Conference Date: May 4 & 5, 1995 Location: Recording for the Blind,
Inc., Princeton, New Jersey USA Proceeding distribution date: May 1995

Proceedings distributed by:

Recording for the Blind, Inc. 20 Roszel Road Princeton, NJ 08540 Internet: research@rfb.org Phone: 609.452.0606 Fax: 609.520.7990

Preface

RFB encourages distribution of these Proceedings in any medium and permission for such copying is hereby given, so long as the proceedings are distributed in their entirety on a non-profit basis.

Print and disk copies are available from RFB at no charge upon request. TABLE OF CONTENTS

PART 1. Executive Summary PART 2. Overhead slides from opening session
PART 3. Working group reports PART 4. Concluding discussion and plans
PART 5. List of conference participants-(e-mail addresses included)

PART 1. EXECUTIVE SUMMARY

The Conference was funded by a grant from the Alfred P. Sloan Foundation.

Opening Session Presentations

Dr. Raman, developer of AsTeR, "Audio System for Technical Readings," updated conference participants on the advancements made by Dr. John Gardner, Dr. Bill Barry, and Dr. Randy Lundquist. These researchers from Oregon State University are funded by the National Science Foundation to develop AsTeR.

A panel presentation by Dr. Jan Engelen, Dr. Bill Barry, and George Kerscher, described the challenges of working with publisher source files. It was clear that significant gains can be made when structural information is present in data to be used by reading and writing systems. The panel concluded that focused research and development of emerging publishing systems would have the greatest long term benefit to persons using accessible reading and writing systems.

Text and figure descriptions of slides from these presentations are included in Part 2.

Working Groups

Digital Libraries, electronic archives of published works, are being designed by leaders in the publishing field. A second working group formed to collaborate with the designers of these digital libraries to ensure that the needs of persons with disabilities can be met. Jeri Uzzo from the IEEE and member of the Association of American Publisher's (AAP) "New Technologies Group" helped to pull together a strategy of collaboration for the coming year.

Throughout the conference, mathematical notation in computer files was a central focus. Of special interest was the proposed draft of HTML 3.0 and the mathematical notation therein. A third working group formed to examine the proposed draft in light of the math reading and writing systems presented to conference participants, and to discuss ways to strengthen the HTML proposed draft to better serve persons with disabilities.

Reports from the three working groups are included in Part 3.

Conclusions

One group will follow up on the HTML 3.0 math working group recommendations. The second will examine publisher archives as suggested by that working group. ICADD (The International Committee for Accessible Document Design) will also be involved.

Project teams and plans for ongoing collaboration are included in Part 4.

PART 2. OVERHEAD SLIDES FROM OPENING SESSION PRESENTATIONS

[The following 6 slides were shown by Dr. Jan Engelen (1-6)]

Slide 1

1. The European Union and its Research Programmes incl.:TIDE, 4th Framework, Antwerp Forum

2. Essentials of SGML, the Standard Generalized Mark-up Language

3. The basis of the TIDE MATHS project GUIB, GRIF

4. How to get to a working prototype of the interactive Maths workstation: Usability Choice of DTD User interface, audio rendering

5. Some preliminary conclusions

Slide 2

Slide 3

= an industrial support programme: stimulating companies to produce devices for the rehabilitation market

Therefore: *outcomes must be products, at least in prototype form *market and other economic issues are considered to be as relevant as the technical developments *work is done within consortia: temporary grouping of companies, research centers and user groups *Cultural differences between partners are real *work is organised in workplans that have very strict timing *Commercial groups are funded for 50% only *High pressure to "show" something

Slide 4

Slide 5

TIDE goes on *currently in planning stage *involvement of non-EU countries is possible (but without EU- funding) *about 30 new projects will start *no provision for rapid actions

Slide 6

Antwerp Forum

-collaboration with USA and Canada is stimulated -Task forces have been set up -major actions: * exchange of information: AZtech, TIDE's WWW server * mutual exchange of programme, project and partner data

[The following 18 slides were used by Dr. Tom Wesley (7-24)]

Slide 7

The key to increasing access:

* linking into commercial processes * allows explicit structure to be made available for browsing and navigation * harness scarce resources by using true international standards

* SGML is the major enabling technology

Slide 8

Much of our access to the printed word makes use of visual representations of logical structure which can be scanned rapidly:

