Annotating educational discussion boards to help students who are blind
ABSTRACT Educational discussion boards are growing in use as they help students share knowledge and doubts in a working/studying environment. Although these interactive tools are highly effective in learning environments, their usability by blind students is very poor. In this paper we develop techniques to improve accessibility of educational content for students who are blind. Threads of messages in discussion boards evolve with new postings, thus just by investigating the subject headings or contents of earlier postings in a message thread, students may not be able to guess the contents of the postings deeper in the hierarchy. In order to overcome the navigation obstacle for users, it is essential to develop techniques that help identify how the content of a discussion board grows. We develop a technique to organise messages in a message board, by automatically classifying and annotating postings to guide users through discussion segments relevant to their navigational goals. she is currently an Associate Professor. Her initial contributions to computer science were in the area of logic programming and artificial intelligence, specifically in the semantics of negation, and in the abductive extensions of logic programs.
- [show abstract] [hide abstract]
ABSTRACT: Users now have access to unprecedented amounts of media data, but it is increasingly difficult to integrate relevant media from multiple and diverse sources. The functioning of a multimodal integration system requires metadata, such as ontologies, that describe media resources and media components. Such metadata are generally application dependent and this can cause difficulties when media needs to be shared across application domains. There is a need for a mechanism that can relate the common and uncommon terms and media components. In this paper, we develop an algorithm to mine and automatically discover mappings in hierarchical media data, meta-data, and ontologies, using the structural information inherent in these types of data. We evaluate the performance of this algorithm for various parameters using both synthetic and real data collections and show that the structure based mining of relationships provides high degrees of precision.
- [show abstract] [hide abstract]
ABSTRACT: In this paper, we present the architecture and describe the functionality of an Intelligent Tutoring System (ITS), which uses an expert system to make decisions during the teaching process. The expert system uses neurules for knowledge representation of the pedagogical knowledge. Neurules are a type of hybrid rules integrating symbolic rules with neurocomputing. The expert system consists of three components: the user modelling unit, the pedagogical unit and the inference system. The pedagogical knowledge is distributed in a number of neurule bases within the user modelling and the pedagogical unit. Another important component of the ITS, for both its development and maintenance, is its knowledge management unit, which provides knowledge acquisition and knowledge update capabilities to the system, that is, offers expert knowledge authoring capabilities to the system.Expert Syst. Appl. 01/2004; 26:477-492.
- [show abstract] [hide abstract]
ABSTRACT: Recently, there is an interest in using associations between web pages in providing users with pages relevant to what they are currently viewing. We believe that, to enable intelligent decisions, we need to answer the question "for a given set of pages, find out why they are associated". We present a framework for reasoning about Web document associations. We start from the observation that the reasons of the web page associations are implicitly embedded in the content of the pages as well as the links connecting them. The association reasoning scheme we propose is based on a random walk algorithm. This algorithm can take both link structure and contents into consideration and allows users to specify a focus. We then show how the proposed algorithm, combined with a logical domain identification technique, can be used for web site summarization and web site map construction to help users navigate through complex corporate sites. We see that, to achieve this goal, it is essential to recover the web authors' intentions and superimpose it with the users' retrieval contexts in summarizing web sites. Therefore, we present a framework, which uses logical neighborhoods, entry pages, and associations of entry pages, in creating context-sensitive summaries and maps of complex web sites.Data Knowl. Eng. 01/2002; 43:121-150.
Int. J. Cont. Engineering Education and Life-Long Learning, Vol. 17, Nos. 4/5, 2007
Copyright © 2007 Inderscience Enterprises Ltd.
Annotating educational discussion boards to help
students who are blind
Maria Luisa Sapino*
Dipartimento di Informatica,
Università di Torino, Torino, 10149, Italy
K. Selçuk Candan and Jong Wook Kim
CSE Department, Arizona State University,
Tempe, AZ 85287, USA
E-mail: firstname.lastname@example.org E-mail: email@example.com
Dipartimento di Informatica,
Università di Torino, Torino, 10149, Italy
Abstract: Educational discussion boards are growing in use as they help
students share knowledge and doubts in a working/studying environment.
Although these interactive tools are highly effective in learning environments,
their usability by blind students is very poor. In this paper we develop
techniques to improve accessibility of educational content for students who are
blind. Threads of messages in discussion boards evolve with new postings, thus
just by investigating the subject headings or contents of earlier postings in a
message thread, students may not be able to guess the contents of the postings
deeper in the hierarchy. In order to overcome the navigation obstacle for users,
it is essential to develop techniques that help identify how the content of a
discussion board grows. We develop a technique to organise messages in a
message board, by automatically classifying and annotating postings to guide
users through discussion segments relevant to their navigational goals.
