Sight is a quite complicated system, built in just two years. How was it possible? Having our own original idea, we use the power of co-operation. Many really good, reliable modules and applications are released under General Public License (GPL) or compatible. GPL rules that we strictly follow allows to integrate them into the open source software. We used:
Stalker algorithm, as described in Muslea et al. This is an algorithm for analysis of html documents and should not be confused with the similary named encryption algorithms.
WebMacro is a nice template engine, normally used to generate dynamic html pages. We however forced it to generate java code and XSLT transforms from our templates. We could not use Apache Velocity, because the current Apache license is not GPL compatible.
KJC Kopi Java Compiler is a good java compiler, used internally by Sight to compile the generated web robots and complete applications. Pizza also works well, but we could not release the version with Pizza - the old Artistic license is not GPL compatible.
JSch M is a simple, reliable library to work with SSH protocol that is now replacing TelNet and FTP. This library allows to create and use robots that log in into server and execute some commands. Hence there is no longer need to have an accessible web form on the server. While HTTP feature is useful integrating public web services, SSH is used to work with resources on the local network of the institution.
Weka is a rich collection of the text mining algorithms. Sight workflow allows modules that classify data using Weka classifiers, and provides some means to choose the optimal classifier for the current task.
BioQuery M is a specialised web robot that currently does just one thing - submits queries to NCBI databases. But it does this really well! With our adaptor the BioQuery can work in Sight workflow together with other robots.
jEdit M Syntax Package can display raw html and xml highlighting the syntax. It really helps when you need to mark something during the agent generation.
M - to use in Sight, we needed to modify this package. The released downloads contain the modified code and not the original code.
Sight also has one "enemy" - Service Pack 6 for MS Windows NT 4.0 . Sigh is not operating properly under this service pack, use the Service Pack 5 or 6a instead.