Search

International audience; Web browsers have come a long way since their inception, evolving from a simple means of displaying text documents over the network to complex software stacks with advanced graphics and network capabilities. As personal computers grew in popularity, developers jumped at the opportunity to deploy cross-platform games with centralized management and a low barrier to entry. Simply going to the right address is now enough to start a game. From text-based to GPU-powered 3D games, browser gaming has evolved to become a strong alternative to traditional console and mobile-based gaming, targeting both casual and advanced gamers. Browser technology has also evolved to accommodate more demanding applications, sometimes even supplanting functions typically left to the operating system. Today, websites display rich, computationally intensive, hardware-accelerated graphics, allowing developers to build ever-more impressive applications and games.In this paper, we present the evolution of browser gaming and the technologies that enabled it, from the release of the first text-based games in the early 1990s to current open-world and game-engine-powered browser games. We discuss the societal impact of browser gaming and how it has allowed a new target audience to accessdigital gaming. Finally, we review the potential future evolution ofthe browser gaming industry.

We measure the data sent to their back-end servers by five browsers: Google Chrome, Mozilla Firefox, Apple Safari, Brave Browser and Microsoft Edge, during normal web browsing on both desktop and mobile devices. Our aim is to assess the privacy risks associated with this data exchange between a browser and its back-end servers. With regard to shared services, all of the browsers make use of a safe browsing service to mitigate phishing attacks and our measurements indicate that this raises few privacy concerns. Similarly, with regard to the Chrome extension update service accessed by Chromium-base browsers (Chrome, Brave, Edge). Overall, we find that both the desktop and mobile versions of Brave do not use any identifiers allowing tracking of IP address over time, and do not share details of web pages visited with backend servers. In contrast, Chrome, Firefox, Safari and Edge all share details of web pages visited with backend servers. Additionally, Chrome, Firefox and Edge all share long-lived identifiers that can be used to link connections together and so potentially allow tracking over time. In the case of Edge these are device and hardware identifiers that are hard/impossible for users to change. On mobile devices, but not desktop devices, Firefox also shares device identifiers.

The aim of the work is the performance analysis of the Unity3D engine in the contex of applications running with a web browser and WebGL technology.. The paper presents literature review was carried out in the scope of creating multimedia applications run in web browsers, graphics engines and WebGL technology . Additionally, technical issues related with converting Unity3D application to WebGL technology are discussed. The research are performed based on computer simulations with five scenes created with Unity3D environment, which include the main capabilities of this engine. Finally, it was presented the results of these studies and the conclusions.

Web performance is important for the user experience and can heavily influence web page revenues. While there are many established Web Performance Optimization (WPO) methods, our work so far has clearly shown that new network protocols, optimized browsers and cutting-edge web standards can have a significant impact on known best practices. Additionally, there is still low-hanging fruit to be exploited, in the form of personalizing performance based on user context (i.e., current device, network, browser) and user preferences (e.g., text reading vs multimedia experience). In our PhD project, we strive to integrate this user-specific metadata into dynamic configurations for both existing and new automated WPO techniques. An intermediate server can (pre)generate optimized versions of a web page, which are then selected based on user context and preferences. Additional metadata is also passed along to the browser, enabling improvements on that side, and used to steer new network protocols to speed up the incremental delivery of page resources. We use the Speeder platform to perform and evaluate full-factorial objective measurements and use subjective user studies across a range of groups to assess the applicability of our methods to end users. Our aim is to provide insights in how WPO can be tweaked for specific users, in the hopes of leading to new web standards that enable this behavior. Robin Marx is a SB PhD fellow at FWO, Research Foundation - Flanders, project number 1S02717N. This PhD project is supervised by Prof. Peter Quax, PhD and Prof. Wim Lamotte, PhD.

Disseny i implementació d'un sistema amb capacitat de generar sons i imatges en temps real inspirat en el funcionament de les consoles làser dels anys setanta. Aquest sistema permet la creació d'una gran varietat de sons de tipus experimental i la seva corresponent visualització en temps real seguint els principis de l'anomenada "música visual". Design and implementation of a system able to generate sounds and images in real time inspired by the operation of the laser consoles of the seventies. This system allows the creation of a large variety of experimental sounds and their corresponding visualization in real time following the principles of the so-called "visual music". Diseño e implementación de un sistema con capacidad de generar sonidos e imágenes en tiempo real inspirado en el funcionamiento de las consolas láser de los años setenta. Este sistema permite la creación de una gran variedad de sonidos de tipo experimental y su correspondiente visualización en tiempo real siguiendo los principios de la llamada "música visual".

