly, the application of background knowledge to improve information processing is rapidly growing — from the improvement of information extraction, natural language processing, and machine learning to better understanding and processing of social and sensor data. We can now apply domain-independent (related to time, space, and geographic concepts, for example) as well as domain-specific models of various complexities and comprehensiveness, such as nomenclatures, taxonomies, and ontologies, to improve information processing. The ability to utilize user- and community-created dictionaries (such as urbandictionary.com) and knowledge repositories (Musicbrainz, for example) to exploit structured information from unstructured data (DBpedia from Wikipedia) — and reuse such knowledge in improving computation — has added significant strength to semantic processing.

Semantics for Abstractions and Human Experience

Increasing amount of data generated by 5 billion mobile phone users (arguably the most important tool in human history, and many now with data connections), millions of social media users, and more than 40 billion mobile sensors, is finding its way to the Web. A single four-hour flight might generate 240 terabytes of data. How much of it is useful for a given human need? The ability to search this much data, however good, simply isn’t scalable in terms of the search results humans can review and absorb. What we want is a few nuggets of information or insights that we can act on. We care about broader and aggregate understanding of events, improved decision making, and getting answers to our questions. And we care about enhancing our human experience.

Semantics is a core component of developing abstraction mechanisms so that we can use computing to support perception and cognition. Semantic approaches support abstractions that convert low-level data and observations into the high-level symbolic representations that constitute our human perception and cognition. Semantics-empowered solutions can now analyze constantly streaming sensor or social data to tell us abstractions and events of human interests (such as icy roads, blizzard conditions, the need for intervention to save crops, chances that a movie will succeed, or the progress of a mass protest). My earlier article “Computing for Human Experience: Semantics empowered Sensors, Services, and Social Computing on ubiquitous Web" (IEEE Internet Computing, Jan/Feb 2010), explores this topic a bit further. You can find examples of such approaches at http://knoesis.org/showcase.

We have entered an exciting time for semantic computing. Semantics is changing contemporary Web applications, such as search, and will play a pivotal role in future computing that will span cyber-physical-social systems.

Citation info.: Amit Sheth, "Semantics Scales Up: Beyond Search in Web 3.0," IEEE Internet Computing, pp. 3-6, November/December, 2011

A companion presentation on Semantic computing in Real World: Vertical and Horizontal applications, withing Enterprise and on the Web.

New Tools for the rest of the 75% of the War

2011-01-13T12:28:00.000-08:00

“After losing 4400 American lives and spending 700 billion directly on Iraq war, we are now losing influence to Turkey and Iran… the real big picture is that this is the sign that US effort in Iraq was too focused on hard power, on military power, and not enough on soft power, on political and economic measures... “[Fareed Zakaria GPS, Jan. 09, 2011.]

Consider these statements from our military and defense leaders: (now ousted) General Stanley McCrystal: “It’s not the number of people you kill, it’s the number of people you convince” [P10]; and Secretary of Defense Robert Gates, our “ultimate success or failure will increasingly depend more on shaping the behavior of others – friends and adversaries, and most importantly, the people in between.” These point to the new imperative for determining the success of our wars. It is to use cyberspace to enhance the soft power to achieve what Henry Crompton, a veteran CIA covert officer said in an interview: "what we're seeing now in the tribal areas is a classic insurgency. And the way you counter that is working with indigenous forces. You use hard power in the beginning, the first 10, 20 percent, and then the next 80, 90 percent is what you might refer to as soft power..." [C08]. In fact, Pentagon now believes that understanding cultural dynamics is at least as important as weapons." [W10]

Robert Nye is credited with coining the term soft power [N90], and increasing number of world leaders (including those from China, India and Taiwan) have advocated the use of soft power as a preferred and more effective mechanism to resolve international conflict.

Mr. Robert Gates spoke of the need to enhance American soft power by "a dramatic increase in spending on the civilian instruments of national security -- diplomacy, strategic communications, foreign assistance, civic action and economic reconstruction and development."

Our ability to project soft power using information and cyberspace will involve:

(a) understanding the reaction of the affected population to the US’s efforts to fight the war

(b) winning the hearts and minds of people who must support our war on terror

(c) identifying and influencing terrorists and bad actors who (plan to) conduct irregular warfare using cyberspace and sensors (including social media); and understanding the use of false narratives disseminated over the internet by jihadists

(d) interdicting to prevent the use of cyberspace to recruit jihadists and terrorists; interdicting any undesirable uses in order to prevent the spread of vile messages and influence; modeling and executing campaigns to convince terrorists and bad actors to change

(e) Combine the above with hard power when persuasion fails

Arianna Huffington wrote an interesting post, arguing “we can now change the conversation to the impact of technology and social media on peace, not just on terror.” [H10] An interesting question is—which of these help deal with terror or promote peace? Education? Poverty alleviation? Better interactions between people and cultures? Democracy, civil and political rights, and open communications? [E10]

