* Move to in-core fts function, instead of the compat wrapper provided by the tsearch2 contrib
* Provide default cover density tuning (config file)
* Move default preferred language settings from storage to search, where they make more sense
More on the CD tuning:
Evergreen uses a cover density algorithm for calculating relative ranking of matches. There
are several tuning parameters and options available. By default, no document length normalization
is applied. From the Postgres documentation on ts_rank_cd() (the function used by Evergreen):
Since a longer document has a greater chance of containing a query term it is reasonable
to take into account document size, e.g., a hundred-word document with five instances of
a search word is probably more relevant than a thousand-word document with five instances.
Both ranking functions take an integer normalization option that specifies whether and how
a document's length should impact its rank. The integer option controls several behaviors,
so it is a bit mask: you can specify one or more behaviors using | (for example, 2|4).
0 (the default) ignores the document length
1 divides the rank by 1 + the logarithm of the document length
2 divides the rank by the document length
4 divides the rank by the mean harmonic distance between extents (this is implemented only by ts_rank_cd)
8 divides the rank by the number of unique words in document
16 divides the rank by 1 + the logarithm of the number of unique words in document
32 divides the rank by itself + 1
If more than one flag bit is specified, the transformations are applied in the order listed.
It is important to note that the ranking functions do not use any global information, so it
is impossible to produce a fair normalization to 1% or 100% as sometimes desired. Normalization
option 32 (rank/(rank+1)) can be applied to scale all ranks into the range zero to one, but of
course this is just a cosmetic change; it will not affect the ordering of the search results.
In Evergreen, these options are set via search modifiers. The modifiers are mapped in the
following way:
* #CD_logDocumentLength => 1 :: rank / (1 + LOG(total_word_count)) :: Longer documents slightly less relevant
* #CD_documentLength => 2 :: rank / total_word_count :: Longer documents much less relevant
* #CD_meanHarmonic => 4 :: Word Proximity :: Greater matched-word distance is less relevant
* #CD_uniqueWords => 8 :: rank / unique_word_count :: Documents with repeated words much less relevant
* #CD_logUniqueWords => 16 :: rank / (1 + LOG(unique_word_count)) :: Documents with repeated words slightly less relevant
* #CD_selfPlusOne => 32 :: rank / (1 + rank) :: Cosmetic normalization of rank value between 0 and 1
Adding one or more of these to the default_CD_modifiers list will cause all searches that use QueryParser to apply them.
git-svn-id: svn://svn.open-ils.org/ILS/trunk@20180
dcc99617-32d9-48b4-a31d-
7c20da2025e4
-->
<estimation_strategy>inclusion</estimation_strategy>
+ <!--
+ Evergreen uses a cover density algorithm for calculating relative ranking of matches. There
+ are several tuning parameters and options available. By default, no document length normalization
+ is applied. From the Postgres documentation on ts_rank_cd() (the function used by Evergreen):
+
+ Since a longer document has a greater chance of containing a query term it is reasonable
+ to take into account document size, e.g., a hundred-word document with five instances of
+ a search word is probably more relevant than a thousand-word document with five instances.
+ Both ranking functions take an integer normalization option that specifies whether and how
+ a document's length should impact its rank. The integer option controls several behaviors,
+ so it is a bit mask: you can specify one or more behaviors using | (for example, 2|4).
+
+ 0 (the default) ignores the document length
+
+ 1 divides the rank by 1 + the logarithm of the document length
+
+ 2 divides the rank by the document length
+
+ 4 divides the rank by the mean harmonic distance between extents (this is implemented only by ts_rank_cd)
+
+ 8 divides the rank by the number of unique words in document
+
+ 16 divides the rank by 1 + the logarithm of the number of unique words in document
+
+ 32 divides the rank by itself + 1
+
+ If more than one flag bit is specified, the transformations are applied in the order listed.
+
+ It is important to note that the ranking functions do not use any global information, so it
+ is impossible to produce a fair normalization to 1% or 100% as sometimes desired. Normalization
+ option 32 (rank/(rank+1)) can be applied to scale all ranks into the range zero to one, but of
+ course this is just a cosmetic change; it will not affect the ordering of the search results.
+
+ In Evergreen, these options are set via search modifiers. The modifiers are mapped in the
+ following way:
+
+ * #CD_logDocumentLength => 1 :: rank / (1 + LOG(total_word_count)) :: Longer documents slightly less relevant
+ * #CD_documentLength => 2 :: rank / total_word_count :: Longer documents much less relevant
+ * #CD_meanHarmonic => 4 :: Word Proximity :: Greater matched-word distance is less relevant
+ * #CD_uniqueWords => 8 :: rank / unique_word_count :: Documents with repeated words much less relevant
+ * #CD_logUniqueWords => 16 :: rank / (1 + LOG(unique_word_count)) :: Documents with repeated words slightly less relevant
+ * #CD_selfPlusOne => 32 :: rank / (1 + rank) :: Cosmetic normalization of rank value between 0 and 1
+
+ Adding one or more of these to the default_CD_modifiers list will cause all searches that use QueryParser to apply them.
