robertj@tekgen.bv.tek.com (Robert Jaquiss) (05/13/91)
Index Number: 15524
III Chapter 3
REQUIREMENTS ANALYSIS & SPECIFICATION
III.A Introduction
We begin our requirements analysis with a brief look at the
reading and writing environment of "an educated person" who does
not have a disability.
An educated person opens a door and invites a friend into
the private library. The friend has never been in this room
before and gazes around. The room is not too large, but
contains shelves of books from floor to ceiling. A
strategically placed, large, oak desk dominates the room.
In addition to the desk chair, a comfortable sofa is placed
so the lamp sheds perfect light on the environment.
The two people take time to examine the volumes in the room.
The library is divided into several areas The literature
section contains great works that must be read by any person
who claims to be educated. Another section contains the
professional titles that relate to this persons chosen
profession. The next section holds reference materials
including, an encyclopedia, the Oxford English Dictionary, A
modern English dictionary, and technical reference
materials. The final section is entirely devoted to
periodicals. This section primarily contains magazines, but
a few old newspapers can be seen. This person's private
library is not unlike the large university library across
town.
Upon examining the desk we find that a manuscript is under
production. Text with drawings and diagrams have been
artfully produced with the assistance of a word processor.
We find that a manual of style is open and has been
consulted for a technicality. The entire well formatted
manuscript is visually appealing. The author realizes that
if the publisher accepts their document, many visual changes
will be made before it appears in print.
The problem, of course, is that this scene could not be acted out
by a person with a print-disability. Our goal is to produce a
reading and writing system that emulates the picture painted
above. We are attempting to create an inviting, comfortable,
easy to use system for all individuals with a print-disability.
In the same way that we described the home library above, we will
attempt to describe the same electronic library that we will
specify in this chapter.
An educated person opens a door and invites a friend into
the private library. The friend has never been in this room
before and gazes around. The room is not too large, but
contains a computer based system designed for this person
with a print-disability. There are three output devices
attached to this computer and two input devices. One output
device is a normal monitor and one input device is a
traditional keyboard.
The person with a print-disability moves to the system and
inputs the word "library." Immediately three options are
presented. The reference section can be selected, the
periodical works can be chosen or individual titles can be
browsed. They select the individual titles and several
categories are presented. Since both these persons are
interested in scientific work, they decide to examine
together a college textbook.
They select a mathematical title in the electronic library.
All three output devices show the table of contents. It is
easy to get an overview of the entire book from this point.
Chapter 3 looks interesting and with a simple movement the
computer brings the opening portion of that chapter to the
three output devices. This is mathematical information and
each of the three output devices displays the information
from the book in three different ways. Both people are
examining the same material, but the representation of that
material is different on each output device. On the next
electronic page, an elaborate graph appears. The person
with a print-disability at this point pulls up a simplified
version of that drawing. Besides the simplified drawing a
textual description accompanies the simplified graphic. The
simplified graphic conveys an overview of the figure and
then we return to the complex drawing and read the
accompanying textual description. From this point they turn
to the index to look up something. It is easily found by
specifying a search string in the index. Another simple
movement brings all output devices to the point in the text
indicated in the index.
As they continue to browse the electronic library, one
thing becomes clear. The books, reference materials and
periodicals that are available to the sighted community have
counterparts in the print-disabled electronic version. What
is striking is that at any point the print-disabled person
can relate to the fully sighted person in their chosen
medium through the computer representation. That is, the
print disabled person may be using one input or output
device and the same information is manipulated and presented
in a different way simultaneously. Writing is handled in
much the same way. What the print-disabled person creates
is represented in different modalities on the attached
output devices. The person with a print-disability may
input data in one way, and the person who is normally
sighted can read the information in the customary way.
The task is to specify the reading and writing system so that an
effective data representation design can be constructed. At some
point all requirements of the reading and writing system must be
recorded. Presently we restrict our system requirements analysis
to the factors that will effect the data representation
requirements and design. The electronic library described above
is a good starting point for the specification of the reading and
writing system. The software controlling the electronic library
manipulates the data represented in the electronic book and
outputs that information in a variety of ways. We must generally
describe those features of output that are dependent on the data
representation. Then we must design a data representation system
that can effectively support those different methods of output.
An example may serve us here.
