fork of https://github.com/sourcegraph/zoekt
ranking: parametrize contribution of document ranks (#479)
We tune the contribution of document ranks to the final ranking with a damping factor. The zero value of the damping factor leads to an equal contribution of scores and ranks, while a damping factor of 1 effectively eliminates the influence of document ranks.
The default is 0 IE equal contribution.
author
Stefan Hengl
committer
GitHub
date
(Nov 9, 2022, 3:22 PM +0100)
commit
2e71aad6
2e71aad6ac9ace4bed7febc928ca466ea3c3afd9
parent
8645eff1
8645eff1f7580ef292a9f0bdb548a01f6d2af425
+44
-14
+5
api.go
+5
api.go
···
803
803
// sorting matches.
804
804
UseDocumentRanks bool
805
805
806
+
// RanksDampingFactor determines the contribution of documents ranks to the
807
+
// final ranking based on RRF. A value in (0,1] reduces the contribution,
808
+
// while a value in (-inf,0) increases it.
809
+
RanksDampingFactor float64
810
+
806
811
// Trace turns on opentracing for this request if true and if the Jaeger address was provided as
807
812
// a command-line flag
808
813
Trace bool
+1
-1
contentprovider.go
+1
-1
contentprovider.go
···
900
900
sort.Stable(fileMatchesByRank{fileMatches: ms, rrfScore: rrfScore})
901
901
902
902
for i := range rrfScore {
903
-
rrfScore[i] += 1 / (k + float64(i))
903
+
rrfScore[i] += (1 - opts.RanksDampingFactor) / (k + float64(i))
904
904
if opts.DebugScore {
905
905
ms[i].Debug += fmt.Sprintf("%d), ", i)
906
906
}
+38
-13
contentprovider_test.go
+38
-13
contentprovider_test.go
···
3
3
import (
4
4
"bytes"
5
5
"fmt"
6
+
"math"
6
7
"testing"
7
8
8
9
"github.com/google/go-cmp/cmp"
···
335
336
{FileName: "d4", Score: 1, Ranks: []float64{0.5}},
336
337
}
337
338
338
-
// Document RRF(Score) RRF(Ranks) SUM Rank
339
-
// d3 1/(60+1) 1/(60+0) 0,0330601092896175 0
340
-
// d2 1/(60+0) 1/(60+3) 0,0325396825396826 1
341
-
// d1 1/(60+2) 1/(60+1) 0,0325224748810153 2
342
-
// d4 1/(60+3) 1/(60+2) 0,0320020481310804 3
339
+
cases := []struct {
340
+
name string
341
+
dampingFactor float64
342
+
wantOrder []string
343
+
}{
344
+
// Document RRF(Score) RRF(Ranks) SUM Rank
345
+
// d3 1/(60+1) 1/(60+0) 0,0330601092896175 0
346
+
// d2 1/(60+0) 1/(60+3) 0,0325396825396826 1
347
+
// d1 1/(60+2) 1/(60+1) 0,0325224748810153 2
348
+
// d4 1/(60+3) 1/(60+2) 0,0320020481310804 3
349
+
{
350
+
"equal",
351
+
0,
352
+
[]string{"d3", "d2", "d1", "d4"},
353
+
},
354
+
{
355
+
"scores only",
356
+
1,
357
+
[]string{"d2", "d3", "d1", "d4"},
358
+
},
359
+
{
360
+
"ranks only",
361
+
math.Inf(-1),
362
+
[]string{"d3", "d1", "d4", "d2"},
363
+
},
364
+
}
343
365
344
-
SortFiles(in, &SearchOptions{UseDocumentRanks: true, DebugScore: true})
366
+
for _, tt := range cases {
367
+
t.Run("", func(t *testing.T) {
345
368
346
-
wantOrder := []string{"d3", "d2", "d1", "d4"}
369
+
SortFiles(in, &SearchOptions{UseDocumentRanks: true, DebugScore: true, RanksDampingFactor: tt.dampingFactor})
347
370
348
-
var haveOrder = []string{}
349
-
for _, f := range in {
350
-
haveOrder = append(haveOrder, f.FileName)
351
-
}
371
+
var haveOrder = []string{}
372
+
for _, f := range in {
373
+
haveOrder = append(haveOrder, f.FileName)
374
+
}
352
375
353
-
if d := cmp.Diff(wantOrder, haveOrder); d != "" {
354
-
t.Fatalf("-want, +got\n%s\n", d)
376
+
if d := cmp.Diff(tt.wantOrder, haveOrder); d != "" {
377
+
t.Fatalf("-want, +got\n%s\n", d)
378
+
}
379
+
})
355
380
}
356
381
}