Addressing the above objectives point to a multidisciplinary research agenda involving social scientists, computer scientists, cyberspace technologists, policymakers and strategists. A small subset of the related items that the multidisciplinary team of researchers at the Ohio Center of Excellence in Knowledge-enabled Computing (Kno.e.sis) is interested in include:

· Socio-cultural-behavioral modeling and its use to study cultural dynamics

o personal identity, social behavior, logic, forms of reasoning, persuasion, etc as well as values and social roles

o study of national and group differences in cognition

· multidimensional, multifaceted situational awareness with support for

o aggregation, extraction, integration of massive amount of content, including increasing amount of dynamic and real-time content over Web, social media/network including human-in-the-loop sensing, multimodal/multi-level sensor (A/v, signal,..), and importantly SMS which are available to 5 billion mobile connections and widely used in developing countries

o spatio-temporal-thematic analysis and pattern mining leveraging high performance, high throughput infrastructure

o people-content-network-event analysis, understanding not just traditional enemies, but populations, developing a "Social Radar" [MM09] that can detect social signals [AS09], public perceptions, social sensing and tracking

I present some related ideas in this talk: Enhancing Soft Power (ESP)

[B09] John W. Bellflower, "The Soft Side of Airpower," Small Wars Journal, January 2009.

[C08] Interview with Henry Crumpton, Frontline, September 8, 2008.

[E10] Exploding misconceptions: Alleviating poverty may not reduce terrorism but could make it less effective, The Economist, December 16, 2010.

[H10] Arianna Huffington, Facebook, Twitter and the Search for Peace in the Middle East, Huffington Post, November 24, 2010.

[M09] Mark Maybury, Social Radar for Smart Power, MITRE Corporation, April 2010.

[My09] Gene Myers, Projecting power: QDR’s aerospace imperatives, Armed Forces Journal, July 2010.

[P10] Michael Phillips, "U.S. Steps Up Missions Targeting Taliban Leaders," Wall Street Journal, February 2, 2010.

[S09] Amit Sheth, Citizen Sensing, Social Signals, and Enriching Human Experience- IEEE Internet Computing, July/August 2009.

[W10] Sharon Weinberger, Pentagon Turns to 'Softer' Sciences - US defence research to focus more on biology, cybersecurity and social sciences to help win conflicts. Scientific American, April 14, 2010.

Rewards for an Educator

2010-07-15T10:00:00.000-07:00

Success of my advisees has always been a primary measure by which I evaluate my academic success. As our university President Dr. Hopkins noted in one of his weekly messages: "And you could understand why he was selected ( for the Trustee Award) from his comments (cf: video) --he focused his remarks on the success of his students, the contributions of his current students, and the accomplishments of the faculty he works with every day. "

I have reasons to be particularly proud and happy this summer, thanks to the hard and smart work of two of the best PhD advisees I have had the opportunity to work with so far--Meena Nagarajan and Satya Sahoo. I think we have proved Kno.e.sis' graduates are second to none. Let me explain.

The job market for PhDs has been tough. A Scientific American article reports that only15 percent of PhDs will get a post at the kind of research university where the nation’s significant scientific work takes place (and fewer yet will get tenure track positions at such places; perhaps situation for CS is slightly better than this overall statistics).

Satya Sahoo received and accepted a tenure track position in Case Western Reserve University’s School of Medicine (and also received multiple post-doc offers). Satya had a single minded focus on an academic option and did not consider an industry option. Here are some of the reasons why this makes an exceptional outcome (besides the excellent terms of the offer itself):

Given the dismal state of budgets/deficits, this appeared to me as the worst academic recruiting season in my 25 year career.
In brand conscious academia, it is possible to recruit graduates from "top" ranked universities/departments, so recruiting from a lesser known/ranked university is a rarity (our department made US News rank of CS departments just this year). Satya’s competition included graduates from top 20 schools, one candidate that had 10+ years of post-PhD experience in top industry and government research labs as well as academia, and those with post-doc experience.
Satya's offer from the School of Medicine also demonstrates Kno.e.sis' success in interdisciplinary research and training.

In like manner, Meena gave me a reason to rejoice as she received three offers (she also was one of of the two finalists for a post-doc with a renowned faculty from a top 5 ranked school, but it was too late to entertain that possibility). Job offers were for a visiting assistant professorship from a highly respected IS department, highly competitive (5-8% selection rate) and prestigious NSF CI Fellows to work with an exceptional faculty in her field, and a Research Staff Member position from IBM Research - Almaden. Most of my colleagues in the know would agree that two of the most attractive industry research options are Microsoft Research and some groups in IBM Research- Almaden. What makes IBM Almaden an attractive option is the ability to spend 50% of time on topics of researcher's own interest, exceptional links with academia (many are familiar with Google's 20% discretionary time) and the strength of the offer (the best terms for a fresh graduate I have seen so far). I know of cases where post-doc from a top department accepted a post-doc position at the same IBM research lab.