The scene above describes the users going to the index of
the book, looking up a word, and jumping to that portion of
the book referenced in the index. Assuming that the index
itself is represented in ASCII characters, finding a
particular item in the index is trivial and requires no
specialized data representation. The move to a particular
location in the textbook found in the index does require
special notation in the original data representation. In
fact the index was probably created by having key words
marked in some way in the text. If the user wants to cut
and paste a portion of text under examination, that function
can be conducted without regard to the data representation
itself. One simply copies the data from the book and moves
it to another location in software. There is no need to
consider the cut and paste function when designing the data
representation system. These two functions show us what
should be considered in this section. The implementation of
an index with references in the data requires maintaining
some code in the data to indicate that the word is in the
index. The cut and paste function requires no modification
of the data and should only be mentioned in passing.
For these reasons we restrict our requirements analysis to those
structural and functional considerations that may effect the data
representation design. How the data is conveyed is not as
important at this time as what data is represented. Our job is
to define the items that must be preserved and eliminate all
extraneous information. We must convert information that is
conveyed visually to its logical equivalent. In a book a heading
may be represented in characters that are one half inch high.
This should translate to information that this is a heading, not
that the characters are a particular size. The visual world uses
size to indicate importance. This will translate to the concept
of a level heading in our data representation system.
In this chapter we intend to describe those characteristics that
affect the data representation design. We will describe these
characteristics from an "Information Hierarchy." From this
description software structure and design can be derived. We
will formally describe the information requirements. From this
description the data design can be established. User interface
components are independent and can be specified at a later time.
We will not define the "library" level structure. The individual
"book" structure will be defined in detail. This is the most
complex document structure. Periodicals and reference materials
will be formally defined at a later time.
III.B Information Structure
Information is organized in a way that permits access to
structured data. If that structure does not exist in the data
representation, that information is not easily accessed. The
defined hierarchy determines the type of information that can be
derived from the data.
III.B.1 Information Hierarchy
The hierarchy of information is represented in a tree structure.
We recognize that a graphical representation of this tree will be
instructive for many sighted individuals. Many advisors working
on this project, including the principal investigator, cannot
take advantage of these graphical tools. Therefore, we will use
a simple notation system to describe the tree structure.
III.B.2 Terminology
The tree structure starts at level 0. Each move down the tree
moves us to another level. Each level on a path may have an
associated name. At a particular level we may have a name used
in other sections of the tree. "Paragraph" is a name of an
object that will occur in many branches of our tree. The naming
of this object helps identify the structure and will aid in
writing reusable computer code. Many possible paths from one
level to the next level may exist. (NOTE: I use the term "down"
to indicate a move from one level to the next level. We start at
the "root," level 0 and move down. The next level down has a
higher number. I know this does not make much sense, but the
idea draws from nature.)
LEVEL 0 BOOK
The tree structure we define here is the structure of a book with
all variations that may occur. It is not necessary that each
component exist in a particular instance of a book, but the
structure should allow for any variation. Anything that may be
found in a book will have a logical equivalent in this structure.
We use legal outline notation for the representation of this
structure. Each number in the chain represents a move down the
tree structure. I will briefly describe the first few levels so
an overview can be gained. I will then go into detail.
1 FRONT MATTER
1.1 Copyright Information
1.2 Title Page
1.3 Dedication
1.4 Preface
1.5 Preface To Previous Editions
1.6 Table of Contents
1.7 Introduction
2 BODY
2.1 Chapter
2.1.1 Level Heading
2.1.1.1 Subheading
2.1.1.1.1 Further Subheading Divisions
3 END MATTER
3.1 Appendices
3.2 Glossary
3.3 Index
3.3.1 Index headings
3.3.1.1 Index entries
The three logical divisions of a book, front matter, body, and
end matter are level 1 items. Each smaller structural item is
represented by an additional "." dot followed by a number. The
final number lacks trailing dot. This notation is used so
advisors can discuss the structural elements in perspective of
where it exists. A simple novel may only have a few of the
divisions described above. A college textbook may have all the
features. We will now describe in detail each item in the tree
structure of a book. This will take us to the lowest structural
levels, the paragraph, table or figure. Separate treatment of
each of these elements will be necessary later in this chapter.
III.B.3 Hierarchial Structural Description
1 FRONT MATTER
Front matter contains all information that comes before the body
of the book. This varies from title to title. The copyright
information is essential in the production of an electronic book.
1.1 Copyright information
Copyright holder, copyright dates, ISBN, and other
traditional legal information must be presented here.
1.2 Title Page
Title specification including authors information and
publisher information should be presented here.
1.3 Dedication
Author's dedication statement should be preserved.
