RE: htdig long searches

Gene Haignere (
Mon, 16 Mar 1998 13:07:51 -0800

Is there an easy way to limit the search to the first n hits (say 1000) even if all of these are in 3 or 4 documents? Frost and Sullivan seem to be willing now to accept this sort of solution.

If most of the time is in the postprocessing phase, would it help to reduce the number of words before and after the key word? What else might help?

>This would be htsearch/ and htsearch/
>Unfortunately, those two files make up the whole guts of the search
>Mike Stabler wrote:
>> Andrew,
>> We still have this problem with the Frost&Sullivan site and they're
>> anxious for a fix. We're going to proceed with your suggestion. I'm
wondering if it would
>> help the programmer that we assign to the task if he/she knew the
>> source code file names.
>> Any info you can provide considering your limited time would be
>> Thanks,
>> Mike Stabler
>> -----Original Message-----
>> From: Andrew Scherpbier <>
>> To: <>; Darren Maglidt
>> <>
>> Date: Thursday, February 12, 1998 1:23 PM
>> Subject: htdig long searches
>> >I just thought of a good way to limit the search result count.
>> >
>> >Currently, all matches need to be found in the database and then sorted
>> >according to the document weight.
>> >Since all information, except the maximum weight of all results is
>> available,
>> >the search algorithm could be modified to start throwing away results if
>> their
>> >weight is below a certain number, relative to what it has already seen.
>> That
>> >way, only a limited number of results need to be actually sorted and
>> the
>> >speed will increase.
>> >Unfortunately, I don't have time to do this. A competent C++ programmer
>> >should be able to do this, however.
>> >
>> >--
>> >Andrew Scherpbier <>
>> >Contigo Software <>
>> >
>> >
>Andrew Scherpbier <>
>Contigo Software <>

This archive was generated by hypermail 2.0b3 on Sat Jan 02 1999 - 16:25:49 PST