Keywords: website accessibility; navigation in message boards; rule-based
Reference to this paper should be made as follows: Sapino, M.L.,
Candan, K.S., Kim, J.W. and Antonelli, F. (2007) ‘Annotating educational
discussion boards to help students who are blind’, Int. J. Continuing
Engineering Education and Life-Long Learning, Vol. 17, Nos. 4/5,
Biographical notes: Maria Luisa Sapino got her MS and PhD Degrees in
Computer Science at the University of Torino, where she is currently an
Associate Professor. Her initial contributions to computer science were in the
area of logic programming and artificial intelligence, specifically in the
semantics of negation, and in the abductive extensions of logic programs.
Annotating educational discussion boards to help students who are blind 295
Since mid-1990s she has been applying these techniques to the challenges
associated with heterogeneous and multimedia data management. She
developed novel techniques for similarity based information retrieval, and she
is currently working on web accessibility for users who are visually impaired,
and on temporal aspects in distributed multimedia presentations in the presence
of resource constraints.
K. Selçuk Candan is an Associate Professor at the Department of Computer
Science and Engineering at the Arizona State University. He received his PhD
in 1997 from the Computer Science Department at the University of Maryland.
He has worked extensively on the integration of heterogeneous multimedia
information. His current research include data processing and indexing for
sensory data, distributed data management for efficient dynamic content
delivery, and adaptive information technologies for helping individuals who are
blind. He has published over 80 papers in respected venues. Most recently, he
is an editorial board member of the Very Large Databases (VLDB) journal and
the ACM SIGMOD Digital Symposium Collection (DiSC).
Jong Wook Kim is a PhD student at the Department of Computer Science and
Engineering, Arizona State University, AZ, USA. He received his BS Degree
from Korea University in 1998 and his MS Degree from KAIST in 2000.
Before joining ASU, he conducted research at Digital Media Research Lab in
LG Electronics. His primary research interests are in the areas of web data
mining, information retrieval, and database systems. In particular, his current
research concentrates on indexing and mining for non-atomic web data, such as
a navigation hierarchy and a discussion boards, and application of these
techniques for development of navigational helps for individuals who are blind.
Fabrizio Antonelli received his degree in Computer Science from the
University of Torino in 2005. In 2003, he earned an internship at Fiat Avio,
where he worked within a European project for knowledge management for
airplane engine factories. In 2005, he visited the Department of Computer
Science and Engineering at Arizona State University and contributed to the
iCare project for improving accessibility of educational websites and tools.
He is currently a researcher at the Telecom Labs, Italia.
Many organisations (such as companies, universities, schools, e-learning societies) are
developing online tools to enable their users share and exchange information. One of the
most important tools in this category is the discussion board, where users can post
messages to ask a question, to provide answers to other’ questions, or to post an
announcement (Figure 1). In general, it is very common for discussion boards to contain
postings about several different topics at the same time (within a single thread or across
multiple threads) and it is difficult for a user who is blind to locate a posting relevant to
her task. Suppose a user who is blind wants to know if anybody knows the answer to a
specific question or can provide suggestions about a specific topic. It would be very
convenient for the user, if the system could assist by providing her with a list of postings
matching her interests (based on the criteria she gave), without making her scan the
M.L. Sapino et al.
Figure 1 An example discussion board tree
With the aim of reducing the navigational load of blind students, we are developing a
software interface, called iCare-Assistant, which provides context- and task-dependent
navigational guidance when accessing online educational materials that are already
available for the use of sighted students. iCare-Assistant provides a transparent
interface between an existing educational system (such as Blackboard, a commercial
educational software (http://www.blackboard.com)) and the blind student, and adds
context and keyword-based navigation (query) facilities to locate the desired information
with a few interactions only, reducing the navigational load in accessing information.
State-of-the-art browser-based interfaces and existing navigational helps, such as site
maps and visual cues, alleviate this load for only sighted users and are generally not
applicable to dynamically growing content. Instead, we employ transparent guidance and
dynamic adaptation techniques (Donderler et al., 2003, 2004) in iCare-Assistant to help
students without sight. Such dynamic adaptation and guidance require an understanding
of the inherent, yet implicit, structures of the content at the educational websites.
Text mining techniques (Cohen, 1996; Liu et al., 2003) have been successfully
developed and applied in many business intelligence applications. Discussion records,
however, cannot fully benefit from these techniques, because of their subtle, often
implicit, dependencies upon each other. It is often the case that an individual posting
carries a very limited amount of information (which makes text mining techniques of
very little use on such postings), whereas a collection of interrelated postings in the
discussion board forms a context, which carries more information, if properly identified.
Moreover, standard text mining techniques and indexing solutions do not apply to
message boards since the collections of messages are highly dynamic. The challenge, in
this case, is the discovery and organisation of such information possibly coming from
different messages, given the fact that:
Annotating educational discussion boards to help students who are blind 297
several different topics can be discussed in a single message, whose title/subject is
not necessarily required to match the message content
different aspects of the same topic may be found in various postings.
Contributions of this paper
In this paper, we address the problem of knowledge discovery and presentation from
message-boards. More specifically, we focus on the challenges associated with accessing
educational discussion boards, i.e., discussion boards used by teachers, assistants, and
students, to exchange information. We concentrate on the problems of classifying the
posted messages and discovering different relationships among messages posted on the
board. Based on the resulting classification, appropriate indexing techniques can be
developed to improve the information accessibility. In particular, we develop techniques
to improve accessibility for students who are blind, and therefore cannot benefit from the
visual help offered by well engineered websites.