What is the secret sauce that makes Kno.e.sis PhDs succeed in getting such attractive options?
Research results demonstrated through publications is a necessary component, but by far publications alone are not sufficient in these more competitive times. Here are some of those components:

Collaborations and networking, partly facilitated by internships at top industry and government labs (e.g. NLM and Microsoft Research for Satya; HP Labs, IBM Research, UC-Berkeley and Microsoft Research for Meena), often resulting in reference letters from renowned researchers (e.g. Prof. Carole Goble from U-Manchester and Dr. Oliver Bodenreider from NLM for Satya, Prof. Marti Hearst of UC-Berkeley for Meena).
Communication, mentoring, teamwork, national and international travel, proposal writing and project coordination/leadership roles
Outcomes and activities that are unusual at PhD or even post-doc level (Satya: organizing international workshops, W3C activities; Meena: keynote (almost unheard of prior to PhD completion), chairing and participating on panels; Both: serving on 7-10 international program committees including mainstream conferences).
Kno.e.sis environment and our emphasis on work culture (here is the open secret: Video, Article).

Some more details of extraordinary student achievements can be found here: student accomplishments flyer, student accomplishment flash

Meena and Satya will soon join me as colleagues and serve as inspiration to current and future Kno.e.sis students...and I look forward to watching their success and get back to the job of mentoring students in anticipation that more of them will join their ranks.

Computing for Human Expereince: Semantics empowered Sensors, Services, and Social Computing on ubiquitous Web

2010-01-01T07:45:00.000-08:00

When planning for my keynote at 2008 Asian Semantic Web Conference, I decided to think about and focus on what might drive my future research. Besides mainstream work on Semantic Web, we at Kno.e.sis are doing a good bit of work on Semantics-enhanced Services and Cloud Computing, Semantics-empowered Social Web, Semantic Sensor Web-- the talk leverages these research thrusts. However, more than these technological perspectives, what excites me more is the evolution along the lines of sensory-perception-semantics dimensions and and disappearance of physical-digital-conceptual divide. As with most of my research, applications drove the thought process-- in this case trying to envisage applications. More recently, I was invited to align my Web Semantics and Services Column with the Jan/Feb 2010's IEEE Internet Computing special issue on "Internet Predictions" that is guest-edited by Vint Cerf and Munindar Singh.

Version of the article on Computing for Human Experience
NCSA Director's talk (2010 - talk and video)
Earlier version: ASWC 2008 keynote talk and video (Jan 2009)

January 1, 2010

Two interesting issues of IJSWIS

2009-11-11T12:26:00.000-08:00

Those interested in this blog are likely to be interested in:

IJSWIS Special Issue on Linked Data (with free access to lead article by Bizer/Heath/Berners-Lee)

IJSWIS Special Issue on Scalability and Performance of Semantic Web Systems

Amit

Citizen Sensing, Social Signals, and Enriching Human Experience

2009-06-20T11:21:00.000-07:00

As I watch a media outlet such as CNN playing a video from YouTube based on a link their reporter/producer sees on a twitter feed [1], I am able to find the same content directly, reported by citizen sensors. The next small step is being able to study social signals through analysis of growing sources of social media and citizen (human-in-the-loop) sensors--to help us study the pattern of discourse, and variations along spatial, temporal and thematic dimensions. For example, we can ask, what are the important topics of discussions and concerns in different parts of the world on a particular day, or how different cultures or countries are reacting to the same event or situation (eg Mumbai Attack), or how a situation such as financial crisis is evolving over a period of time in terms of key topics of discussion and issues of concern (eg subprime mortgages and foreclosures, followed by troubled banks and credit freeze, followed by massive government intervention and borrowing, and so on). Combining the citizen sensors with fixed and mobile sensors (also called multilayer sensing) will lead to even more compresensive information on events and insights into situations. A paper [4] gives more details while a talk [2] and an article [3] give overview.

From computer science research perspective, it is exciting to see that a number of things Kno.e.sis researchers are working on are coming together -- especially the topics of semantics-enabled services, sensor and social computing:

aggregation and integration of social data (coordinated/led by Karthik Gomadam, and leading to Twitris)
analysis of user-generated content (by Meena Nagarajan [5])
extraction/creation of a domain model from Wikipedia or similar community authored content (by Christopher Thomas)
semantic sensor web (by Cory Henson)

From social perspective, here is a thought that comes to my mind. Countries may have authoritative (or dictatorial) regime, but it is becoming increasingly impossible to contain democratic aspiration and expressions of their citizens when they have access to social media. Only a couple of countries have effectively managed to be exceptions-- how long will they remain exceptions? Social scientists have started to ask such questions. Lately I have seen lively discussion on journalism and new media on TV news programs. I see rewarding opportunities for collaborations between computer scientists and social scientists.