* HEADINGS IN CAPITALS * bulleted indents * emphasised sections in italics * large print

Slide 9

LOGICAL STRUCTU

* accessible, but usable only with difficulty * navigation through large documents is almost impossible

Far better usability can be provided with document standards which encode logical structure explicitly

Slide 10

* allows the definition of the logical structure of (textual) documents * being used extensively in the publishing industry * is the basis of HTML (HyperText Markup Language) used on World Wide Web

Slide 11

* CAPS TIDE Project (Communication and Access to information for People with Special Needs) * ICADD (International Committee on Accessible Document Design) * Mathtalk (University of York) using LaTex

Slide 12

Judicious mixture of our needs and those of TIDE

* Based on SGML * Create a marketable product (not just tools) * Integration of blind and partially sighted implies sophisticated WYSIWYG system * Graphical User Interface (Windows)

Slide 13

* Grif, a French company based in Paris is the major European SGML workstation (software) vendor * WYSIWYG SGML editing system running on Unix, Mac and Windows * The underlying software of the Euromaths Consortium * An active contributor to SGML Open

Slide 14

* A major assistive technology manufacturer based in Germany * Active in GUIB (Graphical User Interfaces for the Blind) TIDE Project * Providing expertise (and systemQWindots) for dealing with Windows * Gerhard Weber is responsible

Slide 15

* University of York, UK (Alistair Edwards and Robert Stevens) * Katholieke Universiteit Leuven, Belgium (Jan Engelen and Bart Bauwens) * University College Cork, Ireland (John McCarthy, Helen Cahill, Carol Linehan) * University of Bradford, UK (Tom Wesley) * Electric Brain Company, UK (Mark Elsom-Cook)

Slide 16

Target users are secondary school children

* All children must learn mathematics * Many (most?) blind students don't * We are not primarily targeting mathematicians * Workstation should also be suitable for first year university students studying (non specialist) mathematics * A system for working with maths

Slide 17

Polynomials of arbitrary degree, fractions, arithmetic operators, rrlational operators, roots of arbitrary degree; Logarithms; Trigonometric functions; Calculus: differentials, integrals, delta notation, limits; Boolean calculus: conjunction, disjunction, implication operators, negation, existential and universal quantifiers

Slide 18

Ideally we would like to support any DTD that is (or will be) used to encode mathematics. However,

* limited resources suggested a choice had to be made * our requirements needed to be matched as closely as possible by the DTD

Slide 19

* Able to represent our functionality (and more) * Clear separation of syntactic structures * Possible to add semantic information * Availability of documents directly or by conversion * Ease of conversion to LaTeX * Ease of implementation

Slide 20

We studied (or became familiar) with:

* AAP * Grif * Euromaths * ISO 12083 * CALS * Elsevier * HTML 3.0 (early draft)

Slide 21

* A powerful and elegant DTD designed and used by mathematicians * Implemented on the Grif workstation * Clearly structured * Completely recursive * Easy to add optional semantic information * Fully functional public domain converters to and from LaTeX

Slide 22

* Integral (original slide shows four boxes in a two dimensional structure representing the integral symbol, the upper and lower limits and the integrand)

* Subscript (original slide shows five boxes in a two dimensional structure representing the main base and left and right subscripts and superscripts)

* Limit (original slide shows three boxes in a two dimensional structure representing the limit symbol, the limit and the expression)

Slide 23

We have used the SDAMAP construct to allow the capability of adding optional semantic attributes. We can then distinguish between

* Normal division * Integer division * Derivative * Partial derivative

Slide 24

We have defined an internal tree structure representation (in C) for processing purposes. This closely follows the Euromaths+ DTD but further distinguishes operators and operands in linear textual elements

* Available in C

* INMATH DTD

PRESENTATION 2: AsTeR Presented by: Dr. T.V. Raman

[The following 23 slides were used by Dr. Raman. (25-47)]

Slide 25 Introduction

AsTeR RFB Symposium On Math & Science Access --1995

T. V. Raman Cambridge Research Lab Digital Equipment Corporation

Slide 26 Objective of Talk

Set themes to capitalize on Work and expertize of conference invitees.

Look back to RFB-1994 And look forward to RFB-1995.

