change distance computation between place and address part

Instead of computing the distance to the centroid of the area
compute the distance of the area to the centroid of the feature.
This means we give preference to the area that covers the centroid.
It's still a heuristics but one that is a bit less random.
This commit is contained in:
Sarah Hoffmann
2022-04-21 21:56:59 +02:00
parent 403e6f7e5c
commit 784dad866f
3 changed files with 10 additions and 5 deletions

View File

@@ -39,7 +39,10 @@ BEGIN
END END
$$ LANGUAGE plpgsql IMMUTABLE; $$ LANGUAGE plpgsql IMMUTABLE;
create or replace function getNearFeatures(in_partition INTEGER, feature GEOMETRY, maxrank INTEGER) RETURNS setof nearfeaturecentr AS $$ CREATE OR REPLACE function getNearFeatures(in_partition INTEGER, feature GEOMETRY,
feature_centroid GEOMETRY,
maxrank INTEGER)
RETURNS setof nearfeaturecentr AS $$
DECLARE DECLARE
r nearfeaturecentr%rowtype; r nearfeaturecentr%rowtype;
BEGIN BEGIN
@@ -48,7 +51,7 @@ BEGIN
IF in_partition = {{ partition }} THEN IF in_partition = {{ partition }} THEN
FOR r IN FOR r IN
SELECT place_id, keywords, rank_address, rank_search, SELECT place_id, keywords, rank_address, rank_search,
min(ST_Distance(feature, centroid)) as distance, min(ST_Distance(feature_centroid, geometry)) as distance,
isguess, postcode, centroid isguess, postcode, centroid
FROM location_area_large_{{ partition }} FROM location_area_large_{{ partition }}
WHERE geometry && feature WHERE geometry && feature

View File

@@ -449,6 +449,7 @@ CREATE OR REPLACE FUNCTION insert_addresslines(obj_place_id BIGINT,
maxrank SMALLINT, maxrank SMALLINT,
token_info JSONB, token_info JSONB,
geometry GEOMETRY, geometry GEOMETRY,
centroid GEOMETRY,
country TEXT, country TEXT,
OUT parent_place_id BIGINT, OUT parent_place_id BIGINT,
OUT postcode TEXT, OUT postcode TEXT,
@@ -511,7 +512,7 @@ BEGIN
END LOOP; END LOOP;
FOR location IN FOR location IN
SELECT * FROM getNearFeatures(partition, geometry, maxrank) SELECT * FROM getNearFeatures(partition, geometry, centroid, maxrank)
WHERE not addr_place_ids @> ARRAY[place_id] WHERE not addr_place_ids @> ARRAY[place_id]
ORDER BY rank_address, isguess asc, ORDER BY rank_address, isguess asc,
distance * distance *
@@ -1106,7 +1107,8 @@ BEGIN
END IF; END IF;
SELECT * FROM insert_addresslines(NEW.place_id, NEW.partition, max_rank, SELECT * FROM insert_addresslines(NEW.place_id, NEW.partition, max_rank,
NEW.token_info, geom, NEW.country_code) NEW.token_info, geom, NEW.centroid,
NEW.country_code)
INTO NEW.parent_place_id, NEW.postcode, nameaddress_vector; INTO NEW.parent_place_id, NEW.postcode, nameaddress_vector;
{% if debug %}RAISE WARNING 'RETURN insert_addresslines: %, %, %', NEW.parent_place_id, NEW.postcode, nameaddress_vector;{% endif %} {% if debug %}RAISE WARNING 'RETURN insert_addresslines: %, %, %', NEW.parent_place_id, NEW.postcode, nameaddress_vector;{% endif %}

View File

@@ -34,7 +34,7 @@ BEGIN
NEW.parent_place_id = 0; NEW.parent_place_id = 0;
FOR location IN FOR location IN
SELECT place_id SELECT place_id
FROM getNearFeatures(partition, NEW.geometry, NEW.rank_search) FROM getNearFeatures(partition, NEW.geometry, NEW.geometry, NEW.rank_search)
WHERE NOT isguess ORDER BY rank_address DESC, distance asc LIMIT 1 WHERE NOT isguess ORDER BY rank_address DESC, distance asc LIMIT 1
LOOP LOOP
NEW.parent_place_id = location.place_id; NEW.parent_place_id = location.place_id;