Automatically label topics of legislation

[fgregg]

Much of the legislation is routine and boiler plate. We should be able to learn and label legislation that are just "congratulations", "zoning changes", "signage", etc.

Ideally we could find the major routine types, so we could then identify and highlight non routine legislation.

[danxoneil]

Would love this feature, mainly because it allows for the surfacing of more important items. It's difficult to parse all of the honorary baloney to get to the real stuff!

[fgregg]

14:04 < fgregg_> Derek had the good idea of having the scraper be responsible 
                 for making the label, but I was wondering if you
14:04 < fgregg_> had thought about how councilmatic might do well to handle 
                 these labels
14:05 < fgregg_> maybe this is a tag field?
...
14:17 < mjumbewu> fgregg_, that'd be awesome. my first thought is that it'd be 
                  attached to the metadata objects.
14:17 < mjumbewu> each legfile has a metadata attached to it
14:18 < mjumbewu> and things like locations and references are attached to that
14:18 < mjumbewu> tags/topics could be another model that gets attached
14:18 < fgregg_> that makes sense.
14:21 < fgregg_> what do you think the model might look like
14:21 < fgregg_> an enumerated list
14:21 < fgregg_> or a full on separate table?
14:23 < mjumbewu> lemme see
14:29 < mjumbewu> yup, a new model is the way i'd go. like MetaData_Tag. And it 
                  would look a lot like MetaData_Location, except without a 
                  location

[mjumbewu]

@fgregg heh, yeah, that.

[derekeder]

Logic is in place to handle this:

• MetaData_topics tables
• displaying topic labels for each item
• topics are clickable and searchable
• topic classifier function defined in local_settings.py

Todo:

• finish topic classifier for all common cases

[fgregg]

Can we search for legislation that does not have a routine tag?

[derekeder]

this is being handled right now with a non-routine tag