Slide 27 Outline

Slide 28

The next-generation reading machine.

Information is display independent.

This figures shows a box with the words "Electronic Documents" from the box is two arrows. One arrow is pointing to "LaTeX" which is pointing to "Print." The second arrow is pointing to "AsTeR" which is pointing to "Audio."

Different views of the same document.

Slide 29 What AsTeR Does

Given a LaTeX document,AsTeR

* Constructs a high-level representation. * Renders this using a default style.

Listener can:

* browse this representation. * Switch among predefined rendering styles. * Define new rendering styles.

Electronic books come alive with AsTeR!

Slide 30 Features

* A rich internal representation. * AFL -- Audio Formatting Language.

* A powerful browser. - Object orientation - multiple renderings * An upwardly extensible system.

Slide 31 Synopsis -- RFB 1994

* Port AsTeR to different platforms. * Influence electronic encoding formats. * Interface to incorporate all modalities.

Themes were assigned to working groups

Slide 32 Making AsTeR Widely Available

We have ported AsTeR to CLISP A public-domain Lisp for UNIX.

AsTeR now runs under Linux on a PC. AsTeR is being ported to MSWindows Franz-Lisp.

Since no speech front-end existed to UNIX I wrote one.

Slide 33 Electronic Encodings

This is a high-priority theme for 1995.

Slide 34 Unified Adaptive Interface

* Integrating visual and aural output. * Integrating aural and tactile output. * Tackling scientific graphs and plots.

Hope to have things to show in 1996.

Slide 35 Software Requirements

* Clisp --Public Domain-- on UNIX systems including: * Digital UNIX on the DECALPHa * Linux for Intel X86

* Lucid Common Lisp.

Ongoing ports:

* Franz Lisp under Windows. * Clisp under DOS/Windows.

All ports use an Emacs front-end.

Slide 36 Emacspeak -- Speech for UNIX

Since none existed, I wrote one.

Emacspeak--

* Speech enables Emacs. * Provides fluent feedback as you work. * A speech interface - not a screenreader.

AsTeR and Emacspeak share the Dectalk.

Slide 37 Hardware Requirements

* a 486 class machine or better. * At least 8MB of memory. * A Dectalk synthesizer. * A sound card.

Ongoing work: --AsTeR will use

* Software Dectalk on the DECALPHA. * Dectalk on the SoundBlaster.

Slide 38 Using AsTeR

Steps in reading with AsTeR

* Specify what to read. * Recognize structure and start rendering. * Interrupt rendering and browse. * Optionally switch rendering styles.

Listen -> Browse -> Styles -> Listen ->...

Slide 39 What you can read

AsTeR provides a consistent interface.

The slides for this talk form a document.

Slide 40 Browsing Mathematics

$$ a +\frac{b+c}{d+e} +x$$

$$ e^x =1+x + \frac{x^2}{2!} +\frac{x^3}{3!} + \cdots + \frac{x^n}{n!} + \cdots $$

Slide 41 Speaking Document Content

\begin{theorem} If $x$ has children $y_1,y_2,\ldots,y_m$ ordered by time linked into $x$, then $\rho(y_i)\geq i-2$. \end{theorem}

Using this theorem, we can show that the smallest possible tree of rank $n$ has $F_n$ vertices, where $F_n$ is the Fibonacci sequence: $F_0 = 1$, $F_1 = 2$, and $F_{n+2} = F_{n+1} + F_n$.

Slide 42 Proof Reading

You can preview parts of a document, eg.

A single equation

Or sit back and listen to the entire document.

Slide 43 Doing Math

Emacspeak & AsTeR Edit & Preview Write & Read Check & Validate Compute & Review

A unified interface.

Slide 44 Symbolic Calculator

The taylor series expansion of $\sin x$ is:

x - {x^3 \over 6} + {x^5 \over 120} - {x^7 \over 5,040}

Slide 45 Client/Server Interface

Permits an AsTeR server running at RFB

To be accessed using an WWW client.

Advantages:

* Copyrighted material can be protected at the server. * RFB can provide aural access to subscribers.

Prototypical Library Of The Future

Slide 46 Running AsTeR Remotely

Permits remote logins to a machine running AsTeR

And have the output sent back.

