Catálogo de publicaciones - libros

One very important aspect of computer programming is reading source code. Whilst this is a relatively simple process for sighted programmers, for blind computer programmers this presents a significant problem. Navigating through and comprehending often thousands of lines of code can be time consuming and difficult. Current development environments have many features that aid the reading of source code for sighted users, however, most of these features are visual in nature and are not translated well into audio by general screen-reader applications. Research has suggested that the use of fisheye views could aid navigation and enhance performance in program comprehension activities for blind programmers. This paper reports the results of a study into creating a better fisheye view of Java source code, by improving the method used to determine each line’s ’global importance’ or ’Level of Detail’ (LOD). The traditional LOD determination method uses only the indentation level of a line to calculate its overall importance. This paper describes the results of the study, and suggests some of the issues which may need to be considered in developing an improved LOD calculation for programming source code.

The wide-spread applications of document digitization have lead to the use of structured digital representation methods such as the XML language. Extraction methodologies for the formatting metadata can be used on such structured documents for enhancing their accessibility, including augmented audio representation of documents. To the best of our knowledge, an effort has yet to be made to produce an automatic extraction system of semantic information of the document formatting, solely from document layout, without the use of natural language processing. In this study a corpus of XML representations of several issues of a Greek newspaper is used in order to create and evaluate a semantic classifier of text formatting, based on Bayesian Networks.

We present the Winsight system, a Windows-based software system for completely automatic translation of Nemeth Braille code to The Winsight system takes hard copy Braille input containing Mathematics (written in Nemeth Braille code) and text (written in contracted Braille) via a scanner, performs image recognition and analysis of the scanned file to generate the ASCII Braille file, automatically separates Nemeth Braille coded expressions and contracted Braille text, backtranslates them to math expressions and text respectively, and produces a print output file in pdf format containing the result of backtranslation. The Winsight system comes with tools that allow users to manually intervene during each step, if they desire, to fix any errors reported by the system or seen by the user. In this paper we give an overview of the Winsight system.

In this paper we present research results and discuss recent directions concerning the structural analysis of science texts and cognitive aspects of text elements, such as causal relationships between text elements. We outline the process of structuring text material for the design of dialogue activities for science text comprehension in the educational environment of ReTuDiS (Reflective Tutorial Dialogue System). The system supports text comprehension using reflective dialogue activities, adapted to learners of different levels and its authoring tool, ReTuDiS supports users in authoring dialogue activities.

We present the VoxBoox system, a system for making digital books accessible to visually impaired individuals via audio and voice. This is accomplished by automatically translating a book published in HTML to VoiceXML, and then further enhancing this VoiceXML rendering of the book to enable listener-controlled dynamic aural navigation. The VoxBoox system has the following salient features: (i) it leverages existing infrastructure since the book that is to be made accessible need only be published digitally using HTML on the visual Web, (ii) it is based on accepted Web standards of HTML and VoiceXML and thus books can be made accessible inexpensively, and (iii) it is user-centered in that the listener (the user) has complete control over (aural) navigation of the book. In this paper, we present details of the technologies that make the VoxBoox system possible, as well as the details of the system itself. A prototype of the VoxBoox system is operational.

Our paper discusses research on using multimodal interaction to improve usability of mobile data entry. This can benefit users of mobile technologies such as mobile workers, construction crews, and students collecting data during field trips. The evaluation of a mobile multimodal application for construction field work was conducted in the laboratory. Designing this laboratory experiment required a detailed definition of the type of device to be used, user requirements, including interaction techniques and usage context. During this work we discovered a need to develop a generic evaluation framework to assist in the selection of mobile technology and interaction techniques for further testing. The development of such a framework for municipal inspections with a focus on field inspections is discussed. This paper presents examples of using our framework to gain a better understanding of the contextual aspects influencing the feasibility of using mobile technology in the field.

The Web has turned out to be an interactive and complex information seeking environment where apart from text, layout and visual cues play a significant role in its efficient use. In this paper we present a semantic web based framework for communicating this visual meta-information to blind users who miss it. Based on annotations of web pages blind users are given a set of browsing shortcuts within a web page and this way they can move easily to the desired point of the page. In addition to this mechanism we enhanced these annotations with spatial information and we provide blind users with a spatially enriched variation of the browsing shortcuts mechanism. To test its effects on web browsing of blind users we conducted an experiment with 15 users participating. The results indicate that there is no significant improvement when using the spatially enriched variation of the shortcut feature but there is a decrease in time and keystrokes when the browsing shortcuts mechanism is used alone.

This study was to design an alternative on-screen keyboard and evaluate the efficacy of this innovative layout design for people with severe physical disability. The matrix keyboard layouts was designed based on human computer interaction. A repeated experiment was performed to compare the speed and accuracy of text entry with point-and-click input method between the matrix on-screen keyboard and the Windows XP QWERTY virtual keyboard. Data analysis revealed that the matrix on-screen keyboard provided better performance for the participant. The result also indicates that layout adaptation assessment is a valid tool to confirm proper layout size for users and that alphabetic order is better than QWERTY order for a novice user to learn a new on-screen keyboard. A usability study was undertaken to evaluation the performance of the double click instead of the point-and-click plus Shift key input method. The possible causes of the results and suggestions for further studies are discussed.

The decoding performance of information extraction is important for data analysis and decision making. The objective of this study is to evaluate the effects of display types (TYPE) shown in web pages using different techniques of rapid serial visual presentation (RSVP) on decoding performance for different information extraction tasks (TASK). Twenty-four subjects (12 males, 12 females) participated in the information extraction tasks. Two factorial designs with blocking experiment were used to collect the response time data. In Experiment 1 (TASK includes finding exact value and identifying the trend), the results showed that both the 2-way interactions of RSVP and TASK as well as TYPE and TASK would significantly affect mean response time. In Experiment 2 (TASK includes finding maximum and comparing two values), the results showed that the 3-way interaction of RSVP, TYPE, and TASK would significantly affect mean response time.

This paper presents a large scale study of user search behavior on a popular online dictionary website. Our goal is to understand the current status of online dictionary, especially English-Chinese dictionary user search behavior by analyzing 10 million queries on Dict.cn website during the period form late 2006 to early 2007. We believe our findings will help traditional dictionary publishers and online dictionary providers to improve their existing products and services, and further develop innovative services to better serve the unique needs of their users.