In this paper, we propose a Web recommendation system based on a maximum entropy model. Under the maximum entropy principle, multiple sources of knowledge about users' navigational behavior in a Web site can be seamlessly combined to discover usage patterns and to automatically generate the most effective recommendations for new users with similar profiles. In this paper we integrate the knowledge from page-level clickstream statistics about users' past navigations with the aggregate usage patterns discovered through Web usage mining. Our experiment results show that our method can achieve better prediction accuracy when compared to standard recommendation approaches, while providing a better interpretation of Web users' diverse navigational behaviors.
Recently proposed key management systems (KMSs) provide limited functionality in highly partitioned mobile ad hoc networks. In this paper we present a framework for key management that provides redundancy and robustness for Security Association (SA) establishment between pairs of nodes in mobile ad hoc networks. Our KMS uses a modified hierarchical trust Public Key Infrastructure (PKI) model in which nodes can dynamically assume management roles. The system ensures high service availability for the network members through a number of schemes. A novel behavior grading mechanism provides security criteria for the network nodes and aids the management functions of the KMS to revoke or reissue certificates for nodes. This mechanism is based on the notion of trust, and more specifically on SAs among nodes in the entire network. In this paper we give an overview of the framework of the system and present a performance analysis of the system based on certificate issuance and acquisition. Finally, we compare our KMS with threshold cryptography schemes.
A vast amount of data is available on the World Wide Web. Many companies have started to mine this data to augment datasets used in production. This paper presents an industry study for one such text data mining. The study and analysis of the extracted data from the World Wide Web could be used to help improve products and services. A prototype has been completed and benched marked in the laboratory.
Outdoor location systems based on Wireless Access Point (WAP) signal strength must either know the exact location of the WAPs in order to use a triangulation algorithm, or must construct a radiomap of the signal strengths. While the radio-map technique increases accuracy and can accommodate a wireless network that is not owned by any one entity, conventional methods for constructing a radio-map are labor-intensive and impractical in such environments. We introduce a toolkit for automatically constructing outdoor radio-maps. Our toolkit can easily be carried by delivery personnel and security guards during their normal work duties to obtain signal readings. The scattered data readings are then fed into an interpolation algorithm to construct a more complete grid that can be used as the radio-map.
We describe our experience applying tabular mathematical approaches to software specifications. Our purpose is to show alternative approaches to writing tabular specifications and to help practitioners who want to apply such methods by allowing them to pick the best one for their problem. The object for the case study is software used by Dell Products for testing the functionality of the keyboards on notebook computers. Starting from informal documents, we developed a variety of tabular representations of finite state machine specifications and tabular trace specifications. We found that the discipline required by these methods raised issues that had never been considered and resulted in documents that were both more complete and much clearer. The various tabular representations are compared from a user's point of view, i.e., clarity, consistency, unambiguity, completeness, suitability, etc.
Some multiplayer, on-line games rely on the collaboration of all participating players. If a player's gamble is aborted, the rest of players cannot continue playing. This behavior can be used by fraudulent players to avoid paying by simply quitting the game before its completion. It is difficult to decide whether a player has left the game in a deliberated, fraudulent way, because there are many factors, both intentional and inadvertent, that can cause the abandonment. This paper presents a fraud detection system that specially fits to such scenarios. By gathering and correlating information held by multiple sources, our approach will help the on-line casino administrator to decide if a player leaving a game is actually cheating. Results of our work can be easily adapted for use against other existing on-line gambling frauds.
This paper presents an efficient and secure protocol for key issuing in ID-based cryptosystems using bilinear pairings. We use a simple blinding technique to eliminate the need of secure channel and multiple authorities approach to avoid the key escrow problem. We show that our protocol overcomes the disadvantages of other pairing based schemes and is efficient compared to the existing key issuing protocol.
In this paper we investigate the selection of an optimal threshold level that takes into account security and quality of service requirements for ad hoc networks. We incorporate intelligence into the selection of an optimal threshold for a distributed threshold cryptography scheme for a distributed trust model in ad hoc networks. We investigate both local and global threshold schemes.
Protecting sensitive information (credit card information, social security number... etc.) from being captured is of utmost importance as it can lead to serious repercussions if the collected data lands into unscrupulous hands. Despite tremendous advances in security, plenty of security problems still afflict systems. Existing security approaches do not prevent processes running in the background from capturing sensitive information on the screen. This paper proposes a no-capture hardware security feature that prevents capturing of sensitive data by users that have administration privileges.
We show the design of a network processing system in an IPv6 based media gateway based on abstract state machine (ASM) logic and show how the design features may help in scalability realization for end-to-end QoS issues in next generation networks. The novelty of the approach lies in the formal algorithmic definition of the processing hardware and generation of physical design unit from the data-driven algorithm function computation at compile time. The design in this paper is used to build an implementation prototype of an IPv6 media gateway with a feature of MPEG4 BIFS-MPEG4 LASeR transcoding, standardization activities of which are being currently pursued at the ISO MPEG MDS.
Though software engineering seems to focus on the development of software, most of the funds are spent maintaining and testing the systems. The ability to predict effort required to test and maintain is valuable to an organization interested in allocating project resources accordingly. This paper reviews the metrics cyclomatic complexity and weighted methods per class. The flaws in weighted methods per class are examined and a suite of metrics to determine and justify class complexity is proposed. This paper also discusses the design of a software tool that can be used to extract the values for the proposed metrics suite. The benefits of the software tool are also discussed.
