[Geocoding] help! something went awry with my Nominatim install

Tue Jun 8 15:50:55 BST 2010

Hi there,

hoping someone may have a pointer as to where we went wrong setting up
Gazetteer & Nominatim.

we only used a UK extract, not a full planet.  we used this UK extract:
http://nick.dev.openstreetmap.org/downloads/planet/uk-091028.osm.bz2

The indexing took 2-4 days on an absolute honking machine (200+ gb of ram,
16 processors) - but I am pretty sure we dont have postgres configured
properly to take up memory and processor power (am going to have to learn
that now too )

anyway - Here is the symptom - please compare these to
http://nominatim.openstreetmap.org:

The query "pubs in south benfleet" returns no results.

The query "pubs in benfleet" returns "The Belgrave Arms, Benfleet Court,
Haggerston, City, London Borough of Hackney, Greater London, United Kingdom"
as the top result

The query "pubs in london" returns "The London, High Street,
Weston-super-Mare, North Somerset, United Kingdom"as the top result (in fact
the only result).

Obviously something has gone wrong.  At first I was thinking maybe our
extract didnt really have all the data, until I realized that it seemed to
be keying on the city name as the potential pub name or street name

Putting nominatim into debug mode produced the output at the bottom of this
email.

I am still pretty sure that we dont have all the data, but still - I KNOW we
DO have pubs in London in the data, so "pubs in London" should still have
returned at least some pubs in London, right?

Thanks in advance for any advice or suggestions on where to look for what
went wrong

Ant

search?q=pubs+in+benfleet
Nominatim Debug Output:

string(137) "select word_id,word_token, word, class, type,
location,country_code from word where word_token in (' pubs','pubs','
benfleet','benfleet')" array(2) { [0]=> array(3) { ["string"]=> string(4)
"pubs" ["words"]=> array(1) { [0]=> string(4) "pubs" } ["wordsets"]=>
array(1) { [0]=> array(1) { [0]=> string(4) "pubs" } } } [1]=> array(3) {
["string"]=> string(8) "benfleet" ["words"]=> array(1) { [0]=> string(8)
"benfleet" } ["wordsets"]=> array(1) { [0]=> array(1) { [0]=> string(8)
"benfleet" } } } } array(3) { [" benfleet"]=> array(1) { [0]=> array(7) {
["word_id"]=> string(4) "8801" ["word_token"]=> string(9) " benfleet"
["word"]=> string(8) "Benfleet" ["class"]=> NULL ["type"]=> NULL
["location"]=> NULL ["country_code"]=> NULL } } ["benfleet"]=> array(1) {
[0]=> array(7) { ["word_id"]=> string(4) "8802" ["word_token"]=> string(8)
"benfleet" ["word"]=> NULL ["class"]=> NULL ["type"]=> NULL ["location"]=>
NULL ["country_code"]=> NULL } } [" pubs"]=> array(1) { [0]=> array(7) {
["word_id"]=> string(6) "532099" ["word_token"]=> string(5) " pubs"
["word"]=> NULL ["class"]=> string(7) "amenity" ["type"]=> string(3) "pub"
["location"]=> NULL ["country_code"]=> NULL } } } array(2) { [2]=> array(1)
{ [0]=> array(11) { ["iSearchRank"]=> int(2) ["iNamePhrase"]=> int(1)
["sCountryCode"]=> bool(false) ["aName"]=> array(1) { [8801]=> string(4)
"8801" } ["aAddress"]=> array(1) { [8801]=> string(4) "8801" } ["sClass"]=>
string(7) "amenity" ["sType"]=> string(3) "pub" ["sHouseNumber"]=> string(0)
"" ["fLat"]=> string(0) "" ["fLon"]=> string(0) "" ["fRadius"]=> string(0)
"" } } [11]=> array(1) { [0]=> array(11) { ["iSearchRank"]=> int(11)
["iNamePhrase"]=> int(1) ["sCountryCode"]=> bool(false) ["aName"]=> array(1)
{ [8802]=> string(4) "8802" } ["aAddress"]=> array(1) { [8802]=> string(4)
"8802" } ["sClass"]=> string(7) "amenity" ["sType"]=> string(3) "pub"
["sHouseNumber"]=> string(0) "" ["fLat"]=> string(0) "" ["fLon"]=> string(0)
"" ["fRadius"]=> string(0) "" } } } rankName TokensAddress Tokenscountry
classtypehouse#LatLonRadius2# benfleet#

amenitypub

11#benfleet#

amenitypub

string(4) "
------------------------------
" array(11) { ["iSearchRank"]=> int(2) ["iNamePhrase"]=> int(1)
["sCountryCode"]=> bool(false) ["aName"]=> array(1) { [8801]=> string(4)
"8801" } ["aAddress"]=> array(1) { [8801]=> string(4) "8801" } ["sClass"]=>
string(7) "amenity" ["sType"]=> string(3) "pub" ["sHouseNumber"]=> string(0)
"" ["fLat"]=> string(0) "" ["fLon"]=> string(0) "" ["fRadius"]=> string(0)
"" } rankName TokensAddress Tokenscountryclasstypehouse#LatLonRadius0#
benfleet#

