hans/Nominatim
1
0
Fork 1
mirror of https://github.com/osm-search/Nominatim.git synced 2026-03-10 03:54:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

experimental tweaking of ranks of searches

Browse Source 
- completely removes partial words with spaces from addresses
- ensure that frequent partial search terms appear in NonAddress
- higher penalty for partial words consisting of numbers
This commit is contained in:
Sarah Hoffmann
2014-05-01 17:53:13 +02:00
parent fd6db5a13f
commit 4d6da459da
1 changed files with 8 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
lib/Geocode.php
View File

@@ -932,7 +932,7 @@
{ {
if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
{ {
if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false)
{ {
$aSearch = $aCurrentSearch; $aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1; $aSearch['iSearchRank'] += 1;
@@ -941,8 +941,11 @@
$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
} }
elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version? elseif (isset($aValidTokens[' '.$sToken]) && strlen($sToken) >= 4) // revert to the token version?
{ {
$aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
$aSearch['iSearchRank'] += 1;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
foreach($aValidTokens[' '.$sToken] as $aSearchTermToken) foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
{ {
if (empty($aSearchTermToken['country_code']) if (empty($aSearchTermToken['country_code'])
@@ -959,6 +962,7 @@
else else
{ {
$aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
} }
} }
@@ -966,7 +970,8 @@
if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
{ {
$aSearch = $aCurrentSearch; $aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 2; $aSearch['iSearchRank'] += 1;
if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;
if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];