Slide 47 Conclusion

* Developing the outlined themes * Defining our goals for the year. * Developing a time-line for realizing these.

And the coming year Realizing the vision we develop

PRESENTATION 3: CONVERSION OF PUBLISHER FILES Panel discussion by Dr. Jan Engelen, Dr. Bill Barry and George Kerscher Presented by: George Kerscher

[The following 9 slides were used by George Kerscher. (48-56)]

Slide 48 ADVANTAGES OF USING PUBLISHER SOURCE FILES

* No verifying/correcting of scanned data

* Programmatic identification of structure

DISADVANTAGES OF USING PUBLISHER SOURCE FILES

* Time required to obtain complete files

* Files must be converted before editing

* Last minute changes in print copy may not be reflected in files

* May be more cost effective to scan

HOW MUCH TIME DOES IT REALLY SAVE

* Editing time can be cut by as much as 80%

* Some books require no hand editing

Slide 49 Revised May 1994

RFB E-Text Production

Slide 50

1. Identify file format

2. Find out if conversion tools already exist

3. Find out if you can get documentation on the file format

4. Identify escape character(s)

5. Lexically analyze the files -- identify all tags

6. Assign semantics to the tags

7. Select conversion tool - FastTag, OmniMark, FLEX, PERL, etc.

8. Write program that maps semantic meaning into tag set

9. Test results

10. Document the conversion process

Slide 51 QUICK AND DIRTY CONVERSION PROCESS

5. Identify only needed tags

6. Identify only special characters used in document

7. Select conversion tool

8. Write program that maps semantics to our tag set

9. Write notes about the types of problems editors should look for

10. Keep notes and conversion for possible future use

Slide 52 PROBLEMS WITH CONVERSIONS

* Bogus tags in files

* Same mark up may be used for a different purpose; in various books

* Files may not contain generated text -- Table of contents, index, etc.

* Sections may be missing

* Files may not reflect last minute changes in printed copy

* Tables may have been "cut" in by hand and are not present in files

* For subsequent editions, edition changes may be in separate files

* Math and other complex data may have been put in as a bitmap

Slide 56 PEOPLE POWER /FIXING ALL THE PROBLEMS

* Pull together files that have been converted

* Resolve any ambiguities in mark up

* Compare files with hard copy to check for completeness

* Correct the order of information

* Add missing parts -- scan or key

* Reconstruct tables

* Convert graphs to tables

* Create descriptions of graphics

[The following 4 slides were used by Dr. Jan Engelen. (57-60)]

Slide 57

Slide 5

some European results (no Maths yet)

* wordprocessing into Braille (via ITS, PCBraille) * conversion from EIF/SGML to Braille (Toulouse, RNIB) * contribution to ICADD: the SDA attributes

Slide 59

Fully automatic conversion is required

Mixture of different techniques * conversion of Oracle data into SGML steps: export of structure information, preprocessing of textual data, DI assembly

* reader uses DOS based SGML browser (CAPS/Sensotec)

Slide 60

* ad hoc (Digikrant) * using tools for HTML production (Author for Word, Cyberleaf, Grif Symposia and many others to come soon) * real time search using: at the user end: FORMs

at the service provider end: ODBC, Applescript

PART 3. WORKING GROUPS FINDINGS AND PLANS

HTML 3.0 MATH WORKING GROUP REPORT

1. We should communicate our conclusions to Dave Raggett as suggestions to improve his draft DTD. A cover memo follows this report.

2. Our ultimate objective is to provide for accessibility by persons with print disabilities, in order to facilitate interchange of ideas among all users -- including authors, publishers and readers of mathematics at all levels.

We are concerned that, as it presently stands, the DTD will be inadequate to meet these needs.

Specifically, we wish to see the following more fully realized in the HTML 3.0 DTD:

- The SGML syntax should be clear enough in itself so that the information given in comments is not required for the correct implementation of renderers.

- In alternate formats, such as speech and braille, it is sometimes necessary, or at least desirable, for semantic information to be available for rendering purposes. An example would be "dy/dx" considered as a derivative as opposed to a fraction. In the SGML context of HTML, it would be useful to emulate the SDAMAP facility of ISO 12083, even though it is recognized that users would most likely not use that optional annotation mechanism to its full potential.

