Shangguan Journal CNLPresentation

From Semantic Portal Wiki

Presentation given at CSCI 6966 Advanced Semantic Web (Fall 2008)#Lesson 12

• Speaker: Zhenning Shangguan
• Title: A Controlled Natural Language for the Semantic Web
• Authors: Rolf Schwitter
• Conference: Journal of Intelligent Systems, Vol. 17, No. 1-3, pp. 125-141, 2008
• URL: http://tw.rpi.edu/portal/Image:IntelligentSystems-2008.pdf
• Date of Presentation: 2008/11/20

Questions

ID	Question	Name	Answer
CNL Ankesh	In section 5.2, Property Characterisitcs, author chooses to restrict the range and domain of a symmetric property. (My feeling is that symmetric properties would most of the times have the same domain and range, and therefore the rule may not be useful). However my real question is- does this really capture the semantics of symmetric property. Do we not require similar rule as for inverse property. In this case it would be like If E has the property P1 whose value is V and the property P1 is 'symmetric' then V has P1 whose value is E. Or it is enough to say If the property P1 is 'symmetric' then P1 is 'inverse of' P2. In section 5.3, Class Equivalence, author defines the rule If C1 is an equivalent class of C2 and E has C1 whose value is V then E has C1 whose value is V. Here, what does E has C1 mean? This refers to the same quote as above. Why does author choose to show equivalence between classes only when that class is part of an equivalence axiom. i.e. author chooses to say, C1 equivalent to C1 if there-exists C2 such that C1 is equivalent to C2. why is there no need to declare C1 equivalent to C1, for rest of the classes?	Ankesh Khandelwal
Schwitter2008controlled question 1 by lebo	The Controlled Natural Language presented in this paper seems to be useful for pedagogical purposes, but little more. Although hiding a more obscure syntax /may/ increase the approachability and/or "explainability", it seems that a novice would still require a background understanding of OWL Lite^minus to do anything useful. The novice is at risk for establishing a greater expectation for the language based on the assumption that the English-like sentences are English; these expectations will be unfulfilled by the restricted ("focused"?) capabilities of a logic-based system. On the other hand, the verbosity that provides approachability for the novice would hinder use by "knowledge engineers" who prefer a more concise syntax. However, these issues are most concerning only when authoring the TBox and ABox. When posing questions to the system, the interaction seems more natural. This is done by generating a satisfying model with SATCHMO, which includes all entailed statements. What is a "satisfying model", how does it differ from other kinds of models, and how is it generated?	Tim Lebo
Shangguan Journal CNLPresentation GTW 1	Section 2 claims "it has been shown that about 77% of all current ontologies developed for the Semantic Web fall under the OWL Lite minus subset." However, this claim is based on a 2004 Ph.D. thesis. Since OWL was standardized in early 2004, do you think that subsequent development with OWL might mean ontologies developed since 2004 might use more expressive subsets of OWL?	Gregory Todd Williams
Shangguan Journal CNLPresentation GTW 2	This work seems to suggest that working directly in an RDF syntax (any syntax, although the paper conflates the RDF model with the RDF/XML syntax) isn't a reasonable path forward for the semantic web: "What is urgently required is a high-level interface language to the Semantic Web that abstracts away from these RDF-based formal notations." However, the simplistic examples used in the paper seem to confuse a resource and it's name. For example, section 3 equates "Nora Yuen supervises John Smith" with the RDF "<NoraYuen> ex:supervise <JohnSmith>". Can this model capture RDF's generality, for example describing two different people named "John Smith"? If the model uses names in this way as an identifier, it seems like it would fail to represent this data. If the "John Smith" in this CNL is meant to represent the URI, not the name, wouldn't this require a none-too-natural-language construct such as "John Smith is named 'John Smith'"?	Gregory Todd Williams
Shangguan Journal CNLPresentation Jesse Weaver	Section 1 of the paper proposes CNL as "an ideal candidate to increase transparency of the Semantic Web and to empower non-specialists with a 'seemingly informal' language to work in co-operation with computers." This seems to be the premise of the paper. However, the abstract states: "There is no need to formally encode this knowledge in an RDF-based notation." Section 4 also states: "Specifications in CNL are translated ... into a format that can directly be processed by SATCHMO." It is not actually shown that CNL actually interfaces to current semantic web technologies/standards/syntaxes/etc. Perhaps it is obvious for assertional statements and terminological statements, but conditional statements seems less intuitive (in terms of representation in the current semantic web). Sure, SATCHMO can handle them, but what about other systems? Do the conditional statements correspond to any existing standard rule language?	Jesse Weaver
Shangguan Journal CNLPresentation Jesse Weaver 2	OWL Lite^minus is mentioned frequently throughout the paper and yet is never defined. What is OWL Lite^minus? The paper states that it is "the maximal subset of OWL which can be expressed in the deductive database language Datalog." For those of us less familiar with Datalog, what is OWL Lite^minus in terms of RDFS/OWL features?	Jesse Weaver
Shangguan Journal CNLPresentation Joshua Shinavier 1	I am inclined to disagree with the authors' statement that "what is urgently required is a high-level interface language to the Semantic Web...", with which they motivate their innovation. Is it necessary for novice Web users to express RDF statements/graphs directly? It seems that novice users should deal with RDF only indirectly, through use of special-purpose interfaces which generate metadata on their behalf. Furthermore, the idea of an interface language is misleading, as the CNL tool (the "look-ahead text editor") is actually a GUI in disguise. The CNL expressions alone evidently cannot be decompiled to RDF.	Joshua Shinavier

Attendees

Tim Lebo