Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commit866bad9

Browse files
committed
Add a rank/(rank+1) normalization option to ts_rank(). While the usefulness
of this seems a bit marginal, if it's useful enough to be shown in the manualthen we probably ought to support doing it without double evaluation of thets_rank function. Per my proposal earlier today.
1 parent5858990 commit866bad9

File tree

2 files changed

+32
-15
lines changed

2 files changed

+32
-15
lines changed

‎doc/src/sgml/textsearch.sgml‎

Lines changed: 15 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
<!-- $PostgreSQL: pgsql/doc/src/sgml/textsearch.sgml,v 1.33 2007/11/1418:36:37 tgl Exp $ -->
1+
<!-- $PostgreSQL: pgsql/doc/src/sgml/textsearch.sgml,v 1.34 2007/11/1423:43:27 tgl Exp $ -->
22

33
<chapter id="textsearch">
44
<title id="textsearch-title">Full Text Search</title>
@@ -940,6 +940,7 @@ SELECT plainto_tsquery('english', 'The Fat &amp; Rats:C');
940940
<listitem>
941941
<para>
942942
4 divides the rank by the mean harmonic distance between extents
943+
(this is implemented only by <function>ts_rank_cd</>)
943944
</para>
944945
</listitem>
945946
<listitem>
@@ -953,17 +954,24 @@ SELECT plainto_tsquery('english', 'The Fat &amp; Rats:C');
953954
of unique words in document
954955
</para>
955956
</listitem>
957+
<listitem>
958+
<para>
959+
32 divides the rank by itself + 1
960+
</para>
961+
</listitem>
956962
</itemizedlist>
957963

964+
If more than one flag bit is specified, the transformations are
965+
applied in the order listed.
958966
</para>
959967

960968
<para>
961969
It is important to note that the ranking functions do not use any global
962-
information so it is impossible to produce a fair normalization to 1% or
963-
100%, as sometimes desired.However, a simple technique like
964-
<literal>rank/(rank+1)</literal> can be applied. Of course, this is just
965-
a cosmetic change, i.e., the orderingofthe search results will not
966-
change.
970+
information, so it is impossible to produce a fair normalization to 1% or
971+
100% as sometimes desired.Normalization option 32
972+
(<literal>rank/(rank+1)</literal>) can be applied to scale all ranks
973+
into the range zero to one, butofcourse this is just a cosmetic change;
974+
it will not affect the ordering of the search results.
967975
</para>
968976

969977
<para>
@@ -991,7 +999,7 @@ ORDER BY rank DESC LIMIT 10;
991999
This is the same example using normalized ranking:
9921000

9931001
<programlisting>
994-
SELECT title, ts_rank_cd(textsearch, query)/(ts_rank_cd(textsearch, query) + 1) AS rank
1002+
SELECT title, ts_rank_cd(textsearch, query, 32 /* rank/(rank+1) */) AS rank
9951003
FROM apod, to_tsquery('neutrino|(dark &amp; matter)') query
9961004
WHERE query @@ textsearch
9971005
ORDER BY rank DESC LIMIT 10;

‎src/backend/utils/adt/tsrank.c‎

Lines changed: 17 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@
77
*
88
*
99
* IDENTIFICATION
10-
* $PostgreSQL: pgsql/src/backend/utils/adt/tsrank.c,v 1.8 2007/09/20 18:10:57 teodor Exp $
10+
* $PostgreSQL: pgsql/src/backend/utils/adt/tsrank.c,v 1.9 2007/11/14 23:43:27 tgl Exp $
1111
*
1212
*-------------------------------------------------------------------------
1313
*/
@@ -25,13 +25,14 @@ static float weights[] = {0.1f, 0.2f, 0.4f, 1.0f};
2525

2626
#definewpos(wep)( w[ WEP_GETWEIGHT(wep) ] )
2727

28-
#defineRANK_NO_NORM0x00
28+
#defineRANK_NO_NORM0x00
2929
#defineRANK_NORM_LOGLENGTH0x01
30-
#defineRANK_NORM_LENGTH0x02
31-
#defineRANK_NORM_EXTDIST0x04
32-
#defineRANK_NORM_UNIQ0x08
33-
#defineRANK_NORM_LOGUNIQ0x10
34-
#defineDEF_NORM_METHODRANK_NO_NORM
30+
#defineRANK_NORM_LENGTH0x02
31+
#defineRANK_NORM_EXTDIST0x04
32+
#defineRANK_NORM_UNIQ0x08
33+
#defineRANK_NORM_LOGUNIQ0x10
34+
#defineRANK_NORM_RDIVRPLUS10x20
35+
#defineDEF_NORM_METHODRANK_NO_NORM
3536

3637
staticfloatcalc_rank_or(float*w,TSVectort,TSQueryq);
3738
staticfloatcalc_rank_and(float*w,TSVectort,TSQueryq);
@@ -348,12 +349,17 @@ calc_rank(float *w, TSVector t, TSQuery q, int4 method)
348349
res /= (float)len;
349350
}
350351

352+
/* RANK_NORM_EXTDIST not applicable */
353+
351354
if ((method&RANK_NORM_UNIQ)&&t->size>0)
352355
res /= (float) (t->size);
353356

354357
if ((method&RANK_NORM_LOGUNIQ)&&t->size>0)
355358
res /=log((double) (t->size+1)) /log(2.0);
356359

360+
if (method&RANK_NORM_RDIVRPLUS1)
361+
res /= (res+1);
362+
357363
returnres;
358364
}
359365

@@ -762,7 +768,7 @@ calc_rank_cd(float4 *arrdata, TSVector txt, TSQuery query, int method)
762768
Wdoc /= (double)len;
763769
}
764770

765-
if ((method&RANK_NORM_EXTDIST)&&SumDist>0)
771+
if ((method&RANK_NORM_EXTDIST)&&NExtent>0&&SumDist>0)
766772
Wdoc /= ((double)NExtent) /SumDist;
767773

768774
if ((method&RANK_NORM_UNIQ)&&txt->size>0)
@@ -771,6 +777,9 @@ calc_rank_cd(float4 *arrdata, TSVector txt, TSQuery query, int method)
771777
if ((method&RANK_NORM_LOGUNIQ)&&txt->size>0)
772778
Wdoc /=log((double) (txt->size+1)) /log(2.0);
773779

780+
if (method&RANK_NORM_RDIVRPLUS1)
781+
Wdoc /= (Wdoc+1);
782+
774783
pfree(doc);
775784

776785
pfree(qr.operandexist );

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp