[
https://issues.apache.org/jira/browse/LUCENE-8311?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16484120#comment-16484120
]
Adrien Grand commented on LUCENE-8311:
--------------------------------------
Here is a run with DFR I(ne)L1:
{noformat}
LowPhrase 19.89 (1.2%) 16.59 (1.0%)
-16.6% ( -18% - -14%)
MedPhrase 15.94 (1.2%) 13.36 (1.1%)
-16.1% ( -18% - -14%)
HighTermMonthSort 90.26 (10.9%) 81.72 (11.6%)
-9.5% ( -28% - 14%)
HighSloppyPhrase 1.84 (1.9%) 1.69 (2.2%)
-7.9% ( -11% - -3%)
LowSloppyPhrase 7.87 (2.0%) 7.28 (2.5%)
-7.4% ( -11% - -3%)
MedSloppyPhrase 10.17 (1.6%) 9.43 (2.0%)
-7.3% ( -10% - -3%)
HighTermDayOfYearSort 64.33 (11.6%) 60.25 (10.4%)
-6.3% ( -25% - 17%)
HighTerm 476.13 (2.5%) 452.30 (1.8%)
-5.0% ( -9% - 0%)
Fuzzy1 211.47 (4.1%) 203.28 (3.3%)
-3.9% ( -10% - 3%)
IntNRQ 31.99 (2.5%) 30.96 (7.6%)
-3.2% ( -12% - 6%)
MedTerm 653.93 (2.4%) 634.02 (1.8%)
-3.0% ( -7% - 1%)
Fuzzy2 218.64 (5.9%) 212.25 (5.4%)
-2.9% ( -13% - 8%)
OrHighHigh 17.28 (1.6%) 16.93 (1.7%)
-2.0% ( -5% - 1%)
LowTerm 1405.19 (2.9%) 1380.15 (2.3%)
-1.8% ( -6% - 3%)
AndHighHigh 21.96 (2.1%) 21.62 (2.5%)
-1.5% ( -5% - 3%)
OrHighMed 59.73 (1.5%) 58.89 (1.7%)
-1.4% ( -4% - 1%)
Prefix3 73.07 (4.8%) 72.07 (5.8%)
-1.4% ( -11% - 9%)
Wildcard 64.42 (3.6%) 63.72 (4.5%)
-1.1% ( -8% - 7%)
Respell 181.31 (2.4%) 180.69 (2.3%)
-0.3% ( -4% - 4%)
AndHighLow 982.32 (2.5%) 981.63 (3.1%)
-0.1% ( -5% - 5%)
AndHighMed 47.62 (2.0%) 47.60 (2.5%)
-0.0% ( -4% - 4%)
LowSpanNear 49.59 (3.4%) 49.65 (3.0%)
0.1% ( -6% - 6%)
OrHighLow 314.16 (2.2%) 314.60 (1.7%)
0.1% ( -3% - 4%)
HighSpanNear 5.92 (4.6%) 5.98 (4.1%)
1.0% ( -7% - 10%)
MedSpanNear 5.53 (6.7%) 5.66 (5.5%)
2.2% ( -9% - 15%)
HighPhrase 3.87 (1.5%) 4.36 (1.6%)
12.6% ( 9% - 15%)
{noformat}
> Leverage impacts for phrase queries
> -----------------------------------
>
> Key: LUCENE-8311
> URL: https://issues.apache.org/jira/browse/LUCENE-8311
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Priority: Minor
> Attachments: LUCENE-8311.patch
>
>
> Now that we expose raw impacts, we could leverage them for phrase queries.
> For instance for exact phrases, we could take the minimum term frequency for
> each unique norm value in order to get upper bounds of the score for the
> phrase.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
