Alvaro Semantic Obligation Policies
From Semantic Portal Wiki
- Presentation Given at: CSCI 6965 Advanced Semantic Web (Spring 2009), Week 4
- Speaker: Alvaro Graves
- Title: Specifying and enforcing high-level semantic obligation policies
- Authors: Zhen Liu, Anand Ranganathan, Anton Riabov
- URL: http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B758F-4S7S2RS-1-Y&_cdi=12925&_user=7742305&_orig=browse&_coverDate=01%2F31%2F2009&_sk=999929998&view=c&wchp=dGLbVzz-zSkWz&md5=c6d385041855ad575537a8b1ecc1c926&ie=/sdarticle.pdf
- Date of Presentation: 2009/02/03
Questions
| ID | Question | Name |
|---|---|---|
| Alvaro Semantic Obligation Policies Jesse Weaver | Section 4 says that the authors' notion of a graph pattern is borrowed from SPARQL, but before that they say that it is a set of triple patterns and that "a graph pattern can be used to represent a conjunctive query." This idea doesn't align with a SPARQL graph pattern but rather with a SPARQL basic graph pattern (a specific kind of SPARQL graph pattern; see http://www.w3.org/TR/rdf-sparql-query/#GraphPattern for details). Similar to Greg's question, why wouldn't they want to allow other types of graph patterns in SPARQL, allowing for disjunction/alternation or inclusion of optional information? Consider the example in figure six. Wouldn't it be reasonable to include contact information for the employee (phone number, email, cube location... something to ensure the manager can approach the employee) if such information was available in the stream? Something like the SPARQL OPTIONAL graph pattern might be useful here, would it not? | Jesse Weaver |
| Alvaro Semantic Obligation Policies Joshua Shinavier 1 | The authors state that "the problem of policy concurrency is addressed by the implicit serialization of events as enforced by our streaming model". However, they also claim that the compiler is able to "plan with large numbers of event sources". When there are multiple sources involved, concurrency certainly is an issue. It is not clear how the compiled policy, a "plan" in Stream S, would go about combining events from multiple sources. Under what conditions can event descriptions from two different sources serve as input to be matched by a single event pattern? | Joshua Shinavier |
| Alvaro Semantic Policies GTW1 | Conditions in the Eagle policy specification language are said to be "described as a conjunction of expressions." Is there a reason why disjunction are not allowed? Disjunction could be emulated by simply producing new conditions, but this runs the risk of over burdening the person charged with maintaining the policies. Wouldn't many real-world conditions involve disjunctions, making it a desirable feature?
Why does section 4 restrict the use of variables to just the subject and object of an RDF triple pattern? Are there policies that would benefit from the ability to match on the semantic relationships between individuals, not just the individuals themselves? Is there any provision in the system to support graph pattern matching against multiple discrete events? Could this be handled by a general PE? The evaluation uses what seems like a rather complex setup of randomly generated PEs and RDF graphs with no real discussion of how this might simulate data used in real policies. Is this expected to have any relationship to real-world policies? |
Gregory Todd Williams |
Facts about Alvaro Semantic Obligation PoliciesRDF feed
