[Blindmath] Status Report on BackNem Nemeth Braille Backtranslator

[Susan Jolly]

Since Susan O. was kind enough to ask about the status of my BackNem 
backtranlator here it is.

Let me begin by noting that there has been some discussion on this list 
about backtranslation of braille math not being an appropriate strategy 
since there are numerous modern tools that blind persons can use to produce 
properly typeset print mathematics. Here is my opinion on this issue.  All 
of these tools require writing the math using one of the many different math 
entry systems, such as LaTeX or ASCIIMathML, to create a source file. These 
source file are not intended to be read directly but must be processed by 
software so as to produce typeset math.  I don't see that this approach is 
significantly different from using Nemeth math as a math entry system, 
especially in the context of a braille user who is taking courses in math.

[I understand that braille users may end up in careers or situations where 
they need to be familiar with certain software although this is constantly 
changing.  As I've mentioned before, I never had to learn LaTeX because it 
was thought to be a more efficient use of our time as scientists for our 
secretaries to take care of  such things.]

Now to answer the question. I worked for six or so years developing a Nemeth 
backtranslator although it has now been more than four years since I did any 
significant work on this. What BackNem does is to parse or interpret braille 
and translate it to MathML plus HTML. BackNem uses a combination of Java 
code generated automatically from hand-written ANTLR v2 grammars together 
with hand-written Java. There are a growing number of apps that can render 
MathML so there was no need to develop a renderer.

You can read a somewhat out-of-date overview article about BackNem here and 
either follow the links in that article or go directly to the Table of 
Contents for more information.
http://braille2print.org/backnem2.htm
http://braille2print.org/backnemtoc.htm

Part of the reason BackNem was never released was because of the very poor 
response to the alpha test. Please note the overview article is out-of-date 
as far as its reference to an alpha test. (Also please ignore the 
meaningless pop-up warning message.)

A serious technical issue that arose was the release and widespread adoption 
of ANTLR v3 which is not backward compuatible with ANTLR v2 but which has so 
many advantages that continuing development or even maintenance in ANTLR v2 
did not make sense. I did write some test grammars in ANTLR v3 and had plans 
to convert BackNem to ANTLR v3 but ended up not doing this for various 
reasons.

One reason is that I was discouraged by the ongoing uncertainty as far as 
the braille math code to be used here in the United States.  A lot of the 
complexity of BackNem is related to distinguishing math expressions from 
text since the Nemeth code does not directly distinguish math and text but 
assumes that the human reader will have no problem doing so. However, just 
as I had pretty much finished a first cut at developing this capability, 
there seemed to be the possibility of changes to braille math.

One possibility is that the US will adopts NUBS, which is a modernized 
version of Nemeth that distinguishes math and text in a way that would be 
easy for an automated backtranslator to interpret. BackNem could be 
considerably simplified if NUBS were adopted although, of course, it would 
have to be extended to handle the changes from Nemeth to NUBS.
http://braille2print.org/implement.htm#auto

A second possibility is that the US will adopts UEB, which is totally 
different from Nemeth or NUBS.  Since I have such a negative opinion of UEB 
I cannot imagine being motivated to develop a UEB backtranslator.

There is a large community of ANTLR v3 experts who are also expert Java 
programmers.  If BANA decides either to stick with the Nemeth code or to 
adopt NUBS I would be willing to work (as a volunteer) with one or more 
persons with the appropriate ANTLR plus Java expertise to finish BackNem. I 
estimate this would take no more than six calendar months.

Of course, it would also be necessary to develop an organizational structure 
for maintaining BackNem. I would hope that BackNem could remain open source 
and free which means the organizational structure would need some sort of 
independent support.

Susan Jolly