The document outlines Jie Bao's research background and overview, including work on linked ontologies and linked data using semantic wikis. It discusses a modular ontology approach called P-DL that allows importing between ontologies similar to citation. It also describes using a semantic wiki to generate linked data from wiki revision histories and other semantic data. Future work includes applying these techniques to government data and improving scalability.
1. Towards Linked Ontologies and Data on the Semantic Web Jie Bao , Rensselaer Polytechnic Institute baojie@cs.rpi.edu, http://www.cs.rpi.edu/~baojie Oct 1st, 2009
10. Why not owl:imports? owl:imports (Alice, 2001) (Bob, 2009) Analogy: Paper Writing in the OWL fashion Recent development in Web… In this paper, we extend the algorithm A proposed by (Alice,2001) … Recent development in algorithms … In this paper, we present two algorithms A and B to …
11. Why not owl:imports? (Bob, 2009) Analogy: Paper Writing in the OWL fashion In this paper, we extend the algorithm A proposed by In this paper, we present two algorithms A and B to … Recent development in algorithms … Recent development in Web…
13. P-DL: Semantics People Animals O 1 O 2 Only “relevant” local domains are connected Expressions are interpreted in local domains (e.g., neg)
14. Modular ontology languages: Comparison Distributed DL E-Connection P-DL What can you import? Nothing Concept name, Nominal name Concept , Role, and Nominal name What can you do with the imported names syntactically? (bridge rules between concepts and between roles) Use it in the range of a (link) role Free use, except that imported roles can not be used in role inclusions What is the result semantically? Decidability with BR between concepts. Decidability Decidability , Transitive Reusability, Preservation of Unsatisfiability
21. Private Knowledge Locally visible : Has date Q: Has date? A : Unknown Bob’ schedule ontology Q: Busy? A : Yes Q: Has dinner? A : Unknown indistinguishable
22. Privacy-Preserving Reasoners More informative More general Dummy Reasoner “ Combina-tion Safe” Reasoner Naive Reasoner Gives information Always answer “U” ? Always answer faithfully Safety scope All ontologies Ontologies that have no hidden knowledge
32. Semantic History: Example (Provenance tracking) Who has changed the first name of James Hendler? http://tw.rpi.edu/proj/semhis.wiki/index.php/Example_4
36. Concept Modeling Person Name Role Alias Affiliation Project Name Member Issue ID Project Assignee Jim Hendler Professor hasRole Person rdfs:subClassOf Alice John hasUncle Alice Bob John isParentOf isBrotherOf RDF Modeling Relational Modeling Rule Modeling
37. Rules: Logic Programs RightHanded(x):-Person(x), not LeftHanded(x). {{LP Rule |body= 1::Person; 1:not:LeftHanded |head= RightHanded }} Example : every person is by default right-handed, unless that person is known to be left-handed: See details at: http://tw.rpi.edu/proj/cnl/Template:LP_Rule [ACITA 2009]
42. Case Study: CNL Wiki Class(Rabbit partial intersectionOf(animal restriction(eat someValuesFrom(FreshVegetable))) OWL: “Rabbit is Animal and eats some fresh vegetable” Us wiki templates to create OWL meta-model extensions for SMW Form-based editing interface associated with templates
Let’s see an example. This slide shows Bob’s schedule ontology. Bob’s has a private activity that he is unwilling to share with the public. Let’s see what might happen when his ontology is queried, such as by other people’s software agent, for his availability. While Bob knows he has a date at 3pm by himself, he won’t reveal its details to everyone. When being asked if he is busy, his ontology, in fact, the software for answering queries about ontology, may answer “Yes”. However, when being asked about if he has a date, the software may answer “no” or “unknown” to protect Bob’s privacy. In this work, we assume the answer is unknown. This would not let a reasoner software to lie in protecting privacy. Similarly, when being asked if Bob’ has dinner from 3pm to 5pm, the software may also answer “unknown”, since this piece of information is not in the ontology. We call this incomplete knowledge. Please note that since both hidden knowledge and incomplete knowledge are answered with unknown, a querying agent will not be able to distinguish which case is what. We may use this fact to protect privacy from harmful querying.
Any question so far?
Now let’s see how could we make not only ontologies, but also structured data available on the web that can be linked to each.
We will discuss three problems in dealing with structured data.
Li has presented this on Oct 3, 2008.
This shows… There a couple more interesting use cases for the semantic history extension, please see the demo site for details.
I have also involved in a several other projects that turn data into SMW format, or to export SMW data into another forms…
http://www.cs.rpi.edu/~hendler/LittleSemanticsWeb.html Goal: to have “little semantics” usable by everyone