1.4 Preface
Prefaces vary greatly and require more detailed treatment.
Sometimes level headings will be necessary and perhaps
figures and tables. We will treat these elements later.
Usually a series of paragraphs are used.
1.5 Prefaces To Previous Editions
Preface to previous editions are preserved and follow the
current preface.
1.6 Table Of Contents
The table of contents is extremely important. This
component of the front matter will be used as the "first
screen" of information presented when reading the book. The
table of contents shows the structure of the book. It lists
the front matter components, the book's body structure with
each level heading and subheadings, and all end matter
components. The table of contents provide hyper text links
into the book's parts.
1.7 Introduction
Introductions may occur in the front matter or may be the
first chapter of the book. This is dependent on the book
itself. The structure of the introduction follows the same
hierarchy as described below in the chapter section.
2 BODY
The body of a book is a collection of chapters. In the
simplest form there is only one chapter that may contain
only one level heading. In more complex books there may be
any number of chapters. A chapter continues until the next
chapter or end matter begins or the end of the book is
found. In essence a chapter is the highest level heading.
A novel may not specifically say this is a chapter. Here
the body of the book directly goes to the level heading or
the text level. Most often, the body is divided into
numbered chapters that have a title associated with the
chapter number.
2.1 Chapter
Chapters are major divisions of books that are numbered and
may have an associated title. Chapters are the highest
division of the body. Chapters may have level headings and
subheadings and sub subheadings. The chapter is treated no
differently than other level headings.
2.1.1 Headings
Headings are major divisions of chapters in the same way
that chapters are divisions of the body of the book. They
are numbered and fall below chapters in the hierarchy.
2.1.1.1 Subheadings
Subheadings are further divisions of headings. They are
numbered in descending order following the same pattern used
throughout the book.
2.1.1.1.1 Sub Subheadings
The level headings can be divided as far down as necessary.
Usually it will not be necessary to go more than four levels
deep. There are some applications that will require more
than four levels of division.
2.2 Paragraph
The paragraph is the smallest structural division of a book.
Tables and figures are on the same logical level as a
paragraph. A paragraph may follow any logical division of a
book. The introduction in the front matter may go to the
paragraph level. Paragraphs have a definite beginning and
end. Within paragraphs all textual, mathematical, and
scientific information may be represented.
2.3 Tables
Tables are one of three logical methods of representing
information. Tables are specifically arranged and the
information represented in each element of the table is
similar to what can be represented in paragraphs. Tables
are numbered and may have titles associated with the table.
2.4 Figures
Figures are graphical representations of information. They
are at the same logical level as paragraphs and tables. The
figures have up to four components. A graphic file
associated with the original graphic presented in the paper
version of the book is named. A textual description of this
graphic is included at the point where the graphic occurs in
the book. A second, simplified graphic file, is also named
in the text and the textual description of this graphic
precedes the original description. The order of the
description should be first the simplified graphic followed
by the complex graphic. If the graphic is so simple in its
original form that no further simplification is needed, a
statement in the book should indicate that no simplified
graphic was necessary.
3 END MATTER
End matter is the last of the three logical divisions of a
book. In the simplest form, no end matter is present. In
the complex instances of books many appendices, glossaries,
and one or more indexes may be present.
3.1 Appendix
An appendix is a logical division of a book that provides an
opportunity to provide greater detail than what was
appropriate in the chapter of a book. Sometimes this may
follow the same logical structure as a chapter. In other
cases it may follow the organization of an alphabetical
listing of additional information. This alphabetical list
we shall call a list of "entries."
3.1.1 Entry
The entry shall have a specific notation before the
beginning of the entry. This notation indicates that it is
meant to be located by that entry name. An entry can be
treated as a paragraph, table, or figure.
3.2 Glossary
Glossaries are normally dictionaries that provide more
information than what is provided directly in the text
portions of the body. The glossary is a series of entries.
3.2.1 Entry
Defined above
3.3 Index
The index is very important and needs to be treated with
great care. Some considerations are: that the page numbers
are preserved, exact location of the referenced item is
known, groups of index items are collected under headings,
only those page numbers noted in the paper version are
referenced. Classroom discussions frequently reference page
numbers. The page numbers in the electronic version of the
text should coincide with the pages in the paper copy.
3.3.1 Index Entry
An index entry is normally a specially noted word or phrase.
The special notation allows for direct access to that
location in the text. The index may be created directly by
collecting the noted index items. Electronic versions use
this notation to create hypertext links. The index string
is noted in the text. When creating the index, entries with
the same first word would be sub indexed under that word.
