We consider a collaborative application scenario in Open Hypermedia Systems. We describe a semantic search algorithm to discover semantically equivalent or related resources across distributed link databases, otherwise known as linkbases. Our approach differs from traditional crawler based search mechanisms because it relies on clustering of semantically related entities. It creates clusters of related semantic entities to expedite the search for resources in a random network. It uses a distance-vector based heuristic to guide the search. Our results confirm that the algorithm yields high search efficiency in collaborative environments where the change in content published by each participant is rapid and random.

This paper presents HyperPeer, a framework for developing peerto-peer based hypermedia. The distribution of hypermedia structures is handled through a peer-to-peer (P2P) network, allowing for highly scalable sharing between users. A central challenge of all decentralized systems is to locate material of interest and this paper presents the HyperPeer Hierarchy of Resemblance (HR) searching algorithm, which provides an efficient search as well as partitioning of the network into groups of common interest. Categories and Subject Descriptors:

Abstract. This paper proposes a framework for enabling the usage of Hypermedia services by web developers, and studies the working progress that provides Hypermedia services to the Web. It investigates and analyzes the requirements that will provide web application developers with the ability of easily inserting Hypermedia functionality into their applications. A Service Discovery Mechanism for finding and using services is defined, and solutions for increasing the usage of Hypermedia systems by web developers are proposed. 1

This paper argues that the existence of a developer support framework is a critical issue to the usage of Open Hypermedia Systems (OHSs). For this reason, the OHS Community would benefit by the adoption of both a service discovery mechanism and a set of standards and tools to approach the development of hypermedia clients in a transparent and methodological manner.

Current hypermedia modeling techniques such as OOHDM, RMM or WebML rely on class-based domain models to guide the hypermedia construction process. In this paper we discuss the notion of role-oriented models as particularly suited for capturing different semantic contexts of hypermedia interfaces. The key idea is to allow resources to fill different roles which specify how these resources can be embedded in different link structures. After an overview of some class-oriented models, we discuss in detail the main features and benefits of role-oriented modeling compared to conventional class-oriented approaches, and make clear the distinction between attributes and links associated with objects via their natural types and attributes and relationships for an object specified by the different roles the object can fill. Two detailed examples, one discussing the use of role-oriented models for the specification of link structures in learning materials, another one for the specification of contexts for a tourist information system, show how these role-oriented models can be applied in different areas. Finally we show how such models can be easily expressed and implemented using the W3C standard for web annotation, RDFS, and show that a role-oriented semantics is actually closer to the intended use of RDFS schemas than the "default" class-oriented semantics.

We present Storm, a storage system which unifies the desktop and the public network, making Web links between desktop documents more practical. Storm assigns each document a permanent unique URI when it is created. Using peer-to-peer technology, we can locate documents even though our URIs do not include location information. Links continue to work unchanged when documents are emailed or published on the network. We have extended KDE to understand Storm URIs. Other systems such as GNU Emacs are able to use Storm through an HTTP gateway.

Abstract. Due to the ever increasing number of different digital media types that we use in our daily work, it is no longer sufficient to manage them in an isolated way but desirable to define associations across the media boundaries. While cross-media information systems enable associations between digital and physical information, there is often a lack of support for flexible authoring and sharing of these associations (links) as, for example, required in meeting situations. We present a solution for this kind of ad-hoc information exchange and collaborative authoring in cross-media environments which is based on a link service in combination with peer-to-peer technologies. Our goal was not to replicate hypermedia documents based on peer-to-peer technology, but rather to use a distributed link service architecture to enable the sharing of link metadata in collaborative information spaces. To avoid information overload and guarantee a certain link quality in these highly dynamic information environments, we adopt some general mechanisms for user and link rating. 1