The Java tool jScope has been widely used for years to display acquired waveform in MDSplus. The choice of the Java programming language for its implementation has been successful for several reasons among which the fact that Java supports a multiplatform environment and it is well suited for graphics and the management of network communication. jScope can be used both as a local and remote application. In the latter case, data are acquired via TCP/IP communication using the mdsip protocol. Exporting data in this way however introduces several security problems due to the necessity of opening firewall holes for the user ports. For this reason, and also due to the fact that JavaScript is becoming a widely used language for web applications, a new tool written in JavaScript and called WebScope has been developed for the visualization of MDSplus data in web browsers. Data communication is now achieved via http protocol using Asynchronous JavaScript and XML (AJAX) technology. At the server side, data access is carried out by a Python module that interacts with the web server via Web Server Gateway Interface (WSGI). When a data item, described by an MDSplus expression, is requested by the web browser for visualization, it is returned as a binary message and then handled by callback JavaScript functions activated by the web browser. Scalable Vector Graphics (SVG) technology is used to handle graphics within the web browser and to carry out the same interactive data visualization provided by jScope. In addition to mouse events, touch events are supported to provide interactivity also on touch screens. In this way, waveforms can be displayed and manipulated on tablet and smartphone devices.

This paper reports on a specific Web-based self-report data collection system that was developed for a public health research study in the United States. Our focus is on technical outcome results and lessons learned that may be useful to other projects requiring such a solution. The system was accessible from any device that had a browser that can support HTML5. Report findings include: which hardware devices, Web browsers, and operating systems were used, the rate of survey completion, and key considerations for employing Web-based surveys in a clinical trial setting.

We present a query formulation language (called MashQL) in order to easily query and fuse structured data on the web. The main novelty of MashQL is that it allows people with limited IT skills to explore and query one (or multiple) data sources without prior knowledge about the schema, structure, vocabulary, or any technical details of these sources. More importantly, to be robust and cover most cases in practice, we do not assume that a data source should havean offline or inlineschema. This poses several language-design and performance complexities that we fundamentally tackle. To illustrate the query formulation power of MashQL, and without loss of generality, we chose the Data web scenario. We also chose querying RDF, as it is the most primitive data model hence, MashQL can be similarly used for querying relational databases and XML. We present two implementations of MashQL, an online mashup editor, and a Firefox add on. The former illustrates how MashQL can be used to query and mash up the Data web as simple as filtering and piping web feeds and the Firefox add on illustrates using the browser as a web composer rather than only a navigator. To end, we evaluate MashQL on querying two data sets, DBLP and DBPedia, and show that our indexing techniques allow instant user interaction. © 1989-2012 IEEE. 24 5 783 798 Cited By :15

This paper reports on a specific Web-based self-report data collection system that was developed for a public health research study in the United States. Our focus is on technical outcome results and lessons learned that may be useful to other projects requiring such a solution. The system was accessible from any device that had a browser that supported HTML5. Report findings include: which hardware devices, Web browsers, and operating systems were used; the rate of survey completion; and key considerations for employing Web-based surveys in a clinical trial setting.

In previous work with Bono we introduced a calculus for modelling “environment-aware” computations, that is computations that adapt their behavior according to the capabilities of the environment. The calculus is an imperative, object-based language (with extensible objects and primitives for discriminating the presence or absence of attributes of objects) equipped with a small-step operational semantics. In this paper we define a type and effect system for the calculus. The typing judgements specify, via constraints, the shape of environments which guarantees the correct execution of expressions and the typing rules track the effect of expression evaluation on the environment. The type and effect system is sound w.r.t. the operational semantics of the language.

Over the last few years, the World Wide Web has transformed itself from a static content-distribution medium to an interactive, dynamic medium. The Web is now widely used as the presentation layer for a host of on-line services such as e-mail and address books, e-cards, e-calendar, shopping, banking, and stock trading. As a consequence (HyperText Markup Language)HTML files are now typically generated dynamically after the server receives the request. From the Web-site providers' point of view, dynamic generation of HTML pages implies a lesser understanding of the real capacity and performance of their Web servers. From the Web developers' point of view, dynamic content implies an additional technology decision: the Web programming technology to be employed in creating a Web-based service. Since the Web is inherently interactive, performance is a key requirement, and often demands careful analysis of the systems. In this paper, we compare four dynamic Web programming technologies from the point of view of performance. The comparison is based on testing and measurement of two cases: one is a case study of a real application that was deployed in an actual Web-based service; the other is a trivial application. The two cases provide us with an opportunity to compare the performance of these technologies at two ends of the spectrum in terms of complexity. Our focus in this paper is on how complex vs. simple applications perform when implemented using different Web programming technologies. The paper draws comparisons and insights based on this development and performance measurement effort., © Elsevier

Many identity federations are designed to be used with web browsers. This paper proposes an approach for integrating non web-based applications with web-based identity federations using Kerberos protocol. We evaluate this approach by making NFS server available for users of SAML-based identity federation of Baden-Württemberg state of Germany. We make use of LDAP-Facade software for federating non web-based services. We have modified the web-interface component of LDAP-Facade to enable the registration with kerberized services. Our approach can be used without modifications on the client side. Copyright © 2015 SCITEPRESS - Science and Technology Publications. All rights reserved.

Keyword indices, topic directories, and link-based rankings are used to search and structure the rapidly growing Web today. Surprisingly little use is made of years of browsing experience of millions of people. Indeed, this information is routinely discarded by browsers. Even deliberate bookmarks are stored passively, in browser-dependent formats; this separates them from the dominant world of HTML hypermedia, even if their owners were willing to share them. All this goes against Vannevar Bush's dream of the Memex: an enhanced supplement to personal and community memory. We present the beginnings of a Memex for the Web. Memex blurs the artificial distinction between browsing history and deliberate bookmarks. The resulting glut of data is analyzed in a number of ways. It is indexed not only by keywords but also according to the user's view of topics; this lets the user recall topic-based browsing contexts by asking questions like ‘What trails was I following when I was last surfing about classical music?' and ‘What are some popular pages related to my recent trail regarding cycling?' Memex is a browser assistant that performs these functions. We envisage that Memex will be shared by a community of surfers with overlapping interests; in that context, the meaning and ramifications of topical trails may be decided by not one but many surfers. We present a novel formulation of the community taxonomy synthesis problem, algorithms, and experimental results. We also recommend uniform APIs which will help managing advanced interactions with the browser., © Elsevier

This paper compared the performance of the three web browsers: Google Chrome, Mozilla Firefox and Microsoft Internet Explorer from three different aspects: menu display/history searching, bookmark function and downloading function. A usability test was conducted based on participants from University of North Carolina at Chapel Hill. During the test, the quantitative data was reflected by task execution time and mouse-click counts while qualitative data was analyzed using task load and comments from participants. The results indicated that people were most satisfied with Chrome while some function or design of Firefox and IE may dissatisfy the users to some extent.

Security indicators in web browsers alert users to the presence of a secure connection between their computer and a web server; many studies have shown that such indicators are largely ignored by users in general. In other areas of computer security, research has shown that technical expertise can decrease user susceptibility to attacks. In this work, we examine whether computer or security expertise affects use of web browser security indicators. Our study takes place in the context of web-based single sign-on, in which a user can use credentials from a single identity provider to login to many relying websites; single sign-on is a more complex, and hence more difficult, security task for users. In our study, we used eye trackers and surveyed participants to examine the cues individuals use and those they report using, respectively. Our results show that users with security expertise are more likely to self-report looking at security indicators, and eye-tracking data shows they have longer gaze duration at security indicators than those without security expertise. However, computer expertise alone is not correlated with recorded use of security indicators. In survey questions, neither experts nor novices demonstrate a good understanding of the security consequences of web-based single sign-on.