+ -->
+ <default_CD_modifiers>#CD_documentLength #CD_meanHarmonic #CD_uniqueWords</default_CD_modifiers>
+
+ <!--
+ default_preferred_language
+ Set the global, default preferred languange
+ -->
+ <default_preferred_language>eng</default_preferred_language>
+
+ <!--
+ default_preferred_language_weight
+ Set the weight (higher is "better") for the preferred language. Comment out to remove all lanuage weighting by default.
+ -->
+ <default_preferred_language_weight>5</default_preferred_language_weight>
+
<!-- Baseline number of records to check for hit estimation. -->
<superpage_size>1000</superpage_size>
<max_spare_children>5</max_spare_children>
</unix_config>
<app_settings>
- <!-- default_preferred_language: Set the global, default preferred languange -->
- <default_preferred_language>eng</default_preferred_language>
-
- <!-- default_preferred_language_weight: Set the weight (higher is "better") for the preferred language -->
- <default_preferred_language_weight>5</default_preferred_language_weight>
-
<script_path>LIBDIR/javascript/</script_path>
<script_path>LOCALSTATEDIR/catalog/</script_path>
<scripts>
}
return $self->{rank} if ($self->{rank});
- return $self->{rank} = 'rank_cd(' . $self->table_alias . '.index_vector, ' . $self->table_alias . ".tsq, $cover_density)";
+ return $self->{rank} = 'ts_rank_cd(' . $self->table_alias . '.index_vector, ' . $self->table_alias . ".tsq, $cover_density)";
}
my @ranks;
for my $fts ( $self->fts_query ) {
push @output, join(' ', $self->fts_col, $self->{fts_op}, $fts);
- push @ranks, "rank($column, $fts)";
+ push @ranks, "ts_rank($column, $fts)";
}
$self->{fts_rank} = \@ranks;
}
+ my $config = OpenSRF::Utils::SettingsClient->new();
+
if (!$default_preferred_language) {
- $default_preferred_language = OpenSRF::Utils::SettingsClient
- ->new
- ->config_value(
+ $default_preferred_language = $config->config_value(
+ apps => 'open-ils.search' => app_settings => 'default_preferred_language'
+ ) || $config->config_value(
apps => 'open-ils.storage' => app_settings => 'default_preferred_language'
);
if (!$default_preferred_language_weight) {
- $default_preferred_language_weight = OpenSRF::Utils::SettingsClient
- ->new
- ->config_value(
+ $default_preferred_language_weight = $config->config_value(
+ apps => 'open-ils.storage' => app_settings => 'default_preferred_language_weight'
+ ) || $config->config_value(
apps => 'open-ils.storage' => app_settings => 'default_preferred_language_weight'
);
-
}
# inclusion, exclusion, delete_adjusted_inclusion, delete_adjusted_exclusion
die "No query was passed to ".$self->api_name;
}
+ my $default_CD_modifiers = OpenSRF::Utils::SettingsClient->new->config_value(
+ apps => 'open-ils.search' => app_settings => 'default_CD_modifiers'
+ );
+ $args{query} = "$default_CD_modifiers $args{query}" if ($default_CD_modifiers);
+
my $simple_plan = $args{_simple_plan};
# remove bad chunks of the %args hash
# we expect, and make use of, query, superpage, superpage_size, debug and core_limit args
my $query = $parser->new( %args )->parse;
+ my $config = OpenSRF::Utils::SettingsClient->new();
# set the locale-based default prefered location
if (!$query->parse_tree->find_filter('preferred_language')) {
$parser->default_preferred_language( $args{preferred_language} );
+
if (!$parser->default_preferred_language) {
my $ses_locale = $client->session ? $client->session->session_locale : '';
$parser->default_preferred_language( $locale_map{ lc($ses_locale) } );
}
- $parser->default_preferred_language(
- OpenSRF::Utils::SettingsClient->new->config_value(
+
+ if (!$parser->default_preferred_language) { # still nothing...
+ my $tmp_dpl = $config->config_value(
+ apps => 'open-ils.search' => app_settings => 'default_preferred_language'
+ ) || $config->config_value(
apps => 'open-ils.storage' => app_settings => 'default_preferred_language'
- )
- ) if (!$parser->default_preferred_language);
+ );
+
+ $parser->default_preferred_language( $tmp_dpl )
+ }
}
# set the global default language multiplier
if (!$query->parse_tree->find_filter('preferred_language_weight') and !$query->parse_tree->find_filter('preferred_language_multiplier')) {
- $parser->default_preferred_language_multiplier($args{preferred_language_weight});
- $parser->default_preferred_language_multiplier($args{preferred_language_multiplier});
- $parser->default_preferred_language_multiplier(
- OpenSRF::Utils::SettingsClient->new->config_value(
+ my $tmp_dplw;
+
+ if ($tmp_dplw = $args{preferred_language_weight} || $args{preferred_language_multiplier} ) {
+ $parser->default_preferred_language_multiplier($tmp_dplw);
+
+ } else {
+ $tmp_dplw = $config->config_value(
+ apps => 'open-ils.search' => app_settings => 'default_preferred_language_weight'
+ ) || $config->config_value(
apps => 'open-ils.storage' => app_settings => 'default_preferred_language_weight'
- )
- ) if (!$parser->default_preferred_language_multiplier);
+ );
+
+ $parser->default_preferred_language_multiplier( $tmp_dplw );
+ }
}
# gather the site, if one is specified, defaulting to the in-query version
projects / evergreen / bjwebb.git / commitdiff