--
Datameet is a community of Data Science enthusiasts in India. Know more about us by visiting http://datameet.org
---
You received this message because you are subscribed to the Google Groups "datameet" group.
To unsubscribe from this group and stop receiving emails from it, send an email to datameet+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/a672e807-c8dd-4dba-8d79-f407aa076cb1%40googlegroups.com.
OSM wiki page listing several different software for searching places:
https://wiki.openstreetmap.org/wiki/Search_engines
Osmnames + Sphinxsearch :
This seemed easier to start off so tried it. Once I got it depoyed, didn't get any proper results even for city names. I'm guessing the parameters may require fine-tuning. And I couldn't understand why they are involving wikipedia. It seemed like the search was relying more on data from wikipedia than from OSM.
Didn't explore on it further. (Spare time burned up!)
Nominatim related leads, where I want to dig into in next sprint:
https://www.cascadia-analytics.com/2017/10/01/nominatim-docker.html
https://github.com/mediagis/nominatim-docker
Pelias seems promising too, but we need a proper elasticsearch deployment to run it and that takes up quite some memory.
-----
My personal observations, may be inaccurate:
We should treat geocoding and reverse geocoding as two different applications.
Reverse geocoding would be much simpler to do, since you just need to put in the lat-longs and cough up the closest place names.
Geocoding involves complicated textual search, especially given the variations in ways place names are stored. There has to be high flexibility like making up for spelling differences. So elasticsearch seems like a good fit for it. Then, geo-constraining searches is very important here. We need ways to progressively narrow down the search areas and cut out false positives. Letting the user draw a box on the map to search in, for example, can greatly increase effectiveness of results rather than searching the whole country / planet data.
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/CADbFdW4_1DrFTrQvpTdBuBmW2KMmm3Kxr8D84nmKK9Viwt6fmg%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/CAH7jeuOFsm2NgkWWQt_582OW7UGithgqFAp02-hFC4%3D%3DWO_0eQ%40mail.gmail.com.
--
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/CAKmBd7tjac15zuV7raY5m3AwPGCCxee64FT1qMc_PTOKsjXUmg%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/CALQA-LXH4yNvAgTZnLYA2cgbQQfkTmk3UUDnND5DhYuC3e6L%2BQ%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/CA%2B4SsQAXibDzWSp%3DV1gk%2Btd5bwTSkrycziyLzvvOkcQuMEXd2Q%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/datameet/CAEKTdxrh0S8gYH2cze5Ldo5rXvubT%3D0RCOfAh8h%2BKTkpd5kG5g%40mail.gmail.com.