Conference name:10th International Conference on Detection of Intrusions and Malware, and Vulnerability Assessment, DIMVA 2013 Date of Conference: 18-19 July , 2013 A poorly designed web browser extension with a security vulnerability may expose the whole system to an attacker. Therefore, attacks directed at "benign-but-buggy" extensions, as well as extensions that have been written with malicious intents pose significant security threats to a system running such components. Recent studies have indeed shown that many Firefox extensions are over-privileged, making them attractive attack targets. Unfortunately, users currently do not have many options when it comes to protecting themselves from extensions that may potentially be malicious. Once installed and executed, the extension needs to be trusted. This paper introduces Sentinel, a policy enforcer for the Firefox browser that gives fine-grained control to the user over the actions of existing JavaScript Firefox extensions. The user is able to define policies (or use predefined ones) and block common attacks such as data exfiltration, remote code execution, saved password theft, and preference modification. Our evaluation of Sentinel shows that our prototype implementation can effectively prevent concrete, real-world Firefox extension attacks without a detrimental impact on users' browsing experience. © 2013 Springer-Verlag.

Browser-side templating (BST) is a valid alternative for Web development, even when it comes to building accessible applications. BST processes templates in the browser using a JavaScript-coded engine, providing significant performance improvements and making model-view separation a reality. However, it also has significant drawbacks. BST's dependence on JavaScript affects its accessibility and hides the delivered pages' content from search engines, hampering Web visibility. The authors confront this dilemma and propose a technique that lets BST be accessible and semantically crawlable, while preserving its advantages. © 2006 IEEE.

Editors: Miguel P. Rocha, Juan Manuel Corchado Rodríguez, Florentino Fdez Riverola, Alfonso Valencia., ew analytical methodologies, like the so-called "next-generation sequencing" (NGS), allow the sequencing of full genomes with high speed and reduced price. Yet, such technologies generate huge amounts of data that demand large raw computational power. Many-core technologies can be exploited to overcome the involved bioinformatics bottleneck. Indeed, such hardware is currently in active development. We have developed parallel bioinformatics algorithms for many-core microprocessors containing 64 cores each. Thus, the MC64 web platform allows executing high-performance alignments (Needleman-Wunsch, Smith-Waterman and ClustalW) of long sequences. The MC64 platform can be accessed via web browsers, allowing easy resource integration into third-party tools. Furthermore, the results obtained from the MC64 include time-performance statistics that can be compared with other platforms., This work was supported by “Ministerio de Ciencia e Innovación” [BIO2009-07443-E and AGL2010-17316]; “Consejería de Agricultura y Pesca” of “Junta de Andalucía” [041/C/2007 & 75/C/2009]; “Grupo PAI” [AGR-248]; and “Universidad de Córdoba” [“Ayuda a Grupos”], Spain.

Before making a security or privacy decision, Internet users should evaluate several security indicators in their browser, such as the use of HTTPS (indicated via the lock icon), the domain name of the site, and information from extended validation certificates. However, studies have shown that human subjects infrequently employ these indicators, relying on other indicators that can be spoofed and convey no cryptographic assurances. We identify four simple security indicators that accurately represent security properties of the connection and then examine 125 popular websites to determine if the sites' designs result in correctly displayed security indicators during login. In the vast majority of cases, at least some security indicators are absent or suboptimal. This suggests users are becoming habituated to ignoring recommended security indicators.

Current web browsers are ill-prepared to manage execution of scripts embedded in web pages, because they treat all JavaScript code executing in a page as one unit. All code shares the same namespace, same security domain, and shares uncontrolled access to the same heap; some browsers even use the same thread for multiple tabs or windows. This lack of isolation frequently causes problems that range from loss of functionality to security compromises. ScriptSpace is an abstraction that provides separate, isolated execution environments for parts or all of a web page. Within each ScriptSpace, we maintain the traditional, single-threaded JavaScript environment to provide compatibility with existing code written under this assumption. Multiple ScriptSpaces within a page are isolated with respect to namespace, CPU, and memory consumption. The user has the ability to safely terminate failing scripts without affecting the functionality of still-functional components of the page, or of other pages. We implemented a prototype of ScriptSpace based on the Firefox 3.0 browser. Rather than mapping ScriptSpaces to OS-level threads, we exploit a migrating-thread model in which threads enter and leave the ScriptSpaces associated with the respective sections of the document tree during the event dispatching process. A proportional share scheduler ensures that the number of bytecode instructions executed within each ScriptSpace is controlled. Our prototype can isolate resource-hogging gadgets within an iGoogle Mashup page as well as across multiple pages loaded in the browser and still retain interactive response. Master of Science