The principal goal of near video-on-demand system is to minimize the average user waiting time. The user waiting time of popular videos can be reduced significantly by using broadcasting schemes such as fast broadcasting. Instead of allocating channels uniformly to all the videos, channels are nonuniformly distributed according to the relative popularity of the videos to minimize the average user waiting time. Since the demand for videos changes from time to time, the number of channels allocated to a video needs to be changed dynamically. The channel transition should be seamless so that the users currently watching the video do not experience any disruption due to this transition. Existing seamless channel transition scheme pads a dummy video stream at the end of the original video stream to produce exact correspondence between segments, but this approach causes some wastage of bandwidth. This paper proposes an improved seamless channel transition scheme which minimizes the wastage of bandwidth by preloading a postfix of the original video stream in the client buffer.
We propose a novel, computationally efficient generative topographic model for inferring low dimensional representations of high dimensional data sets, designed to exploit data sparseness. The associated parameter estimation algorithm scales linearly with the number of nonzero entries in the observations while still learning a truly nonlinear generative mapping. The latent variables of the model lie in a 2D space that can be used for visualisation. We discuss related work and we provide experimental results on text based documents visualisation as well as the exploratory analysis of Web navigation sequences.
This paper develops a new idea, traces, which speeds up the search for candidate Blackjack betting systems by >400X. We then use traces to enable a new application of scatter search, which identifies alternative betting systems that are statistically identical to the best professional ones and exceed others in mean betting returns in a study of 21 systems.
Software Architecture (SA) provides a formal basis to describe and analyze a software system as a collection of components and their interactions (connectors). However, the current understanding and specification for a connector are still insufficient. Furthermore, there still exists a gap between architecture design and component-based implementation. This paper attempts to tackle these problems by providing an Architecture Description Language (ADL), xSADL. In the specification of a connector, xSADL can help designers to deal with two kinds of architectural mismatches: behavior mismatch and message mismatch. xSADL is also supported by a GUI-based CASE tool. To help transform the architecture design by xSADL to a component-based software implementation, a component framework, 3CoFramework, is also introduced. xSADL and 3CoFramework complement each other and construct an architecture-based component development environment.
Resource brokers on the grid consult a number of distributed information services to select the best data source and/or computational resource based on user requirements. The consultation task increases the design complexity of Resource Brokers. This work is the first attempt to unify the distributed grid information services in one framework. The grid query service (GQS) is composed of information services layered on OGSA-DAI grid services, which are generic data access and integration services. Grid resource brokers can consult just one service, the GQS service, to obtain indexed information about grid resources.
The authors described the progress extending the undergraduate computer science (CS) curriculum to include a deep understanding of techniques for information and knowledge management systems (IKMS). In a novel five-course sequence, students build and work with techniques for data mining, information retrieval, and text analysis, and develop a large-scale IKMS project. The authors taught in a hands-on lab setting where students use tools they have built, performing experiments that could extend the field. Hence undergraduates have firsthand knowledge of performing CS research using scientific methods. Second, a rigorous set of evaluation criteria developed in the Psychology Institute was utilized to evaluate how well students learn using our approaches. Ultimately, it is believed that this specialization warrants inclusion as an option in the standard undergraduate CS curriculum.
The complexities of the basic unit of OO systems, the class, have resulted in a plethora of implementation-based testing techniques (IBTTs) to test the characteristics of a class. Surprisingly, there has been little or no research in the area of automatically identifying those IBTTs suitable for testing a class. In this paper we present a tool that automates the process of mapping IBTTs to a class under test (CUT). The mapping process identifies those IBTTs that can suitably test characteristics of the CUT and, more importantly, provides feedback to the tester that facilitates identification of the characteristics of the CUT that are not suitably tested by any of the IBTTs in the list.
Concept-based access to information promises important benefits over keyword-based access. One of these benefits is the ability to take advantage of semantic relationships among concepts in finding relevant documents. Another benefit is the elimination of irrelevant documents by identifying conceptual mismatches. Concepts are mental structures. Words and phrases are the linguistic representatives of concepts. Due to the inherent conciseness of natural language, words can represent multiple concepts and different words may represent the same or very similar concepts. Word Sense Disambiguation attempts to resolve this ambiguity using contextual information. The use of an ontology facilitates identification of related concepts and their linguistic representatives given a key concept. Latent semantic analysis, on the other hand, attempts to reveal the hidden conceptual relationships among words and phrases based on linguistic usage patterns. In this work we explore the potential of concept-based information access via these two methods. We examine under what circumstances concept-based access becomes feasible and improves user experience.
Access control software must be based on a security policy model. Flaws in them may come from a lack of precision or some incoherences in the policy model or from inconsistencies between the model and the code. In this paper, we first present a formalisation of access control models based on the work on an algebra of security models by J. McLean (1988). Then, we describe the implementation of this framework and show how it can be used to obtain a particular security model: the Bell and La Padula security model. Last, as an example, we show how such a program can be integrated for secure databases. All our development is done within the Focal (Rioboo et al., 2004) programming environment which provides a language with object-oriented features allowing to write formal specifications, proofs and programs at the same level.
