From b3abb355ebcf06fe47d450e7c2fc5074ae6f4b7c Mon Sep 17 00:00:00 2001 From: Sarah Hoffmann Date: Thu, 29 Sep 2022 14:55:46 +0200 Subject: [PATCH] docs: add customization hints for secondary importance Removing the download links for now as the tile importance is still too experimental. --- docs/admin/Import.md | 23 +++-------------- docs/customize/Importance.md | 49 ++++++++++++++++++++++++++++++++++++ docs/mkdocs.yml | 1 + 3 files changed, 53 insertions(+), 20 deletions(-) create mode 100644 docs/customize/Importance.md diff --git a/docs/admin/Import.md b/docs/admin/Import.md index 91b92728..8b6d6baa 100644 --- a/docs/admin/Import.md +++ b/docs/admin/Import.md @@ -78,27 +78,10 @@ This data is available as a binary download. Put it into your project directory: The file is about 400MB and adds around 4GB to the Nominatim database. -### OSM views -OSM publishes aggregate map access numbers that are generated based on the users’ -behavior when viewing locations on the map. This data is also optional and -it complements wikipedia/wikidata rankings to further enhance the search results -if added. -OSM views data is avalaible as a GeoTIFF file. Put it into your project directory: - - cd $PROJECT_DIR - wget https://qrank.wmcloud.org/download/osmviews.tiff - -The file is about 380MB and adds around 4GB to the Nominatim database. Importing -OSM views into Nominatim takes a little over 3 hours. - -!!! warning - Importing OSM views is currently an experimental feature. OSM views data are - not yet included in the importance values calculations. - !!! tip - If you forgot to download the wikipedia rankings or OSM views, then you can - also add importances after the import. To add both, download their files, then - run `nominatim refresh --wiki-data --osm-views --importance`. Updating + If you forgot to download the wikipedia rankings, then you can + also add importances after the import. Download the SQL files, then + run `nominatim refresh --wiki-data --importance`. Updating importances for a planet will take a couple of hours. ### External postcodes diff --git a/docs/customize/Importance.md b/docs/customize/Importance.md new file mode 100644 index 00000000..d12bfc86 --- /dev/null +++ b/docs/customize/Importance.md @@ -0,0 +1,49 @@ +## Importance + +Search requests can yield multiple results which match equally well with +the original query. In such case Nominatim needs to order the results +according to a different criterion: importance. This is a measure for how +likely it is that a user will search for a given place. This section explains +the sources Nominatim uses for computing importance of a place and how to +customize them. + +### How importance is computed + +The main value for importance is derived from page ranking values for Wikipedia +pages for a place. For places that do not have their own +Wikipedia page, a formula is used that derives a static importance from the +places [search rank](../customize/Ranking#search-rank). + +In a second step, a secondary importance value is added which is meant to +represent how well-known the general area is where the place is located. It +functions as a tie-breaker between places with very similar primary +importance values. + +nominatim.org has preprocessed importance tables for the +[primary Wikipedia rankings](https://nominatim.org/data/wikimedia-importance.sql.gz) +and for a secondary importance based on the number of tile views on openstreetmap.org. + +### Customizing secondary importance + +The secondary importance is implemented as a simple +[Postgis raster](https://postgis.net/docs/raster.html) table, where Nominatim +looks up the value for the coordinates of the centroid of a place. You can +provide your own secondary importance raster in form of an SQL file named +`secondary_importance.sql.gz` in your project directory. + +The SQL file needs to drop and (re)create a table `secondary_importance` which +must as a minimum contain a column `rast` of type `raster`. The raster must +be in EPSG:4326 and contain 16bit unsigned ints +(`raster_constraint_pixel_types(rast) = '{16BUI}'). Any other columns in the +table will be ignored. You must furthermore create an index as follows: + +``` +CREATE INDEX ON secondary_importance USING gist(ST_ConvexHull(gist)) +``` + +The following raster2pgsql command will create a table that conforms to +the requirements: + +``` +raster2pgsql -I -C -Y -d -t 128x128 input.tiff public.secondary_importance +``` diff --git a/docs/mkdocs.yml b/docs/mkdocs.yml index e89c32d5..ab7dec30 100644 --- a/docs/mkdocs.yml +++ b/docs/mkdocs.yml @@ -30,6 +30,7 @@ nav: - 'Configuration Settings': 'customize/Settings.md' - 'Per-Country Data': 'customize/Country-Settings.md' - 'Place Ranking' : 'customize/Ranking.md' + - 'Importance' : 'customize/Importance.md' - 'Tokenizers' : 'customize/Tokenizers.md' - 'Special Phrases': 'customize/Special-Phrases.md' - 'External data: US housenumbers from TIGER': 'customize/Tiger.md' -- 2.45.1