This paper presents the core of a distributed document-sharing environment with particular focus on distributed searching and retrieval of research papers. Problems with Current p2p Systems Despite their obvious popularity, systems like Napster and Gnutella suffer from many problems. Napster uses centralised indexing servers, an approach which is vulnerable to failure. Gnutella avoids Napsters weakness by using a decentralised indexing technique, however this leaves Gnutella with the problem of locating objects within its network. Gnutella uses a flood-based search technique where each search request blindly hops across the network from one node to another searching for the requested file. The search request will only reach an ever-decreasing section of the network as more users join; as a result this technique doesnt scale very well. In recent times, more scalable object location algorithms have emerged that are based on Distributed Hashtables (DHT). Let us consider these useful constructs and how they can be used to facilitate searches in a distributed environment. 3 Existing Structured p2p Overlays Based originally on the research of Plaxton et al in the late 1990s [4], DHT overlay network implementations first appeared during 2001. Projects such as Pastry [5], Tapestry [6], CAN [7] and Chord [8], all produced implementations that adhered to the principles of P2P, decentralisation, robustness and scalability. These systems may be used to form the foundation for functional p2p systems. They provide a routing substrate; a mechanism that efficiently locates objects within a certain number of routing hops. The subject of keyword searching is of particular importance if such DHT systems are to become part of the more mainstream p2p systems ...

In this paper we describe our participation in the Second Web People Search workshop (WePS2) and detail our approaches. For the clustering task, our focus was on replicating the lessons learned at WEPS1 on the data set made available as part of WEPS2 and on experimenting with a voting-based combination of clustering methods. We found that clustering methods display the same overall behavior on the WEPS1 and WESP2 data sets and that a hierarchical clustering approach delivers the best performance, even outperforming voting-based combinations. For attribute extraction, we explore approaches using pattern matching with manually and automatically constructed patterns. Manual patterns were constructed using expert knowledge and following analysis of sample data. Automatic pattern construction extracts textual and syntactic context around training samples and selects patterns which are expected to perform well based on leave-one-out evaluation. Experimental results show that manually constructed patterns are very effective for obtaining high recall. For automatically extracted patterns performance varied widely depending on the attribute type. Larger amounts of training data may help improve these approaches in the future.

ABSTRACT. We present Flycatcher, a prototype system illustrating the idea of gazebased image processing in the context of object segmentation for wearable photography. The prototype includes a wearable eye tracking device that captures real-time eyetraces of a user, and a wearable video camera that captures first-person perspective images of the user’s visual environment. The system combines the deliberate eyetraces of the user with hierarchical image segmentation applied to scene images to achieve reliable object segmentation. In evaluations with certain classes of real-world images, fusion of gaze and image segmentation information led to higher object detection accuracy than either signal alone. Flycatcher may be integrated with assistive communication devices, enabling individuals with severe motor impairments to use eye control to communicate about objects in their environment. The system also represents a promising step toward an eye-driven interface

In this chapter I shall attempt to present a coherent account of classification in such a way that the principles involved will be sufficiently understood for anyone wishing to use classification techniques in IR to do so without too much difficulty. The emphasis will be

There has been a tremendous growth in the amount of information and resources on the World Wide Web that are useful to researchers and practitioners in science domains. While the Web has made the communication and sharing of research ideas and results among scientists easier and faster than ever, its dynamic and unstructured nature also makes the scientists faced with such problems as information overload, vocabulary difference, and lack of analysis tools. To address these problems, it is highly desirable to have an integrated, “one-stop shopping ” Web portal to support effective information searching and analysis as well as to enhance communication and collaboration among researchers in various scientific fields. In this paper, we review existing information retrieval techniques and related literature, and propose a framework for developing integrated Web portals that support information searching and analysis for scientific knowledge. Our framework incorporates collection building, meta-searching, keyword suggestion, and various content analysis techniques such as document summarization, document clustering, and topic map visualization. Patent analysis techniques such as citation analysis and content map analysis are also incorporated. To demonstrate the feasibility of our approach, we developed based on our architecture a knowledge portal, called NanoPort, in the field of nanoscale science and engineering. We report our experience and explore the various issues of relevance to developing a Web portal for scientific domains. The system was compared to other search systems in the field and several design issues were identified. An evaluation study was conducted and the results showed that subjects were more satisfied with the NanoPort system