The files contained in this zip file are provided for researchers
interested in verifying or extending the analyses in my ICCHP 2012 
paper, "Modeling Text Input for Single-Switch Scanning".

Included are ScanningKeyboardSPC.java and supporting files,
including the four word-frequency files used for the analyses
described in the paper:

	d1-wordfreq.txt       (BNC-1 corpus)
	d2-wordfreq.txt       (BNC-2 corpus)
	bc-wordfreq.txt       (Brown corpus)
	phrases2-wordfreq.txt (Phrases corpus)

To begin, unzip all the files into a directory, then compile all and 
build the javadoc API:

	PROMPT>javac *.java
	PROMPT>javadoc *.java

Obviously, the java SDK must be available on the host system.  

A good place to start is by reading the API for ScanningKeyboardSPC,
which is in index.html, as created by javadoc.

The SPC statistics in the ICCHP paper can be generated by
running ScanningKeyboardSPC from the command line.  For example,

	PROMPT>java ScanningKeyboardSPC d1-wordfreq.txt -ck_earduW-tnsfwbW-ohcpvjW-imykq.W
-lgxz..W-......W -spc -wp -kl

will produce the following output on the console:

	Keyboard layout...
	   _ e a r d u W
	   t n s f w b W
	   o h c p v j W
	   i m y k q . W
	   l g x z . . W
	   . . . . . . W
	WordFreqFile=d1-wordfreq.txt, SPC= 3.35

The statistic above (SPC = 3.35) is for the Koester-Levine optimized keyboard
operating with word prediction using half-and-half scanning.  See Figure 3 in 
the ICCHP paper.

Alternatively, the statistics in Figure 3 can be generated using the batch file
ScanningKeyboardSPC.bat:

	PROMTP>ScanningKeyboardSPC
	
At the very least, using the batch program will save quite a bit of command-line typing.

The batch program will take a very long time to execute on most 
systems.  The bottleneck is the generation of SPC for keyboard configurations
using word prediction.

Edit the batch file to add or remove "rem" statements, as
desired, to generate a subset of the statistics.

Feel free to modify the source code to suite your needs.

Also include is an Excel spreasheet (ScanningKeyboardSPC.xlsx) that collects
together many of the statistics and figures in the paper.  The spreadsheet
also includes images of the keyboards from the original
publications.

Please contact me with any questions or comments.

Scott MacKenzie, March 16, 2012
(mack "at" cse.yorku.ca)