Amit

[1] e.g., http://twitter.com/#search?q=%23iranelection
[2] Semantic Integration of Citizen Sensor Data and Multilevel Sensing: A comprehensive path towards event monitoring and situational awareness, February 17, 2009.
[3] Citizen Sensing, Social Signals, and Enriching Human Experience- IEEE Internet Computing, July/August 2009.
[4] M. Nagarajan et al., Spatio-Temporal-Thematic Analysis of Citizen-Sensor Data - Challenges and Experiences, Tenth International Conference on Web Information Systems Engineering, Oct 5-7, 2009, Poland, to appear.
[5] What are people talking about, Why people write, How people write: Meena Najarajan's research

Maturing semantic technologies for information systems

2009-06-02T17:38:00.000-07:00

In 2002, I organized one of the earliest workshops on Semantic Web. It was nice to see that Reasoning Web 2009 [1] has chosen to focus on something I wrote in the Amicalola Report [2].

"Researchers and practitioners in the database, information systems and internet fields over the years have made significant progress towards the building of solutions that involve such systems for a wide range of application domains. In doing this, solutions necessarily concentrated mainly on syntax as the readily available unifying formalism for representation and structure, rather more than on the broad variety of semantics involved. One of the recent unifying visions is that of Semantic Web, which proposed semantic annotation of data, so that programs can understand it, and help in making decisions. Researchers have subsequently seen the value of using semantics to understand information and decision making needs of humans, so that data and human's needs can be semantically intermediated. The scope of semantics-based solutions has also moved from data and information to services and processes. Semantics has not been new to the database and information systems community. Semantics in data models was studied intensively in the 1980s, and applied to problems such as query processing, view management, schema transformation, schema integration and transaction processing. Semantic heterogeneity and interoperability have been studied as part of all major information systems architectures during the last three decades, including federated, mediator, and information brokering architectures. Many projects in information interoperability and integration have addressed semantic heterogeneity. In addition to the study of semantics, we believe there are several important areas of expertise within the database and information systems community developed as part of successful database management, information interoperability, information retrieval and workflow management systems that will be important to build large scale, high performance and practical Semantic Web and Enterprise solutions. A partial list of relevant technology for, e.g., semantic web services includes transaction management, query planning and optimization, distributed scheduling, exception handling, dynamic changes and adaptation, and security."

The corollary is that much more still needs to be done in Semantic Web to make its technologies as relevant and ubiquitous as a database management system, a workflow management system, or an IR system.

[1] http://reasoningweb.org/2009/Objectives.html

[2] http://knoesis.wright.edu/library/resource.php?id=00175

CI Fellows

2009-05-22T06:24:00.000-07:00

For students who have completed their PhD between 5/1/08 and
8/31/09 and are interested in a postdoc position at Kno.e.sis
can check out my posting at: http://bit.ly/U4R72

Relationship Web

2008-10-08T15:18:00.000-07:00

Relationships are at the heart of semantics and we envision a Web of relationships
to relate multimodal content across the Web. Following the first
generation of Web content access characterized by keyword driven document-retrieval, and the more recent process in entity awareness, we believe this third generation of relationship centric framework will support insight elicitation, semantic analytics and knowledge discovery over Web resources not possible so far. Relationship web incorporate the vision of trail blazing outlined by Dr. Bush in 1945!

More in our Internet Computing article: Relationship Web Blazing Semantic Trails between Web Resources (also available here).

What is Semantic Computing?

2007-10-24T08:16:00.001-07:00

I have seen broadening of the scope of where "semantics" is applied. In mid-80s to early 90s, we had explored semantics for data(base) modeling, interoperability and integration (for examples of our work: see [1]). Others also looked at similar problems as well as use semantics in AI, programming language and several other areas in Computer Science. As Web started to become the global information system, some of us explored semantics (esp use of ontologies) for searching and integrating Web accessible information (see [2] for our efforts that predated coining of the term Semantic Web). All these activities then came under the umbrella of Semantic Web after TBL coined the term. When Philip Sheu organized the Semantic Computing conference last year, it seems the scope of semantics as it related to computing further expanded. And the question was asked: how do we define Semantic Computing?

Phil Sheu in ICSC2007 cfp described it as:

"The field Semantic Computing applies technologies in natural languageprocessing, data and knowledge engineering, software engineering, computer systems and networks, signal processing and pattern recognition, and anycombination of the above to extract, access, transform and synthesize the semantics (contents) of multimedia, texts, services and structured data."

