Knewton is an interesting company providing a recommendation service for adaptive learning applications. In a recent post, Jonathon Goldman describes an algorithmic approach to generating questions. The approach focuses on improving the manual authoring of test questions (known in the educational realm as “assessment items“). It references work at Microsoft Research on the problem of synthesizing questions for a algebra learning game.
We agree that more automated generation of questions can enrich learning significantly, as has been demonstrated in the Inquire prototype. For information on a better, more broadly applicable approach, see the slides beginning around page 16 in Peter Clark’s invited talk.
What we think is most promising, however, is understanding the reasoning and cognitive skill required to answer questions (i.e., Deep QA). The most automated way to support this is with machine understanding of the content sufficient to answer the questions by proving answers (i.e., multiple choices) right or wrong, as we discuss in this post and this presentation.
Recently, John Sowa has commented on LinkedIn or in correspondence with some of us at Coherent Knowledge Systems on the old adage due to Shanks concerning the Neats. vs. the Scruffies. The Neats want nice formal logics as the basis of artificial intelligence. This includes anyone who prefers classical logic (e.g., Common Logic, RIF-BLD, or SBVR) or standard ontologies (e.g., OWL-DL) for representing knowledge and reasoning with it. The Scruffies may use well-defined technology, but are not constrained by it. They’ll do whatever they think works, now, whether or not it is a good long term solution and despite its shortcomings, as long as it can obtain immediate objectives.
Watson is scruffy. It doesn’t try to understand or formally represent knowledge. It combines a lot of effective technologies into an evidentiary framework that allows it to effectively “guess”.
Today, in response to continued discussion in the Natural Language Processing group on LinkedIn under the topic “This is Watson”, I’m posting the following presentation on Project Sherlock and the Linguist vs. Google and IBM.
Essentially, the neat approach is more viable today than ever. So, chalk one up for the neats, including Dr. Sowa and Menno Mofait’s comment in that discussion.
During a presentation at CMU after winning the game show,, IBM admitted that in order to get the last leg of improvement needed to win Jeopardy!, they needed to do some “neat” ontological knowledge acquisition, too!
The folks from Knowledge Partners have a post that I found thanks to Sandy Kemsley, whose blog often provides good pointers. This article talks about the decision perspective on business rules. It makes some good points, on which I would like to elaborate albeit at a more semantic or knowledge-level.
Every language has three kinds of statements: questions, statements, and commands. There are also some peripheral types, such as exclamations (Yikes!), but in business processes and decisions only declarative and imperative sentences matter.
From a process- or decision-oriented perspective, decisions are always phrased as imperative sentences. Otherwise, the statements reflected in any business process, whether you are using BPMN or a BRMS, are declarative sentences.
Decisions are imperative sentences because they state an action to be taken. For example, decline a loan or offer a discount. It’s really pretty simple. A decision is an action. Rules that don’t take actions are statements of truth. Such declarative statements of truth are perfect for formal logic, logic programming, and semantic technologies. It’s the action that requires the production rule technology that dominates the market for and applications of rules.
The authors of the aforementioned article use the following diagram to explain the benefits of the decision-oriented approach in simplifying business processes:
The impact is very simple. If you eliminate how you reach decisions from the flow that you diagram in BPMN things get simpler. It’s really as simple as realizing that you have removed all the “if” parts (i.e., the antecedents) of the rule logic from the flow chart.
So who in their right mind would use a business process tool to express any business logic? Continue reading “What could be more strategic than process or decision management?”
Work on acquiring knowledge about science has estimated the cost of encoding knowledge in question answering or problem solving systems at $10,000 per page of relevant textbooks. Regrettably, such estimates are also consistent with the commercial experience of many business rules adopters. The cost of capturing and automating hundreds or thousands of business rules is typically several hundred dollars per rule. The labor costs alone for a implementing several hundred rules too often exceed $100,000.
The fact that most rule adopters face costs exceeding $200 per rule is even more discouraging when this cost does not include the cost of eliciting or harvesting functional requirements or policies but is just the cost of translating such content into the more technical expressions understood by business rules management systems (BRMS) or business rule engines (BRE).
I recommend against adopting any business rule approach that cannot limit the cost of automating elicited or harvested content to less than $100 per rule given a few hundred rules. In fact, Automata provides fixed price services consistent with the following graph using an approach similar to the one I developed at Haley Systems.
Continue reading “The $50 Business Rule”