- In mathematics, precise structure and semantics are often closely related. It is necessary, for example, to tell whether an index is left, right, or directly over or under the base element. The examples of the structures present in the well-proven LaTeX system, or DTDs that are essentially isomorphic, such as Euromath, could serve as a checklist of the desired structures.

3. We should express our willingness to enter into ongoing dialog and review of the DTD. Specifically,

- We will invite and encourage the International Committee for Accessible Document Design (ICADD) to undertake a review of the DTD, which would lead to a formal ICADD comment and endorsement.

- We will invite and encourage current projects and efforts involved in information access by people with print disabilities to build bridges and interfaces to test and prototype braille and speech outputs.

Cover Memo to Dave Raggett and the HTML 3.0 DTD Developers:

On May 4 and 5, 1995, participants in the Math Research Working Conference, held at Recording for the Blind, continued examination of the HTML Specification 3.0 description of mathematics notation. The goal of one of the three working groups at the conference was to identify suggestions for the math notation which would make it more usable when considering the needs of persons who are print disabled, including those who are blind and visually impaired.

Attached are the conclusions of this working group, as presented to the entire group of conference participants. We hope to enter into a productive and ongoing collaboration with you, and look forward to continued discussions.

USER INTERFACE WORKING GROUP REPORT

* Primary target platform is to be MS windows. * Provide integrated mechanism for reading, writing and doing math. * Include flexible design considerations: - Adaptable to user computer familiarity. - Adaptable to level of math sophistication. - Support of a variety of input/output modes. * Development of a stand alone solution to math access. - Current developments underway: Maths Project (European project) Science Access Project (Oregon State University) Mega Dots and others * Stand alone application to serve as a model and resource of access technology for mainstream applications. * Work to influence a math OLE specification. * Applications should be OLE compliant. * Client/Server to have the same look and feel as the stand alone implementations. * Must provide solutions which can be readily placed in the hands of end users.

Display Modes

The working group defines an accessible math application to support the following display modes. * Speech * Non-speech audio * Braille; hard copy and online. * Visual display; must include adaptive technology (ie. screen enlargement and enhancement) * Incorporate other technology as available. - DotsPlus - Full screen tactile displays - Directional audio - ????

Action Items

Progress on the following items will be demonstrated by May 1996.

* Provide information on human interface and usability issues to those groups working actively in this field. This is to be done through the mailing list of participants at this conference. Robert Stevens will direct this effort.

* Explore development of creating an authoring tool for structured figures. This years conference focused on the representation of expressions. We would like to see what we have learned here broadened to include more abstract document elements. This is to include such constructs as data plots and charts, block diagrams, flow charts and illustrations. Doug Forer will direct this work.

* Develop an accessible graphing calculator as a PC application. Current high school and college math courses rely heavily on the use of calculators with the ability to plot and display expressions.These are completely inaccessible to the visually impaired students. An accessible alternative will be developed under the Science Access Project at Oregon State.

DIGITAL ARCHIVE WORKING GROUP REPORT

Goal: Determine the attributes we need in publisher's DTD. Goal: Modify the DTD accordingly. GOAL: Get them to put attributes into document instances.

* = Sub-project - = Task

* Project team development - Set up e-mail distribution list RESPONSIBLE: Jan Engelen DATE: May 15, 1995 - Distribute procedures to join project team RESPONSIBLE: Jan Engelen DATE: May 15, 1995

* IEEE Transfer of Technology - Non-disclosure and transfer of information agreement RESPONSIBLE: DATE: - Produce list of information needed. (flow chart, SGML DTD, Sgml document instances, conversion software, documentation, postscript output files) RESPONSIBLE: George Kerscher DATE: - Deliver requested IEEE information RESPONSIBLE: Jeri Uzzo DATE: - Distribution of IEEE information to cooperating parties. RESPONSIBLE: George Kerscher DATE: - Test the HTML 3.0 output of IEEE files RESPONSIBLE: DATE: - See if more semantics can be obtained from the original files. RESPONSIBLE: DATE: - Write report making recommendations to IEEE. RESPONSIBLE: DATE: - Establish ongoing cooperation in the development of their delivery system. RESPONSIBLE: DATE:

* Promote the use of HTML 3.0 for K-12 publishing - Report to Texas Commission on the success of HTML 3.0 RESPONSIBLE: DATE:

