Article View: ch.admin
Article #14262Re: Search engine for ch.* hierarchy
From: Marc SCHAEFER
Date: Thu, 12 Apr 2001 09:17
Date: Thu, 12 Apr 2001 09:17
15 lines
1658 bytes
1658 bytes
Marc SCHAEFER <schaefer@alphanet.ch> wrote: > Thank you for feedback. Infact, current implementation is quite efficient > because it is based on indexes, and indexes don't really work very well > with regular expressions or wildcards. Implementing this would probably > be a much more difficult task than implement `and', `not' and grouping. Thinking more, I have extracted all the words and put them in a flat file, line by line. For fr.* hierarchy, one year, this is about 13 MBytes. This can be easily cached in memory, and then doing simple regexp takes very little time on a modern processor (about 10ms for the test case I tried with a simple egrep 'SCH(A|AE)FER'). With the returned values, normal queries can then be done. So it looks realizable without too much penalty.
Technical Details
Message-ID:
<9b3rqv$ofn$1@vulcan.alphanet.ch>Path:
arc1-ca.pugleaf.net!supernews.google.com!sn-xit-03!supernews.com!cyclone-sjo1.usenetserver.com!news-out-sjo.usenetserver.com!newsfeed.wirehub.nl!news.stealth.net!news.imp.ch!unet.urbanet.ch!alphanet.ch!not-for-mailReferences:
<9b1hm3$815$1@vulcan.alphanet.ch> <9b26l8.3vs79st.1@news.pflock.de> <9b273m$muo$1@vulcan.alphanet.ch>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 11008d,bc7eb2087db0b33b X-Google-Attributes: gid11008d,public X-Google-ArrivalTime: 2001-04-12 02:38:05 PST Path: arc1-ca.pugleaf.net!supernews.google.com!sn-xit-03!supernews.com!cyclone-sjo1.usenetserver.com!news-out-sjo.usenetserver.com!newsfeed.wirehub.nl!news.stealth.net!news.imp.ch!unet.urbanet.ch!alphanet.ch!not-for-mail From: Marc SCHAEFER <schaefer@alphanet.ch> Newsgroups: ch.admin Subject: Re: Search engine for ch.* hierarchy Date: 12 Apr 2001 11:17:19 +0200 Organization: ALPHANET NF -- Not for profit telecom research Lines: 15 Message-ID: <9b3rqv$ofn$1@vulcan.alphanet.ch> References: <9b1hm3$815$1@vulcan.alphanet.ch> <9b26l8.3vs79st.1@news.pflock.de> <9b273m$muo$1@vulcan.alphanet.ch> NNTP-Posting-Host: vulcan.alphanet.ch X-Newsreader: TIN [UNIX 1.3 unoff BETA 970705; i586 Linux 2.0.38]