The AMI Meeting Corpus is a multi-modal data set consisting of 100 hours of meeting recordings. It is being created in the context of a project that is developing meeting browsing technology and will eventually be released publicly. Some of the meetings it contains are naturally occurring, and some are elicited, particularly using a scenario in which the participants play different roles in a design team, taking a design project from kick-o' to completion over the course of a day. The corpus is being recorded using a wide range of devices including close-talking and far-field microphones, individual and room-view video cameras, projection, a whiteboard, and individual pens, all of which produce output signals that are synchronized with each other. It is also being hand-annotated for many different phenomena, including orthographic transcription, discourse properties such as named entities and dialogue acts, summaries, emotions, and some head and hand gestures. We describe the data set, including the rationale behind using elicited material, and explain how the material is being recorded, transcribed and annotated.

In this paper, we present the design and implementation of Ovid, a browser for Grid-related information. The key goal of Ovid is to support the seamless navigation of users in the Grid information space. Key aspects of Ovid are: (i) A set of navigational primitives, which are designed to cope with problems such as network disorientation and information overloading (ii) A small set of Ovid views, which present the end-user with high-level, visual abstractions of Grid information these abstractions correspond to simple models that capture essential aspects of a Grid infrastructure. (iii) Support for embedding and implementing hyperlinks that connect related entities represented within different information views (iv) A plug-in mechanism, which enables the seamless integration with Ovid of third-party software that retrieves and displays data from various Grid information sources, and (v) a modular software, design, which allows the easy integration of different visualization algorithms that support the graphical representation of large amounts of Grid-related information in the context of Ovid's views. © 2006 IEEE. 2006

The AMI Meeting Corpus is a multi-modal data set consisting of 100 hours of meeting recordings. It is being created in the context of a project that is developing meeting browsing technology and will eventually be released publicly. Some of the meetings it contains are naturally occurring, and some are elicited, particularly using a scenario in which the participants play different roles in a design team, taking a design project from kick-o' to completion over the course of a day. The corpus is being recorded using a wide range of devices including close-talking and far-field microphones, individual and room-view video cameras, projection, a whiteboard, and individual pens, all of which produce output signals that are synchronized with each other. It is also being hand-annotated for many different phenomena, including orthographic transcription, discourse properties such as named entities and dialogue acts, summaries, emotions, and some head and hand gestures. We describe the data set, including the rationale behind using elicited material, and explain how the material is being recorded, transcribed and annotated.

In this paper a system for goal directed integration of web services based on automated planning is presented. The increasing number of web services available on the net poses the problem of having efficient tools in order to integrate existing services for obtaining complex services which reflect user goals and needs. In this scenario, automated planning techniques represent promising components of such dynamical and evolutionary systems. In the proposed architectural model, web services and user goals are modeled as planning operators and goals, while the generated solution plans are used for directly generating web service scripts. An extended planning model based on the notion of output variable has been introduced in order to take into account of results produced by services invocations. A technique called semantic wrapper has been developed for modeling services as operators. In addition to a simple automation tasks, web services plans also optimise's the use of resources, and they provides useful information on variables bindings and the parallel structure of web services calls. The implementation of P4WS, a planner with output variables which demonstrated the model is described and experimental results are presented.

Net-dbx-G is a tool that utilizes Java and other World Wide Web tools as an interface to Grid services to help Grid application developers debug their MPI programs from anywhere in the Internet. Net-dbx-G is a source level debugger with the full power of gdb (the GNU Debugger), providing the user with full Grid functionality. The portability of the tool is of great importance as well because it enables the tool to be used on heterogeneous nodes that participate in an MPI enabled Grid environment. The users of our system simply point their browser to the Net-dbx-G webpage and authenticates to the Virtual Organization that they are part of. Having at their disposal the shared resources participating to that VO, they start debugging by interacting with the provided GUI environment of the tool. The users can dynamically select which MPI processes to view/debug. 35 42 Sponsors: Institute of Electrical and Electronics Engineers, IEEE Conference code: 63577 Cited By :6

The effectiveness of a content browsing system, based on semantic content analysis, user profiling, and adaptation, is discussed. The various aspects of electronic road are also described. An user's navigation path through the multimedia content is represented by an electronic road. Both the multimedia material and metadata, used by the recommendation modules, are included in the information stored in the system to provide the user with context-based browsing. 10 4 49 59 Cited By :15

