Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

input text interpretation #48

Closed
missinglink opened this issue Dec 2, 2014 · 5 comments
Closed

input text interpretation #48

missinglink opened this issue Dec 2, 2014 · 5 comments

Comments

@missinglink
Copy link
Member

When entering location strings where the text is delimited by a symbol, eg. 'London, UK' or 'London, ON' the context suggester fails to see the difference between the named component and the admin component(s) which are conventionally delimited using a comma.

We need to start looking at some very basic NLP even if it's just something like only using the tokens before the first comma.

Here's a visual illustration of the problem:
ss

The gifs shows 2 UX issues:

  1. Not reproducible: when clicking a successful match you cannot use the string returned to search and find that same place, you must remove everything after the first comma.
  2. You cannot paste an address string directly in to the input, eg. pasting "Hackney Town Hall, Hackney, Greater London" yields 0 results.

thoughts @hkrishna ?

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

@missinglink
Copy link
Member Author

Since we are now fully converted to suggest_plus_mget approach, we could do something fancy with the tokens after the comma.

When using /search, this could be pretty easy, simply querying against the admin fields with the remaining text and boosting the results accordingly.

Doing it in /suggest is another story completely...

I'm planning on allowing the suggester FST to be filtered by alpha3, which could at very least support querying alpha3 values after the comma, still that's not very user-friendly.

In the future we could also have context suggester entries for common abbreviations.
ref: pelias/schema#25

@hkrishna
Copy link
Contributor

How about we use the context suggested and instead of the category be one
of the datasets, could we have dataset and alpha3? For ex, geoname:US, locality:IT

And in the query you could maybe use wildcard character * to match on
things??

@dianashk dianashk added this to the Pelias v1.0.0 milestone May 4, 2015
@dianashk
Copy link
Contributor

dianashk commented May 6, 2015

Should be closed in light of newer issues.

@missinglink
Copy link
Member Author

We could close this in favour of:

That ticket only covers one of the issues and only partially, plus it doesn't have as clear a description as this.

@missinglink
Copy link
Member Author

this issue has been resolved

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants