We present our approach in building WonderDesk - a semantic desktop for resource sharing and management, which is a component of our WonderSpace application and designed to be an e-Science tool. We propose a desktop architecture for resource sharing and management using semanticenriched RSS model and P2P technology. To enable fine-grained metadata for e-Science, an ontology about resource is presented as basic vocabulary to be used in constructing resource metadata along with RSS. Further, group-specific discipline ontology is proposed to enhance the semantic description of information resource within user group.
Nowadays there are many ways for resource sharing. It will be beneficial to the
resource consumers if resources are provided with richer metadata. Some efforts have
been made on centralized resource sharing with rich metadata, such as Forum Nokia1.
Other conspicuous efforts are made on desktop, such as Haystack [
In literature [
WonderDesk is designed to be an e-Science tool and mostly concern about
nontrivial resources that are important for research. Born as an e-Science tool,
WonderDesk is strongly affected by the visionary prescribed by Vannevar Bush in his
“As We May Think” [
to construct fine-grained resource metadata. We can also add metadata about topic information to an e-Science resource. Since topics of resource differ a lot, it is hard to design a comprehensive topics category. An ontology about group-specific discipline is proposed to enhance the semantic description of information resource within user group, and each topic of a resource can be point to a class in the ontology. Different groups may use different discipline ontologies, which entirely depend on their research areas. 2
WonderDesk is a component of WonderSpace, which is a Semantic Web application developed in our Lab2. Its potential users are researcher groups.
WonderSpace is still under development, of which WonderDesk and WonderServer are two important components. WonderDesk is a desktop tool focusing on resource sharing and management, while WonderServer acts as an information integrator for all the WonderDesk peers in a group and undertake some computation extensive tasks. It affords some web services and P2P services like indexing service simultaneously. The reason we design in this way is that: when we are trying to obtain scalability, we still want to utilize some server technology to handle issues like concurrent requests and gain high performance.
The typical scenario of WonderDesk is a combined usage with a WonderServer within a research group. We named it a stand-alone WonderSpace application. 3
In WonderDesk, RSS3 model is used as the model of resource sharing, and we use RSS 1.0 specification as the syntax to describe resources metadata. This idea is from Blog (or Weblog) systems. In WonderDesk, we enrich RSS 1.0 Vocabulary to represent specific metadata of various kinds of resources.
WonderDesk is developed basing on an open source RSS Aggregator – RSSOwl4 (The “Owl” here has no relation with the Web Ontology Language), We extended its RSS parser and equipped it with ontologies to make our vocabularies in RSS file comprehensible to the computer. More details about RSS and our ontologies will be discussed in following sections.
WonderDesk peer has the following functionalities: • Presents an user interface to let desktop users annotate their resources; • Shares the annotated resource with the members in a group; • Present an enhanced query interface to help desktop user search for the shared annotated resources within a group; • Requests and retrieves resources from other peers.
4 http://www.rssowl.org
It is typical but not required to use WonderDesk peers combined with
WonderServer. We can behold the typical scenario as an implementation of hybrid
P2P architecture [
P2P Listener
Web ServiceClient Metadata Generator
Extended RSSParser
Sesame
P2P Service Controller
PolicyStorage Policy Maker Policy Checker User Profile
Manager RDFStorage
JXTA/CMS
We have built ontology of resource, which defines terms to be used in constructing resource metadata along with other modules in RSS 1.0 specification. The other is group-specific discipline ontology describing the topic of resources. 4.1 Resource Ontology 5 http://openrdf.org 6 http://openrdf.org/doc/sesame/users/ch06.html resources with WonderDesk at present, this kind of hierarchy concerning more general resources provides scalability for potential extension by any third-party. ^^xsd:String ^^xsd:String
^^xsd:String ^^xsd:String ^^xsd:String
crecreaattinogDrate description identifier Resource
^^xsd:String 4.2 Group-specific Discipline Ontology As mentioned before, topics of shared resources vary within different research groups. And in most cases, a certain group concerns about limited number of research area.
We build an ontology presenting a simple and incomprehensive classification of Computer Science research areas, where each area corresponds to a node in ACM CCS7. Each topic of a certain resource shared within our group can be linked to an area in this ontology.
It is nature that each stand-alone WonderSpace application will share a unified group-specific discipline ontology. We prefer the ontology to be small and specialized than huge and general for the sake of efficiency.
We have demonstrated our approach in build WonderDesk - a semantic desktop for resource sharing and management, which is a component of our WonderSpace application. We design it to be an e-Science tool and benefit a lot from the experiences of previous research, including Haystack, Chandler and Gnowsis.
Among other semantic desktops, WonderDesk mainly contributes to the following aspects: first, we originally propose a desktop architecture for resource sharing and
managing using RSS model and P2P technology; second, to enable fine-grained metadata for e-Science, a resource ontology is presented as basic vocabulary to be used in constructing resource metadata along with RSS. Further, group-specific discipline ontology is proposed to enhance the semantic description of information resource within user group. These ideas have been illustrated to be feasible in our primary implementation.
Our research is just an initial attempt in building Semantic Desktop. A lot of issues are still under consideration. In the future work, we will focus on the semi-automatic or automatic annotation for resources, which will bring great facility to WonderDesk users as discussed in foregoing sections. We will also focus on the connection and collaboration of different stand-alone WonderSpace applications, and model a social researcher network. Trust and reputation issues will be addressed to help users making decision in their researching collaborations.
The work is supported in part by National Key Basic Research and Development Program of China under Grant 2003CB317004, and in part by the NSF of Jiangsu Province, China, under Grant BK2003001, and in part by Hwa-Ying Culture and Education Foundation. We would like to thank Dr. Zhiqiang Gao, Dr. Yuqing Zhai, Prof. Jianming Deng and Dr. Yanbing Wang for their suggestions.