* Survey of College and Scholarly Works divisions of AAP - Obtain previous Survey RESPONSIBLE: Jeri Uzzo DATE: - Create our own survey (include the results and example of the IEEE work.) RESPONSIBLE: DATE: - Get address from AAP, SGML Open, SGML Forum of New York RESPONSIBLE: DATE: - Distribute the survey RESPONSIBLE: George Kerscher DATE: - Identify other publishers for cooperative agreements RESPONSIBLE: DATE:

* Establish cooperative agreements with other organizations - ACM(Jerry Murray) Jeri Uzzo: DATE: - AMS (Bob Kelly) RESPONSIBLE: Helmut Juergensen DATE: - APS RESPONSIBLE John Gardner DATE: - AIP RESPONSIBLE: John Gardner DATE:

PART 4

CONCLUDING DISCUSSION AND PLANS

The discussion emphasized the importance of influencing HTML 3.0 and Digital Archives. Everybody agreed that these developments would profoundly impact access to math and science for generations to come. Two project teams evolved from the discussions.

Tom Wesley agreed to lead a team that would carry out the HTML Working group recommendations. The others that agreed to participate in the project are: Charles Halperin-Hamu, Al Blank, T.V. Raman, Joe Sullivan, and George Kerscher.

The Digital Archive discussion looked at the problem of closely examining publisher's archives. It will require a great deal of time to report to publishers about what should be done to improve their system for use by persons with disabilities. Several people volunteered to be on this project team. They are: Jan Engelen, David Holladay, Lloyd Rasmussen, Helmut Juergensen, T.V. Raman, Joe Sullivan, and George Kerscher. Jeri Uzzo, in the working group discussion, agreed to work on tasks associated with the AAP and her publisher, IEEE.

ICADD is the natural organization to further develop these projects, and we all agreed that ICADD should be involved. All participants were enthusiastic about the progress and accomplishments from the conference, and expressed interest in meeting again next year.

PART 5. CONFERENCE PARTICIPANTS

Randy Lundquist, Oregon State University e-mail: randyl@physics.orst.edu

Bill Barry, Oregon State University e-mail: wab1@physics.orst.edu

Robert Stevens, University of York e-mail: robert@minster.york.ac.uk

Jan Engelen, Katholieke Universiteit Leuven e-mail: engelen@cc1.kuleuven.ac.be

T.V. Raman, Digital Equipment Corporation e-mail: raman@crl.dec.com

Tom Wesley, University of Bradford e-mail: T.A.B.wesley@bradford.ac.uk

Al Blank, The College of Staten Island e-mail: U15430@f.nersc.gov

Helmut Juergensen, University of Western Ontario e-mail: helmut@uwo.ca

Tom Kahlisch, Dresden University of Technology e-mail: kahlish@iis350.inf.tu-dresden.de

Rich Cox, AT & T Bell Lab., RFB TAC Chairman e-mail: rvc@research.att.com

Lloyd Rasmussen, NLS for the Blind & Physically Handicapped e-mail: lras@loc.gov

Beth Goodrich, AAAS e-mail: bgoodric@aaas.org

John Boyer, Raised Dot Computing e-mail: 76025.1265@compuserve.com

Joe Sullivan, Duxbury Systems Inc. e-mail: duxbury@world.std.com

David Holladay, Raised Dot Computing e-mail: dnavy@well.com

Jeri Uzzo, IEEE e-mail: juzzo@ieee.org

Charles Halperin-Hamu, InfoDesign Corporation e-mail: charlie@idc.com

Gunthild Vogel, Dresden University of Technology e-mail: vogel@iis350.inf.tu-dresden.de

Jim Fruchterman, Arkenstone e-mail: jim@arkenstone.org

Charles Davis, Educational Testing Service e-mail: cdavis@ets.org

Gary Day, NSA e-mail: grday@afterlife.ncsc.mil

Doug Forer, Educational Testing Service e-mail: dforer@ets.org

Chris Brooks, RFB e-mail: cwbrooks@rfb.org

George Kerscher, RFB e-mail: cbfb_gwk@selway.umt.edu

Bill Robinson, RFB e-mail: robinson@rfb.org

Cindi Krnac, RFB e-mail: ckrnac@rfb.org

Return to SEM Page

Home