The logical organisation of the messages is described by means of a labelled tree,
whose nodes are associated to the postings and whose labelled edges characterise
the different inter-dependency relations existing between messages. The labelling
of the messages is realised by means of a rule-based system (JESS), which associates
(possibly multiple) scores to the automatically extracted interdependency relations
between pairs of messages. Among all possible labels for a given message, the ones with
the highest score are chosen as the assigned classification for the message. One of the
advantages of the use of a rule-based system is its dynamic adaptation to the board
content: at any point in time, the arrival of new messages might fire previously inhibited
rules and induce an appropriate revision/update of the classification for some message, on
the basis of the newly added information.
The paper is organised as follows. In Section 2 we give a brief overview of the related
work. In Section 3 we introduce the architecture of the navigation support module.
Section 4 describes the method we use to represent the information associated to the
messages, both content and metadata. The rule-based classification module is presented
in Section 5. Section 6 presents usage strategies for the module, and Section 7 illustrates
experimental results. Finally, Section 8 contains concluding remarks.
2 Related work
In this section, we present the related work in the domains of adaptive and assistive
web and educational technologies, adaptive hypermedia, and message classification and
Accessible educational tools
Recently, there has been an increase in the internet-based delivery of course
materials, even when courses themselves are delivered in classrooms. Blackboard
(http://www.blackboard.com), for example, provides software and services to schools,
colleges, universities, and other educational institutions. While it is involved in various
accessibility related projects, such as Web Accessibility In Mind (WebAIM) and Standards
For Accessible Learning Technologies (SALT), these attempts and projects do not directly
address the issue of navigational overload posed on students. Instead, most current
M.L. Sapino et al.
technologies aim to make a given single page accessible. Technologies commonly relied
upon by the users with visual impairments include screen readers (JAWS, WindowEyes),
screen magnifiers (Magnum, ProVi-sion32, ZoomText), voice recognition software,
hypermedia-to-hypertext transformers (DragonNS, IBMViaVoice), and refreshable
Braille displays (ALVA, PBraille).
Adaptive hypertext and hypermedia
Adaptive hypermedia relies on two different but complementary methods, namely
adaptive presentation and adaptive navigation (Brusilovsky, 2001). Adaptive presentation
is manipulation of content fragments in a hypertext document. Order of fragments can be
changed, or fragments can be made invisible or less visible within a page. Adaptive
navigation support, on the other hand, is the manipulation of links. Direct guidance, link
sorting, link hiding, link annotation, link generation, and map adaptation are the
techniques used. Detailed discussion of all these approaches, both for adaptive
presentation and adaptive navigation support, can be found in Brusilovsky (1996, 2001)
and Cavanaugh (2002).
Page-accessibility research includes (Huang and Sundaresan, 2000; Mukherjee et al.,
2004; Pontelli et al., 2002; Ramakrishnan et al., 2004; Takagi et al., 2002). Takagi et al.
(2002) focuses on segmentation and annotation of a given page based on accessible
layouts manually predefined using an annotation editor. Huang and Sundaresan (2000)
also transforms a given page to render it more accessible. In particular, the
transformations may include splitting a single page into multiple units guided through an
index. Pontelli et al. (2002) provides a contextual graph for navigation within the
segments of a given page. Bookmarking and dialog-based navigation through page
segments are supported in Mukherjee et al. (2004) and Ramakrishnan et al. (2004).
As opposed to these techniques, our goal is to provide navigational assistance through a
dynamically (i.e., by multiple authors) generated and diversely (i.e., unpredictably) linked
collection of information units, such as messages, course pages, and notes.
Researchers in the AI community have developed web navigation tour guides, such as
Web Watcher. Web Watcher utilises user access patterns in a particular website to
recommend users proper navigation paths for a given topic. Adaptive hyperbooks,
such as KBS (Henze and Nejdl, 2000), and guidance systems, like TANGOW
(Carro et al., 1999) and ML Tutor (Smith and Blandford, 2003), take into account tasks
and user needs, profiles, and access patterns while adapting for learners. Hatzilygeroudis
and Prentzas (2004) describes how intelligent tutoring systems can benefit from hybrid
knowledge representation formalisms (neurules, which integrate symbolic rules with
neural networks based approaches) in classifying users, giving pedagogical decisions, and
adapting the teaching material. More specifically, rule conditions are assigned
significance factors, while rules are assigned bias factors. Both of these are parameters in
the computation of the activation value associated to the rule. In this paper, we also
benefit from a rule-based system for achieving modular, incremental classification and
annotation of educational material.
Classifying and indexing messages
The problem of classifying and indexing messages in a message board is strongly
related to the one of classifying and indexing web documents or pages.
One approach to organising web query results based on available web structure is topic
distillation proposed in Bharat and Henzinger (1998), Chakrabarti et al. (1998) and