Here is my take:

Semantic computing is a vision of computing based on semantics shared between machines and people. It supports and exploits intrinsic, intended, and emergent meanings (content) in all aspects of computing, encompassing programming, algorithms, information management, and human interactions within devices, as part of communications, and across the Web. Semantics involves the use of formal descriptions, languages, and models, often encoded in metadata, knowledge, and representation of agreements (as in ontologies) to capture the content of multimedia, texts, services, and structured data so that it may be extracted, shared, synthesized and transformed. Semantic techniques foster the development emerging forms of computing, such as semantic Web, and entirely new forms, such as bio-inspired computing, as well as enhance traditional techniques of information retrieval, management of data (including multimedia and multimodal) and artificial intelligence (e.g., natural language processing machine learning, and computational intelligence), leading to more efficient and scalable information processing and higher-quality computer-human interaction.

Perhaps there is a better way to state what I intend to say in the above description of Semantic Computing. I am sure there will better attempts to define this.

[1] mid 80s to early 90s: So far yet so Near, Schematic and Semantic Similarities between Database Objects
[2] early 90s to about 1998: Semantic Information Brokering, InfoHarness, InfoQuilt, OBSERVER

SAWSDL becomes a W3C recommendation

2007-09-04T12:59:00.000-07:00

August 28, 2007 is a special day for our research group (METEOR-S project, LSDIS lab and Kno.e.sis Center). On this day, we saw the culmination of an activity I started a little over three years ago. In July 2004, we wrote a white paper"WSDL-S: Adding Semantics to WSDL - White Paper". In November 2005, in collaboration with IBM, we made a W3C member submission Web Service Semantics - WSDL-S. Our presentation at a W3C workshop held at Innsbruck received support from a number of key members of Semantic Web Services community that includes the OWL-S, WSMO and SWSF groups, which then led to the Semantic Annotations for WSDL (SAWSDL) Working Group in which Kno.e.sis (at Wright State University - a W3C member) and LSDIS (at the University of Georgia) also participated. After timely and extensive contributions of working group members (and especially the groups chair and editors- Jacek, Joel and Holger), the W3C declared Semantic Annotations for WSDL and XML Schema as a recommendation.

Interested readers can find more details in the Internet Computing column I wrote with my former advisee and a key early technical contributor to our work in this area, Kunal Verma (Semantically Annotating a Web Service), by giving this tutorial at Semantic Technology conference (Using SAWSDL for Semantic Service Interoperability), or by playing with implementations and test suites.

So what's next? My own group has defined SA-REST (more on this soon), and Charles Petrie and I have started (with healthy dose of encouragement from Dieter Fensel) a W3C SWS Testbed Incubator Group to develop the awareness and further agreements on post-SAWSDL issues that we have to address before more developers will find Semantic Web Services ready for the prime time.

Marrying Social Media with Semantic Media

2007-05-17T12:11:00.000-07:00

This is a recreation of the original blog at LSDIS lab, UGA, before I moved to Kno.e.sis Center, Wright State University posted on 12/03/05.

A lot has been written about social networking and we can see a number of success stories around this phenomenon. Flickr exemplified taking social networking to media (photos in this case), and hence the emerging focus on social media (see The Flickrization of Yahoo). The power of users playing the roles of authors and editors is undeniable, but the lack of organization it creates is an antithesis to how Yahoo! organized the early web around a directory (that we can now replace by ontologies) and human cataloging (which we can to a good extent replace by automatic and ontology-assisted semi-automatic semantic annotation). Is it possible to get such organization back into naturally unconstrained and entropic collection of stuff (in this case media and associated human given tags without the supporting nomenclature)? I think it quite is, and the current approaches for and technologies behind the Semantic Web provide the most promising paths.

Indeed there is already an independent set of activities in developing semantic multimedia and semantic media. What can we attain by marrying semantic media with social media? Quite a lot. What we get is a multiplicative outcome that combines social media organized to make it much easier to search, integrate and exploit, with semantic media enhanced by the power of people. This will also make it much easier to integrate all shared information whether in text or in digital media of any format. It will be easier to serve up multimedia and multimodal applications. And it will unlock a lot of untapped potential with targeted advertisements and refined personalization that underlying semantic infrastructure could provide, significantly enhancing the business potential associated with shared media.

Here is just one enticing example. You are looking a photo of a church in Innsbruck. And you want to sell an advertisement that takes you on a landing page on Priceline that has a flight from US gateways to Munich. (I had a chance to checked out what types of Google ads are servied today in this context of Innsbruck, and it did not seem to me that anyone would likely close a sale, and especially a sale with a high value). What will tell you that the nearest airport with transatlantic flights is Munich, do not bother to try to sell a ticket to Innsbruck especially from a non European distination, and how to get the specific page on Priceline for North America to Munich flights? Context, semantic metadata, ontologies that have a load of real world and factual knowledge, and a set of rules.

Some of the lovers of information retrieval and dumb keyword search would tell you all this is too complex, not scalable or not maintainable. It is not. Contact me if you want to see a demonstration of how we are working to realize this promise.