The Wireless environment requires new type of services and new ways for structuring the needed content. Personalization comes into aid via the creation of personalized portals that directly tones down factors that break up the functionally of the Internet/wireless services when viewed through wireless devices factors like the "click count", user response time (the "choice" factor) and the size of the wireless network traffic. In this demo we present a flexible personalization system tuned for the wireless Internet taking into consideration user mobility and not only the user profile but the device profile as well. We demonstrate appropriate, to the wireless environment, metrics and our initial performance evaluation, using a real content provider's content, indicates improvement ranging from 33% to, for certain metrics, 60%. 792 794 Conference code: 61820 Cited By :1

Intermediaries are software entities deployed on Internet hosts of the wireline and wireless Web that intervene in the flow of information from clients to origin servers at the application level of the WWW. Intermediary systems fall under the general term of «middleware». Intermediaries represent a useful abstraction for the design and study of emerging software infrastructures for «next-generation» Web services. Their importance is increasing with the increasing demand for personalization, localization, and support for ubiquitous access over different physical media and protocols. We present an overview of a wide range of systems that can be described as intermediaries, classifying them in a number of broad categories according to their basic functionalities. Going beyond simple WWW proxies, we examine the requirements arising from the need to support personalization, mobility and ubiquity under high loads. We identify and refine a set of important properties and characteristics of intermediary systems. Based on these properties, we introduce a detailed taxonomy of characteristic systems and identify a number of key components of emerging intermediary infrastructures. © 2002 IEEE. 231 236 Sponsors: IEEE Communications Society IEEE Computer Society Conference code: 99371 Cited By :2

Net-Console is an integrated program development environment that can be used as a front-end for High Performance Computing (HPC) sites. It consist of an MPI-aware editor, an execution console, a debugger, monitoring tools and an account and file manager. Using Net-console the user is able to edit, execute, debug and evaluate the performance of parallel programs from anywhere in the Internet. The user interface is done through a normal Java-enabled browser. Net-console can also support processing through wireless and lightweight devices with the use of mobile agent technologies. The tools included in Net-console and their functionality, the languages used and the overall structure of the project are presented in this paper. 11 1 139 150

The undeniable popularity of the web makes the efficient accessing of distributed databases from web clients an important topic. Various methods for web database integration have been proposed but recently there is an increasing interest on those based on Java-based ones. This is due to the inherent advantages of Java, which supports platform independence and secure program execution, and produces a small size of compiled code. In this experimental paper, we evaluate all currently available Java-based approaches. These include Java applets, Java Sockets, Servlets, Remote Method Invocation, CORBA, and Mobile Agents. To this end, we implemented a Web client accessing a remote database using each of these approaches and compared their behavior along the following important parameters: (1) performance expressed in terms of response time under different loads, (2) transparency of communication expressed in terms of complexity of networking API, and (3) extensibility expressed in terms of ease of adding new components. 1 290 293 Sponsors: IEEE Conference code: 57861 Cited By :1

This paper describes Net-dbx, a tool that utilizes Java and other WWW tools for the debugging of MPI programs from anywhere in the Internet. Net-dbx is a source level interactive debugger with the full power of gdb augmented with the debug functionality of LAM-MPI. The main effort was on a low overhead but yet powerful graphical interface that would be supported by low bandwidth connections. The portability of the tool is of great importance as well because it enables us to use it on heterogeneous nodes that participate in an MPI multicomputer. Both needs are satisfied a great deal by the use of Internet Browsing tools and the Java programming language. The user of our system simply points his browser to the URL of the Net-dbx page, logs in to the destination system, and starts debugging by interacting with the tool just like any GUI environment. The user has the ability to dynamically select which MPI-processes to view/debug. A working prototype has already been developed and tested successfully. 1470 LNCS 181 189 Sponsors: Association of Computer Machinery International Federation of Information Processing Conference code: 98814 Cited By :3

Date of Conference: 27-30 June 2004 Conference name: IEEE International Conference on Multimedia and Expo (ICME), 2004 Detection and removal of commercials plays an important role when searching for important broadcast news video material. In this study, two novel approaches are proposed based on two distinctive characteristics of commercials, namely, repetitive use of commercials over time and distinctive color and audio features. Furthermore, proposed strategies for combining the results of the two methods yield even better performance. Experiments show over 90% recall and precision on a test set of 5 hours of ABC and CNN broadcast news data.