[OSM-dev] Related terms in the wiki for improved tag search

[Stefan Keller]

We are currently experimenting with our TagFinder
(http://152.96.56.32/poiservice/tagfinder ) and want to improve the
search for OSM tags.

One of the crucial point there is that synonyms and related terms are
found (term being one ore more words). Example: A search for "church"
will show "amenity=place_of_worship".

In order to achieve this a controlled word list (a thesaurus) becomes
necessary. This is a list of terms where some are OpenStreetMap
specific. This can not be found in a general purpose thesaurus.

=> Because of this, I propose to capture in the OSM wiki so-called
"related terms". They can be kept up to date by the community (while I
would like to set a good example).

Of course it would be more precise to introduce also synonyms and
broader/narrower terms. But that seems too complicated for many users.

=> In order to store related terms I propose to use wiki templates.
That would look like this on a OSM wiki page (eg.
http://wiki.openstreetmap.org/wiki/Tag:amenity%3Dplace_of_worship ):

  {{RelatedTerm|place of worship}}
  {{RelatedTerm|church}}
  {{RelatedTerm|mosque}}
  {{RelatedTerm|synagogue}}
  {{RelatedTerm|en_GB|devotion building}}
  {{RelatedTerm|en_US|prayer house}}

The related terms are parsed from the Wiki pages by Taginfo during the
update process and stored in Taginfo database (as it's already doing
for the "tag" template). The terms lists are then passed over to the
Taginfo API. BTW: The "preferred term" ("Preferred tag") can be
estimated by the tag statistics API of Taginfo.

Theoretically parts of this code could be ported over to the search
Taginfo (the Taginfo API is already used now by our TagFinder).

=> What do you think?

Yours, Stefan