Can semantics make Web Services more useful for businesses

2007-05-17T12:09:00.000-07:00

This is a recreation of the original blog at LSDIS lab, UGA, before I moved to Kno.e.sis Center, Wright State University on 10/02/05.

There is a lot of discussion on what Web services can or cannot help business achieve [0]. Earlier simplistic views of Web services as silver bullet are being replaced by more temperate views, as exemplified by Andrew McAfee's recent article [1]. This posting is primarily focused on sharing some thoughts related to [1].

Prof. McAfee offers two insights: "Web services allow construction of modular and interchangeable building blocks software" is happening, but "how companies collaborate and compete" is not happening. He also observed that "the application-integration challenges that remain unaddressed by Web services are the really difficult ones and can only be overcome by the work of managers and leaders, not technologies and consortia".

Prof. McAfee is right on mark when he says "Web services, however, will not create this world," but I disagree when he says "nor will any technology on the horizon". My main argument is that if we add semantics to the mix, I think we will see more progress than what Prof. McAfee has seen so far. I will give a technical (computer scientist) view on these points that is validated by real-world deployments by the company I co-founded (Semagix), some collaborations in bioinformatics between the LSDIS lab and the biologists at UGA, and by some interactions with industry collaborators, mainly at IBM.

A different perspective inteoperability

Interoperability is the key to collaboration, integration and interoperability (between application and application, as well as human and application/system). It comes at four levels: infrastructure/system, syntax, structure/representation and semantics [2] [3]. I feel that this dimension is more meaningful when looking at technical issues compared to the three levels of transport, payload, and process used in [1].

XML and Web services deal with infrastructure, syntax and some representational issues. Concurrently, in several scientific and business domains (e.g., biology [OBO], health care [OpenClinical], risk & compliance [4], digital content applications on mobile networks and so on; also see- Why are we still pushing Semantic Web?), we are finding increasing success in developing and exploiting ontologies. These ontologies are the key enabler of semantic technologies, and embody human-agreement such that machines can interpret (mimic, reuse or in a rather limited form, "understand") this agreement to replace some of the human interactions (when such agreements exist). While I will spare the details, ontologies related semantic techniques and technologies (e.g., disambiguation with approximate/fuzzy matching, probabilistic relationships, etc.) enable semantic integration that can help deal with inconsistencies and other problems pointed out by Prof. McAfee. And while the difficulty in technological solution in Prof. McAfee's IBM case study was attributed to "Midrange systems were simply too complex", human bodies or biological description (e.g., pathways) are at least as complex, and yet it has been possible to develop very useful ontologies in these areas. Interestingly, parts of RosettaNet PIPs (the subject of the Case Study in [1]) has been represented as an ontology by us and others [RosettaNet ontology].

Using semantics to the mix

The need for painstaking work does not disappear when developing ontologies, but but many tools and techniques are available to build and maintain populated ontologies. Furthermore, ontologies defined in formal representation languages such as OWL are highly sharable and reusable form of knowledge representation. Ontologies, policies and rules provide a medium for capturing and reusing the knowledge and experience gained from prior integration efforts and negotiated agreements, thereby leading to greater level of automation at the semantic level,. This can enable companies such as IBM to reuse its experience for more efficient interactions and integration in the future.

To capture the breadth of issues necessary in modeling complex systems, our approach to semantics for Web services/processes consists of four components: functional (the capabilities of a business), data (how to talk to it), non-functional/QoS (policies, rules, ontologies capturing domain knowledge from previous experiences and business goals), and execution (issues of run time behavior, eg errors/exceptions) [5]. This gives a framework to deal with the complexity, capturing some of the semantic level inconsistencies, and dealing with run-time issues in application-application interactions (implemented as part of Web processes).

Commercial successes of Semantic Web Services/Processes are not evident because we are in a rather early technical and technological development process for this emerging technology. Nevertheless, evolutionary approaches to extend WSDL (the current standardard for Web service description) with semantics (e.g., WSDL-S) have been conceived, and I predict that we will be able to address the limitations identified by Prof. McAfee in next 3-5 years. I also expect good case studies from early adoption to be available within next 12 months.

[0] S. Staab, R. Benjamins, C. Bussler, D. Gannon, A. Sheth, W. van der Aalst. Web services: Been there, Done that? IEEE Intelligent Systems, Trends & Controversies, 18(1), Jan/Feb 2003.

[1] A. McAfee: Will Web Services Really Transform Collaboration http://sloanreview.mit.edu/smr/issue/2005/winter/16/

[2] A. Sheth: Focus on Interoperability in Information Systems: From System, Syntax, Structure to Semantics, 1998

[3] A. Sheth: Semantic Meta Data For Enterprise Information Integration

[4] A. Sheth, Enterprise Applications of Semantic Web: The Sweet Spot of Risk and Compliance

[5] A. Sheth, Semantic Web Proicess Lifecycle: Role of Semantics in Annotation, Discovery, Composition and Orchestraction, 2003 Abstract, Presentation

Semantic Web: A different perspective on what works and what doesn't

2007-05-17T12:08:00.000-07:00

This is a recreation of the original blog at LSDIS lab, UGA, before I moved to Kno.e.sis Center, Wright State University posted on 01/13/05.

Peter Norvig's view published on AlwaysOn [1] seems to be colored by a decidedly web search engine perspective. If we start looking at Enterprise Semantic Applications (semantic applications developed for targeted enterprise/corporate/scientific/engineering user base, whether the data comes from the enterprises, or is a syndicated/licensed content, or is a open Web content), you can start to see some exciting alternative perspectives and realties. Let us review Peter's belief that the point that there is not enough RDF and SW content. This is growing at an extremely rapid pace (I am sure others will put out numbers; and there are specialized search engines such as Swoggle that focus on Semantic Web content with rapidly growing size of indexed documents: check ). More importantly, the promise of Semantic Web is closely tied to having the tools for semantic annotations of heterogeneous content, i.e., create semantic metadata automatically. This is much easier to do when you have high quality domain ontologies that bound the scope of automatic extraction. And I really do not see content suppliers putting the metadata in (as we do not see Web page authors using metatags), or at least this will be optional and just one form of input. Instead, metadata will be created with respect to potential use (e.g., there are some definite concepts when we deal with WorldNews, USNews, TechnologyNews, and so on). Commercial technologies (example) can process millions of pages per day and extract semantic metadata, and all these can be represented as RDF (and that is a good idea because of the benefits esp. for high end semantic applications such as analytics). Granted this is hard to do on a panWeb scale where you have no single domain or even a limited set of domains, and huge diversity of users who may need to see the content from different perspectives. Even here, I believe much can be done, but it will take a little more time - maybe 3 years.

Let me next respond to the comment about ontologies. There are many cases within Enterprises and even for consumer applications (e.g., see examples; also expect to see use of Ontologies soon by Amazon and this types of companies). Rather than focusing on common sense or general purpose ontologies, the immediate future is with domain and task/application specific ontologies (latter are appropriate for enterprises, eg., SOX or Anti-Money Laundering (AML) applications). This is done now successfully, with line of business applications (e.g., AML application deployed at one of the largest banks). These types of ontologies routinely have millions of instances (look at SWETO, NCI ontology, GlycO, as very different types of examples: think of SWETO with about a million instance as a poor sample of what is being done for real world Enterprise class Semantic Applications representing only a 10th of population and with diffused focus that a typical enterprise deployment; with 767 classes, think of GlycO has well-focused scientific ontology developed by domain experts; NCI ontology involves more community involvement). Additional thoughts on SW adoption are in my earlier item on this blog.

Postscript: Just before posting this, I stumbled across Danny Ayer's views in response to Peter Norvig's item; I agree with those view almost completely. After my initial posting, I came across another set of comments by Tim Finin.

[1] Original post on AlwaysOn is no longer available, but here is what appears to be a verbatim copy.

Why are we still pushing Semantic Web?

2007-05-17T12:03:00.000-07:00

This is a recreation of the original blog at LSDIS lab, UGA, before I moved to Kno.e.sis Center, Wright State University posted on 01/07/05.

This was the question a panelist asked at the W3C Advisory Committee meeting that I attending at the beginning of December 2004. In other words, the panelist and others discussing this question were wondering, why is it taking so long for the industry to get it (its importance)? Or that, by now, we would have expected it to have seen much wider adoption, a clear indication that the Semantic Web is here for good, transforming the Web into its next logical incarnation.

The essence of my comment at that time was that the rate of progress is quite robust and pervasive, and there are prominent signs that the Semantic Web is not just a fad, that this time, semantics as applied to information (which predates the Semantic Web as defined today) is indeed likely to affect many businesses in not-too-distant a future, and even common Web user in intermediate future. Here is an extended perspective on the adoption of the Semantic Web, which also incorporates a nice dinner discussion that some of the Semantic Web technology/product vendors (who are members of the W3C) had with the W3C Semantic Web team members (Eric Miller and others).

Research

Although funding from NSF, DARPA and the premier funding agencies have now waned, DAML program gave excellent and timely start to the Semantic Web research in the US. The funding initiative moved to Europe with Framework V, and is firmly entrenched with Framework VI. The number of new conferences, conference attendance, sessions related to the Semantic Web in older and more established conferences, number of published papers and new scientific journals devoted to the Semantic Web (such as Web Semantics, Semantic Web & Information Systems, and Applied Ontology) all point to broad and increasingly entrenched interest in this new area.

Standards

One of the nicest things that have happened to our area is timely standards activity. Note the emphasis on "timely", as it is helpful to have basic standards before the area matures and before industry interest peaks, reducing the chances of clashes between the entrenched interests. Not having activities being taken to competing standards bodies, as is the case in Web Services area, helps too.

Technology and Products

One of the most exciting things to have happened in our area is the number of technologies commercialized from academic research (Taalee's MediaAnywhere A/V Semantic Search and Semagix's Freedom from University of Georgia's SCORE technology, Network Inference's relationship with University of Manchester, Ontoprise's relationship with Karlsruhe, to name a few). Now, at least twenty vendors claim to use or support Semantic Web technologies, and the list is growing quite rapidly. And perhaps most importantly, scientific and business communities are building targeted (i.e., with clear purpose) and large ontologies at an impressive pace.

Industry Recognition

The informative panel at the W3C 10th anniversary celebrations (http://www.w3.org/2004/09/W3C10-Program.html) on the "Web of Meaning" illustrated how the thought leaders and industry executives buy into the vision of the Semantic Web. Panelists Tim O'Reilly (O'Reilly Media, talk) and Bill Ruh (Cisco Systems, talk) presented a fairly encouraging perspective on how Semantic (Web) technologies are needed for key applications, such as Regulatory Compliance, B2B Exchange, Workflow and BPM, and Business Intelligence. What is interesting is that some of these are "selling aspirin" rather than "selling vitamins", something that does better in low to moderate economic growth environments.

I would add several other fields of rapid adoption, including life sciences (see the W3c workshop in Semantic Web for Life Sciences), bioinformatics, healthcare, content management, national intelligence and homeland security. Just look at the number of large ontologies that cover the broad range of schema size, descriptionbases (instances) and expressiveness of representation, developed by community or a small number of domain experts, that are now being put to practical use. Some illustrious examples are NCI Cancer Ontology with over 17,000 concepts, or GlycO ontology for complex Carbohydrates with 767 Classes that is up to 11 levels deep and utilizes all expressive power of OWL, or ontologies with over 10 million instances developed for enterprise semantic applications using Semagix Freedom. Researchers interested in finding ontoloiges to play with can consider TAP or SWETO that are based on real-world facts, or get their hands on software to generate synthetically generated ontologies.

At the industry events, such as those organized by TopQuadrant and MITRE, or the user group initiated events, such as those for the US Department of Defense or the Life Science Community, 100 to 300+ people have shown up, which indicated fairly high level of industry and user group interest.

Industry Deployment and Early Successes

Since some very early deployment examples that were discussed at the WWW2004 Developer's day, there are now increasing number of examples of deployments both in Enterprises (e.g., see my KMWorld talk) or for more 'common' web users. It is this topic what garnered the main attention during our dinner discussions (mentioned above). One exciting observation that came up is the stealth inclusion of the Semantic Web technologies in applications. Eric Miller gave the example of Creative Common's use of RDF (also see Shelly Parker's earlier article). This is an example of simpler SW applications involving embedding license metadata and validating it so millions of content items would in essence be using at least limited Semantic Web technology for enforcing licenses! Another example is that of semantic annotation of syndicated contents and Web Services (e.g., the WSDL-S semantic proposal (early draft, currently being revised in an academic-industry partnership) and corresponding tools (e.g., MWSAF and ASSAM) for annotation of Web Services). Such applications can quickly lead to a wide spread and pervasive use of RDF in a fairly short time. What is interesting is that some of the applications are not being deployed by early adoptors; instead the SW technologies have been part of the pain killer types of main-stream IT applications and solutions (such as Anti-Money Laundering, compliance and risk management)! Enecdotal successes are starting to come. For example, a compliance related semantic application(implemented with a semantic technology platform from Semagix) is live at one of the largest banks in the world in the line of business. And I have heard of companies such as Amazon inserting ontologies in their main stream applications, so we can expect to see large scale consumer centric applications exploiting essential components of Semantic Web in the near future.

Final thoughts

One perspective that some in the community, particularly Tim Berners-Lee-TBL, seem to promote it that Semantic Web is "not interesting in the smaller scale". As more and more things connected by a "semantic way" it becomes more and more important. This makes sense from the perspective of global scale Web and non-enterprise applications. But from an industry perspective, I believe Semantic Web is equally interesting at the intra- and inter-Enterprise scales, and for Enterprise applications. This view is the same as the adoption and importance of Web technologies in Intranets. If at all, given the ability to constrain or limit the domain, deeper domain semantics can be put to use, agreements to build ontologies can be reached faster, industry specific metadata standards can be readily used, and facts and knowledge to populate ontologies can be obtained more easily. Today's enterprises have millions of documents, and access to massive amounts of high-quality or targeted syndicated contents and data (e.g., through Lexis-Nexis, ChoicePoint, NewsML and RSS News Feeds, and so on). The ontologies developed to support targeted enterprise scale Semantic Applications are currently exploiting ontologies with millions to tens of millions entity and relationship instances. And yes, the promise of scaling these Enterprise and industry scale islands by interconnecting them (and achieve what TBL called network effect) exists anyways.