[uf-discuss] [pf] picoformats and extracting event data from text
email
Chris Messina
chris.messina at gmail.com
Sun Jul 30 21:28:24 PDT 2006
Hi Michael,
Thanks for your thoughts -- please note that messages pertaining to
picoformats should be prefixed with [pf].
As for your comments, this kind of system is already implemented in a
number of places, such as Skobee and Backpack. Skobee's syntax can be
found here: http://skobee.com/email_howto
What I think this is getting at is a system for streamlining natural
language processing (NLP) by using conventions for certain types of
information. Basically, this is what YAML does.
That said, there's some value in creating some consistency in what
"tags" people can use to perform certain actions.
While the thrust of this idea is a bit off from the original use case
for picoformats, I appreciate you bringing it up and suggest that it
be added to the wiki under picoformats-brainstorming
(http://microformats.org/wiki/picoformats-brainstorming).
Thanks again,
Chris
On 7/30/06, Michael MD <mdagn at spraci.com> wrote:
> I'm thinking about easy (for the user) ways an event promoter could add
> machine-readable data to their emails for adding events to spraci.com
>
> I get a lot of emails about upcoming events but there is no time to do data
> entry
> (I say to them the correct way to add events to spraci.com is to use the
> forms on the website or to provide a data feed).
> However, it would be good if at least some of the stuff sent by email could
> be included in the listings (as that could be a significant amount of extra
> data).
>
> Obviously for html emails they could use hcalendar, but for plain text
> emails and for users who are not familiar with html,xml, etc I'm thinking an
> easy way for them might be just to include something like this in their
> email.
>
> Name: (event name)
> Date: event dates (fill date or start-end including the year)
> Location: (including city and country)
> Description: (short description, lineup, etc)
> Categories: (comma separated list of tags for things such as event-type,
> music, genres, etc)
>
> This could be added after each event blurb with at least one blank
> line separating it from any other text.
>
> The important thing here is it has to be easy for any event promoter to do
> without too much thought and must not need any special authoring tools. (so
> the names must be obvious - this could get complicated if support for other
> languages is added so initially I guess its just English)
>
> If they omit the location field I could also make it check for '@' and use
> anything between that and a new line.
>
> Of course the names would have to be loose because it has to be very easy
> for people to do without having to look them up. (otherwise I would probably
> think of using the iCal/hCal names - A parser could accept both).
> For date formats I think it would have to accept most common formats, such
> as those accepted by commonly used perl modules or strtotime in php.
> (with a warning to people not to use dd/mm/yyyy or mm/dd/yyyy as those are
> ambiguous). It will probably be hard to explain to event promoters why the
> year is required (they seem to be so used to leaving it off), I will
> probably have to provide them with a list of accepted formats for dates.
> I don't really see any easy way around that.
>
> For the city/country a parser could check words against a list of known
> cities/countries. If both a city and country (or state) are found it could
> reduce the possibility of an incorrect match. (of course they will have to
> be careful to spell them correctly and I probably would have to make it
> somehow check for common alternate spellings of names and common
> abbreviations for state names). If there are no matches it probably should
> default to the area/city associated with the user.
>
> This is similar to something I had for processing incoming events from a
> certain contributor back in the mid 90s.
> I needed something like this because the data from that person was
> basically
> just plain text that was sent to me by email and I wanted to reduce the
> amount of manual retyping.
>
> I think this idea is closely related to picoformats so I'll be watching that
> closely!
>
> I'm thinking about this (yet again) for incoming emails and possibly also
> future mobile applications where people are manually typing text as a single
> blob.
>
>
>
>
>
>
> _______________________________________________
> microformats-discuss mailing list
> microformats-discuss at microformats.org
> http://microformats.org/mailman/listinfo/microformats-discuss
>
--
Chris Messina
Agent Provocateur, Citizen Agency &
Open Source Ambassador-at-Large
Work: http://citizenagency.com
Blog: http://factoryjoe.com/blog
Cell: 412 225-1051
Skype: factoryjoe
This email is: [ ] bloggable [X] ask first [ ] private
More information about the microformats-discuss
mailing list