Inconsistent Autocomplete about api

[hkrishna]

This is similar to the empire state bug. As you type empire state you see
that st is analyzed as street and sta to station.. similarly, I think ci is
being analyzed as citadel or something and since we use synonyms as one of
our token filters on name.default field this is a known problem.

I have a few solutions in mind that involves extending the synonyms
filter..
On Feb 19, 2015 11:45 AM, "Peter Johnson a.k.a. insertcoffee" <
[email protected]> wrote:

This UX issue is somewhat subjective, so I'll try to explain why I think
it is a problem.

Consider the gif below:
[image: ss]
https://camo.githubusercontent.com/97d330f3a80c972d0cd63101be7461920a6482da/687474703a2f2f70657465722e6a6f686e736f6e2e73332e616d617a6f6e6177732e636f6d2f756e7072656469637461626c652e676966

As the amount of characters are added to the query "Hackney City Farm" we
get very different results back, those results differ both in which layers
the data comes from; the ratio of local/distant entries and the total
amount of items returned.

For all characters up to and including "hackney " we only get
administrative entries returned, I personally think this is fine with the
exception of a large amount of distant or low population admin entries that
appear before your first word is completely formed.

When you reach 8 characters, eg. "hackney c" you seem to get an odd mix of
admin entries and places, this is also where some entries from over 6000km
appear in the results.

When you type "hackney ci" you don't get any results at all? I find this
behaviour very odd and it seems like a bug?

After typing "hackney cit" or more characters the correct local POI is
returned.

note, to reproduce the above query bias:
https://mapzen.com/pelias#loc=18,51.54503,-0.05639

—
Reply to this email directly or view it on GitHub
#69.

[dianashk]

Waiting on ES 2.0.0

[missinglink]

this issue was resolved by the ngrams release