For example:
@index"abstract data types linked list"
@index "abstract data types trees"
@index "abstract data types queue"
These three index items would be listed under the common
index heading ABSTRACT DATA TYPES. The words linked list
would have page numbers after the entry and an associated
hypertext link could be created. Trees and queue would be
entries under the linked list entries.
ABSTRACT DATA TYPES
linked list 93, 105, 233
trees 35, 199, 266
queue 110, 244, 301
4 Internal references
Three types of internal references are provided cross
references, footnotes, and endnotes.
4.1 Cross References
Cross referenced items are similar to index entries. A
cross reference allows for easy movement to other portions
of the text. In printed material you commonly find
information like "see page 193." This type of cross
reference is created by noting a particular location in the
book and then referring to that named location. These cross
references are created by the computer at time of printing.
If pages are added or deleted the pages referenced are
automatically changed. A portion of text may contain a
label that notes that location. Anywhere in the future the
label name may be used in the same way that you may name a
page. Named labels may reference forward or backward in the
text. There are two types of named labels, place-labels
and call-labels. The place-label command is a location in
the text. The call-label points to a place label and
provides for a hypertext link to be created. For example
three separate locations in a book may contain the following
lines:
George Bush is President of the United States. @place-
label bush1
President Bush ordered the liberation of Kuwait.
@place-label bush2
What two things are George Bush known for? See pages
@call-label bush1 @call-label bush2
The words "@place-label bush1," "@place-label bush2" mark a
specific location in the text. The place labels are not
shown in the paper version of the book and will not be seen
in the electronic version. The data representation will
need to show these locations. The @call-label command
followed by a unique label name allows for the replacement
of page numbers by the called labels. In the electronic
version pages can also be indicated and direct jumping to
these locations can be made simple. (NOTE: I use the "at"
sign,"@," as an "escape character," the data representation
design will address the issue of escape characters fully.)
4.2 Footnotes
Footnotes will be noted in text and associated with a
number. They will be accessible as a hypertext jump or as
traditional footers.
4.3 Endnotes
Similar to footnotes, but may be read as traditional end of
chapter material.
The three remaining portions of the structural requirements
are the paragraph, tables and figures.
III.B.3.a Paragraph
The paragraph is the smallest structural element. Paragraphs
consist of collections of sentences, characters, mnemonic
symbols, or mathematical and scientific information. The
grammatical meaning of a paragraph is the most common form of the
structural paragraph. Collections of sentences will be found
most frequently at this level. Paragraph styles must be
mentioned here.
1. Standard paragraph has one or more sentences
2. Indented paragraphs are set off from other paragraphs for
a variety of reasons. Indentation is a technique for
drawing attention to specific information.
3. Paragraphs in list form. A paragraph may consist of a
list of items. The items may be sentences, words, phrases
or whatever information the author chooses. These list are
sometimes numbered and other times a special character is
placed before the entry to draw attention to each item.
Classically this character is called a "bullet" The bullet
can take on almost any form. Bullet draw visual attention
to each entry. Again it is a technique for drawing attention
to an item.
4. A paragraph may take the form of a mathematical or
scientific formula. The author may select to represent the
equation in a paragraph by itself rather than noting the
equation within a sentence or paragraph. Often a formula
may be represented at the paragraph level and subsequent
paragraphs refer to parts of the formula. The ability to
represent mathematical and scientific information within a
sentence or as a paragraph itself must be preserved.
III.B.3.b Tables
Tables represent related information in a structured form. The
structure of the table is in rows and columns. The most common
notion of a table is a spreadsheet. The columns have an assigned
letter starting at "A." The numbered rows start at "1." Any
location can be precisely referenced by a letter and number.
Normally the titled rows and columns have information below and
across logically associated with the titles. The variety of
information in tables create difficulties for visual
representation as well as logical representation. Any cell may
contain a single character, or many paragraphs of information
may be represented.
It should be possible to read any column heading and the
information organized under the column. Similarly, it is
necessary to read across rows. Realize that any one cell may
contain more information than can be represented on a screen at
any one time. Using a Row column indicator allows for software
to present information in a manner selected by the reader.
III.B.3.c Figures
Figures are drawings, charts, pictures and any other visual
information. These figures have up to four components. 1) a
graphic file that is a simplified version of the graphic in the
book. 2) a text description of the simplified figure as it
relates to the information contained in the book. 3) a graphic
that closely represents the original graphic in the book. 4) a
text description of the graphic in the book as it relates to the
context of the book.
