3 Commits

Author	SHA1	Message	Date
Jean-Francois Dockes	0860b559ee	get rid of a few garbage terms during indexing. Set a threshold for conversion errors after which we discard the doc. Stabilize the new termproc pipeline but no commongrams for now	2011-10-12 17:55:58 +02:00
Jean-Francois Dockes	4a7ff398b2	comments	2011-10-07 08:05:36 +02:00
Jean-Francois Dockes	5fd31172f5	New text to terms processing pipelines: results identical to 1.16 when used with empty stopfile	2011-10-07 07:53:49 +02:00