Hello,
I am research KinoSearch items, today, and came across this post. I have a couple comments for your question. I am no guru, but what I have done with my implementation of KinoSearch is the following:
Wrote script to perform KinoSearch indexing on a file
Downloaded and installed Apache Tika
Have Tika convert files to text version of files
Inform KinoSearch index script to pull in files
If file has a text brother, then index the text brother, instead of the raw file
Has been very successful, so far.