Mailing List Archive

Author: creamyg
Date: 2008-09-18 21:20:03 -0700 (Thu, 18 Sep 2008)
New Revision: 3882

Modified:
trunk/c_src/KinoSearch/Posting/MatchPosting.bp
trunk/c_src/KinoSearch/Posting/MatchPosting.c
Log:
Factor in TF and IDF (but not boost) in MatchPostingScorer. This is only for
testing and benchmarking purposes -- MatchPostingScorer should really produce
a constant score.


Modified: trunk/c_src/KinoSearch/Posting/MatchPosting.bp
===================================================================
--- trunk/c_src/KinoSearch/Posting/MatchPosting.bp 2008-09-19 03:48:00 UTC (rev 3881)
+++ trunk/c_src/KinoSearch/Posting/MatchPosting.bp 2008-09-19 04:20:03 UTC (rev 3882)
@@ -52,9 +52,17 @@
class KinoSearch::Posting::MatchPostingScorer cnick MatchPostScorer
extends KinoSearch::Search::TermScorer {

+ float *score_cache;
+
static MatchPostingScorer*
init(MatchPostingScorer *self, Similarity *similarity,
PostingList *posting_list, Compiler *compiler);
+
+ public Tally*
+ Tally(MatchPostingScorer *self);
+
+ void
+ Destroy(MatchPostingScorer *self);
}

/* Copyright 2007-2008 Marvin Humphrey

Modified: trunk/c_src/KinoSearch/Posting/MatchPosting.c
===================================================================
--- trunk/c_src/KinoSearch/Posting/MatchPosting.c 2008-09-19 03:48:00 UTC (rev 3881)
+++ trunk/c_src/KinoSearch/Posting/MatchPosting.c 2008-09-19 04:20:03 UTC (rev 3882)
@@ -9,6 +9,7 @@
#include "KinoSearch/Posting/RawPosting.h"
#include "KinoSearch/Search/Similarity.h"
#include "KinoSearch/Search/Compiler.h"
+#include "KinoSearch/Search/Tally.h"
#include "KinoSearch/Store/InStream.h"
#include "KinoSearch/Util/MemoryPool.h"

@@ -116,12 +117,44 @@

MatchPostingScorer*
MatchPostScorer_init(MatchPostingScorer *self, Similarity *sim,
- PostingList *plist, Compiler *compiler)
+ PostingList *plist, Compiler *compiler)
{
- return (MatchPostingScorer*)TermScorer_init((TermScorer*)self, sim,
- plist, compiler);
+ u32_t i;
+
+ /* Init. */
+ TermScorer_init((TermScorer*)self, sim, plist, compiler);
+
+ /* Fill score cache. */
+ self->score_cache = MALLOCATE(TERMSCORER_SCORE_CACHE_SIZE, float);
+ for (i = 0; i < TERMSCORER_SCORE_CACHE_SIZE; i++) {
+ self->score_cache[i] = Sim_TF(sim, (float)i) * self->weight;
+ }
+
+ return self;
}

+Tally*
+MatchPostScorer_tally(MatchPostingScorer* self)
+{
+ Tally *const tally = self->tally;
+ MatchPosting *const posting = (MatchPosting*)self->posting;
+ const u32_t freq = posting->freq;
+
+ /* Calculate initial score based on frequency of term. */
+ tally->score = (freq < TERMSCORER_SCORE_CACHE_SIZE)
+ ? self->score_cache[freq] /* cache hit */
+ : Sim_TF(self->sim, (float)freq) * self->weight;
+
+ return tally;
+}
+
+void
+MatchPostScorer_destroy(MatchPostingScorer *self)
+{
+ free(self->score_cache);
+ TermScorer_destroy((TermScorer*)self);
+}
+
/* Copyright 2007-2008 Marvin Humphrey
*
* This program is free software; you can redistribute it and/or modify


_______________________________________________
kinosearch-commits mailing list
kinosearch-commits@rectangular.com
http://www.rectangular.com/mailman/listinfo/kinosearch-commits