| Hi Bradley,
Thank you for this opportunity. This was just the sort of question that I was looking to answer and that has helped me revisit my own client base --- clients, in my personal experience, are not as savvy as they otherwise might become with this Ontology Summit effort, about the process and need for spending direct money on "ontology development" as a bona-fide and separated component of product development, so we have to build-in the costs into the commercial world into the software product development lifecycle for our clients without getting into the onerous process of education (where clients tend to leave and find other vendors that do not explicate the complexities). Even with companies that should know better, employees or their consultants are in large part unaware of the ontological status of their software commitments and that is always dangerous water to tread in a client-vendor discussion ... with those notes, here are my answers:
1) I would like to add the prefatory note that our specific *clients* at the C-executive level do not care about Ontology per se --- they care about results. So our approach has been to factor the cost of ontology development through the R&D investment in semi-automated tools to give our company, VivoMind Research, a competitive edge over others in order to retain and secure our client base: in other words, the ability to rapidly build ontologies is important to the success of our projects because we can do it faster to get to results faster. As a recent example, we built an 800,000 term ontology within a week using our tools without which the client would never have been able to complete the analytic tasks needed. Our client sees ontologies as a part of the software development process and in particular, sees "models" as a part of the *analysts* process --- so there is a divide between these views.
2) Yes. We can point to our work with the DOE on the BREM (Beyond Rare Earth Magnets) program where we do machine-reading of Science-and-Technical journals to extract and identify compounds, processes, relationships, functions and metrics that could help the DOE scientists gain an advantage in rpaid understanding of which data points are relevant in building the next-generation of magnetic materials. We are continuing this work as it provides a 600x force multiplier from having to manually read and extract the same data so what it takes a human 50 hours to do, we do in 2-3 minutes. We are interested in further expanding this capability with interested partners.
3) We analyze massive collections of PDF (millions of documents at a time) and use intelligent algorithms to reverse-engineer PDF structures into machine-processable databases that are then analyzed using our NLP that fit an object model that is stored in Oracle DBMS systems.
4) The most exciting initiative within our own project is the integration of logic-programming with JAVA and SCALA to provide super-scalar performance beyond thread-based environments. We have also noted that a large part of ontology synthesis, another exciting area, depends on high-speed hypothesis formation with evidence-collection from open-sources (such as Wikipedia and other data) to confirm or disconfirm the plausibility of a hypothesis as to whether a give ontological structure proposed by the system is functionally workable against data (i.e. that results produced are not inconsistent).
5) We call our killer-app a Proto-Ontology Synthesizer --- we use the term "proto" ontology because a human is needed in the feedback loop along with reference data to build a final ontology.
6) Big data analysis needs to be dealt with using fundamental principles and, today, Ontologies are way to capture, curate, manage and exploit those principles.
7) Ontologies themselves can be caste into an ontology of ontologies --- ie. a meta-ontology of systems of ontologies. These structures enable us to avoid automation-surprises (when systems fail to behave as expected). It also allows us to inter-operate systems whose internal structures are heterogenous.
8) Ontologies are emerging as a means for inter-disciplinary collaboration by requiring that implementors, users, designers and other stakeholders synchronize together in a project based on the ontology specifications which become the grounding for the use-cases.
9) Yes. Ontologies cost a lot of money to build (for example, the money spent on the Cyc ontology is quite public) and for many domains, such as SNOMED and others, the costs are also public. The costs run into the tens of millions of dollars. Therefore, automatic solutions are a de-facto need. In our own work we have saved the customer about 7 million dollars by using automated ontology synthesis.
I hope this helps.
Thanks,
Arun Majumdar CEO, VivoMind Research LLC
On Apr 4, 2012, at 8:36 AM, Bradley Shoebottom wrote: 1. Why are ontologies important to the success of your project? They bring order to what would be otherwise an incomplete PDF search query through technical publications. 2. Can you share any specific success metrics from the application of ontologies? An implemented ontology in a search interface improved new Tier 1 tech support agents (less than 1 year on the job) searches for information by 78%, 3 out of 4 times. It also was a performance support tool for “slow” searchers. Ontologies were not as useful for employees with more than 1 year on the job because they have learned how the information is structured, its key terms etc. Tier 2 tech support agents had improved search times whether new or older, in part because they did less searching for info. Their knowledge often mapped closely to the ontology terms so they were able to use it more easily. Their benchmark PDF search time was also quite high since they did not search PDF tech pubs as much as the Tier 1s. 3. How does your project fit into the Big Data picture (if at all)? We deal in the millions of pages of unstructured XML to search through for product technical documentation. We need to be able to connect from software engineering design feature IDs and descriptions to topics that discuss the concepts, tasks and reference materials related to each design ID. 4. What are some of the exciting initiatives within your project that you've seen - who should we watch / why? SmartLogic, Semaphore product http://www.smartlogic.com/: a simple ontology model (zThes) but good enough for large organizations and has workflow to get new instances/classes into the ontology. Ontotext: KIM search interface. http://www.ontotext.com/kim Offers various search types including facets, patterns and structured search 5. What is the killer app using your project that you're hoping to see emerge over the next few years? Semi-autonomous ontology and term list builder. Ontology builder has to be able to suggest object properties and data properties. 6. Why are ontologies essential for modeling systems? Because often staff are only experts in their part of the system and a system ontology helps show the linkages to other parts of the system. 7. Why are ontologies essential for integration / federation of systems? They are faster to create that SQL tables and more flexible. You can reuse the ontology as needed. 8. How do ontologies help address Interdisciplinary Collaboration? You can include definitions, restrictions etc. so that others can explore the model and discover what is there. They can also link technical writers to the software engineers better because the tech writers do not have to rad lengthy design specs or ask as many questions because the model contains how the feature relationships occur. 9. Has anyone quantified and is willing to share any of the costs associated with the lack of ontologies in {modeling, integration, federation, collaboration}? From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Ali SH
Sent: April-03-12 6:17 PM
To: Ontology Summit 2011 discussion
Subject: [ontology-summit] Quotes for Ontology Summit 2012 Dear All, I'm writing to this group to solicit some human quotes for the 2012 Ontology Summit. I've written a number of you individually (especially those that presented applications), but wanted to open the floor to all 2012 Ontology participants. I am writing a number of little stories / narratives to try to get more publicity for the summit, any good story needs quotes from people.... While I'm focusing on highlighting a number of the projects that were presented at the summit, anyone who feels like they have a good insight to the questions below should feel free to chime in. One or two lines at most per question would be of tremendous help. It would be especially helpful if the style of your response is as though we were talking to one another (conversational, not academic). 1. Why are ontologies important to the success of your project? 2. Can you share any specific success metrics from the application of ontologies? 3. How does your project fit into the Big Data picture (if at all)? 4. What are some of the exciting initiatives within your project that you've seen - who should we watch / why? 5. What is the killer app using your project that you're hoping to see emerge over the next few years? 6. Why are ontologies essential for modeling systems? 7. Why are ontologies essential for integration / federation of systems? 8. How do ontologies help address Interdisciplinary Collaboration? 9. Has anyone quantified and is willing to share any of the costs associated with the lack of ontologies in {modeling, integration, federation, collaboration}? I would appreciate a response at your earliest convenience and again, please keep in mind that the more conversational your answer, the more usable it will be! _________________________________________________________________ Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/ Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/ Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxxCommunity Files: http://ontolog.cim3.net/file/work/OntologySummit2012/Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012 Community Portal: http://ontolog.cim3.net/wiki/
|