I've google'd for this stuff and already found some info. But, I want to ensure that my interpretation is right.
To my understanding,  the threshold word used by LZH or other dictionary compression algorithm explanation/specification means the minimum number of match-length in bytes, i.e. the minimum number of  bytes that match within the dictionary and the bytes in the current look-ahead buffer that have to be "surpassed" to carry-out the "encoding". So, if the match-length is less than the threshold, the phrase in the look-ahead buffer won't be encoded, but will stay as literal in the compression result. Is this correct?

There must be some compression guru around here to help me ;)

PS: sorry for the bad english
Posted on 2005-06-13 03:56:15 by Pinczakko

There must be some compression guru around here to help me

If not, try efnet, #compression, and get hold of Jibz :)
Posted on 2005-06-13 05:54:40 by f0dder
The terminology differs a bit between the descriptions you find, but I think you're most likely right.

There is often a minimum match length, below which it is impossible (or unlikely) to obtain any compression due to the encoding, so you ignore shorter matches.

E.g. if you use 20 bits to encode each match, there is no point in being able to encode matches of length 2 :).
Posted on 2005-06-13 10:25:53 by Jibz