The text should be labeled as a description of the figure and a
named graphic computer file should be associated with the text
description.
III.C Emphasized Text
Emphasized text can be seen throughout any modern textbook.
Making books "look inviting" is an important task of the
publisher. Preserving all emphasized text complicates books
designed for the print-disabled beyond usefulness. Balance is
needed when deciding what to preserve and what to eliminate in
emphasized text. A brief example is needed here.
A publisher may choose to emphasize a particular word. Let
us select the word "book" for this example. Visually the
entire word is in bold face print. The "B" is printed in a
larger size font and the style of the "ook" is slightly
different from the first character in the word. Spacing
before and after is somewhat larger than other spacing. The
publisher has certainly drawn attention to the word.
What are we to do with this emphasized word in our system? We
wish to preserve the fact that this word is emphasized and there
must be an associated description of the emphasis type. This
emphasis also must have a definite beginning and end. Let us
consider a grammar textbook that uses emphasis to indicate
different parts of speech. The subject of the sentence may be
bolded, the verb in italic print, the object may be highlighted,
and all nouns are underlined. Information is presented by the
use of emphasized text.
First let us decide on broad categories to eliminate.
1. Size need not be preserved. Usually size of print
indicates structural components. We have dealt with this
logically by using level headings.
2. Font style is meaningless to the visually impaired. If
the presentation of a character is in "Dutch," "Hampton,"
"Gothic" or another style, little information is conveyed.
This may be more visually pleasing in one font more than
another, but I doubt whether an author made decisions of
this nature when they created the book.
Second let us decide on the types of emphasis to preserve. The
following is a list of the types of named emphasis we will
preserve:
1. Bold face print
2. Italic print (Italic "leans" right.)
3. Slanted print (Slanted "leans" left.)
3. Underlined text
4. Double underlined text
5. Highlighted text
We expect that software will associate the emphasized text with
a screen attribute: Reverse video for bold, underline with
underline, flashing with italic and so on. Any type of emphasis
will be mapped to one of these characteristics. If a publisher
uses more than one type of bold, all bold styles will map to the
one bold style in this electronic version of the book.
III.D Characters, Symbols and Punctuation
ASCII is the rock we will build our system upon. There are 256
characters defined in the ASCII character set and the extended
character set. Yet there are well over a thousand characters,
mathematical and scientific symbols, and punctuation marks used
in the printing industry today. Each character or symbol must
map to an associated ASCII character or mnemonic defined in our
system. As a rule the printing industry is complex. This
complexity will not serve to convey information effectively if
text is filled with mnemonics that confuse more than clarify.
For example there are at least five types of dashes used in
printing: the dash, the minus sign, a hyphen, a symbol known as
an em-dash and another called an en-dash. Our system will map
the minus sign and the hyphen to the "-," ASCII 45. The others
will map to a mnemonic for a dash. Similarly, typeset books
normally have opening quotation marks and closing quotation
marks. These will map to ASCII 34, the quotation mark. In this
way we can make the text less cluttered. Information is lost by
this simplification, but it is questionable how many fully
sighted people have noticed that there are at least five types of
dashes. The specific mapping will be explicitly defined in the
design section of the next chapter.
Mathematical and scientific notation uses spacial relations to
define equations precisely. We must be capable of translating
the spacial relations to a linear, text based format. For these
purposes we borrow heavily from TeX discussed in Chapter 2. If
we say, "square root of three over four," we have not precisely
defined what equation we are talking about. Is three over four
inside the square root symbol or is the square root symbol over
four? The system we define must be precise and translatable to a
correct visual representation and to a correct tactile
representation. It seems that it will be necessary to have a
special "math and science" mode. This mode will use certain
symbols for scientific purposes. These requirements are
necessary and form the basis for a reading and writing system
that can take us into a new level of education for the print-
disabled.
III.E Conclusion
We have defined the structural components required for reading
books. In addition to the structure we have included
requirements necessary for hypertext links. Limitations were
placed on the complexity of emphasized text. The requirement for
logical transformation of visual clues was constantly emphasized.
Simplification of enhancements and characters that logically
preform the same task is encouraged. Graphic figures are treated
fully allowing maximum use of existing vision for all disability
groups. Finally, we require that all mathematical and scientific
information be represented in a linear text based system that
allows for no ambiguity. These features implemented in a
logical, easy to use data representation system will provide a
foundation for a reading and writing system for the print-
disabled.