amenitypub

string(523) "select
place_id,ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-225,83.14),ST_Point(225,-76.06)),4326),
centroid) as
in_small,ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-675,242.34),ST_Point(675,-235.26)),4326),
centroid) as in_large from search_name where name_vector @> ARRAY[8801]
order by
ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-225,83.14),ST_Point(225,-76.06)),4326),
centroid) desc,
ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-675,242.34),ST_Point(675,-235.26)),4326),
centroid) desc, search_rank ASC limit 10" string(127) "select place_id from
placex where place_id in (19334,3794) and class='amenity' and type='pub'
order by rank_search asc limit 10" string(241) "select l.place_id from
placex as l,placex as f where f.place_id in (19334,3794) and
ST_DWithin(l.geometry, f.geometry, 0.01) and l.class='amenity' and
l.type='pub' order by ST_Distance(l.geometry, f.geometry) asc, l.rank_search
ASC limit 10" array(0) { } string(4) "
------------------------------
" array(11) { ["iSearchRank"]=> int(11) ["iNamePhrase"]=> int(1)
["sCountryCode"]=> bool(false) ["aName"]=> array(1) { [8802]=> string(4)
"8802" } ["aAddress"]=> array(1) { [8802]=> string(4) "8802" } ["sClass"]=>
string(7) "amenity" ["sType"]=> string(3) "pub" ["sHouseNumber"]=> string(0)
"" ["fLat"]=> string(0) "" ["fLon"]=> string(0) "" ["fRadius"]=> string(0)
"" } rankName TokensAddress Tokenscountryclasstypehouse#LatLonRadius0
#benfleet#

amenitypub

string(523) "select
place_id,ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-225,83.14),ST_Point(225,-76.06)),4326),
centroid) as
in_small,ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-675,242.34),ST_Point(675,-235.26)),4326),
centroid) as in_large from search_name where name_vector @> ARRAY[8802]
order by
ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-225,83.14),ST_Point(225,-76.06)),4326),
centroid) desc,
ST_Contains(ST_SetSRID(ST_MakeBox2D(ST_Point(-675,242.34),ST_Point(675,-235.26)),4326),
centroid) desc, search_rank ASC limit 10" string(185) "select place_id from
placex where place_id in
(19334,22243,22244,453174,738688,738689,772432,698300,1239202,1033786) and
class='amenity' and type='pub' order by rank_search asc limit 10"
string(299) "select l.place_id from placex as l,placex as f where f.place_id
in (19334,22243,22244,453174,738688,738689,772432,698300,1239202,1033786)
and ST_DWithin(l.geometry, f.geometry, 0.01) and l.class='amenity' and
l.type='pub' order by ST_Distance(l.geometry, f.geometry) asc, l.rank_search
ASC limit 10" array(10) { [0]=> string(5) "38547" [1]=> string(6) "120706"
[2]=> string(5) "92761" [3]=> string(5) "92765" [4]=> string(5) "41184"
[5]=> string(5) "37195" [6]=> string(5) "34616" [7]=> string(5) "34615"
[8]=> string(5) "92770" [9]=> string(5) "41176" } string(4) "
------------------------------
" string(1192) "select
osm_type,osm_id,class,type,rank_search,rank_address,min(place_id) as
place_id,country_code,get_address_by_language(place_id,
ARRAY['name:en-us','name:en','name','ref','type']) as
langaddress,get_name_by_language(name,
ARRAY['name:en-us','name:en','name','ref','type']) as
placename,get_name_by_language(name, ARRAY['ref']) as
ref,avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry)))
as lat from placex where place_id in
(38547,120706,92761,92765,41184,37195,34616,34615,92770,41176) group by
osm_type,osm_id,class,type,rank_search,rank_address,country_code,get_address_by_language(place_id,
ARRAY['name:en-us','name:en','name','ref','type'])
,get_name_by_language(name,
ARRAY['name:en-us','name:en','name','ref','type'])
,get_name_by_language(name, ARRAY['ref']) order by
rank_search,rank_address,CASE when min(place_id) = 38547 then 0 when
min(place_id) = 120706 then 1 when min(place_id) = 92761 then 2 when
min(place_id) = 92765 then 3 when min(place_id) = 41184 then 4 when
min(place_id) = 37195 then 5 when min(place_id) = 34616 then 6 when
min(place_id) = 34615 then 7 when min(place_id) = 92770 then 8 when
min(place_id) = 41176 then 9 ELSE 10000000 END ASC"
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/geocoding/attachments/20100608/df650323/attachment.html>