Tweak the Tweet for Chile

[Sophia Liu]

Thank you for hosting the CrisisCommons conference call today. I know it was a small group but it was really helpful to understand what has been happening in the background for Chile. Please send out information about the conference call tomorrow at 1 pm.

Here is an update on what we are doing with Tweak the Tweet for Chile.

Tweak the Tweet:

Our Project EPIC team is helping to deploy and distribute the Tweak the Tweet syntax or format specific to #chile. We have gotten volunteers to help us translate our Tweak the Tweet prescriptive examples and instructions to spanish and chilean (or rather spanish slang). 

This is a Google Doc with the prescriptive tweets in spanish and chilean: http://tinyurl.com/Chile-TTT

This is a Google Doc with the TtT Instructions (we need more translation help on this): http://tinyurl.com/TTT-Chile

We created a TtT Editor in english and spanish to help people create tweets with the TtT syntax: http://tinyurl.com/TTT-editor

Kate Starbird is updating this Google Spreadsheet with tweets using the TtT syntax: http://tinyurl.com/ttt-sebusca

We are currently trying to push the syntax to our Chilean contacts via Twitter. On the phone, I think Noel suggested that we have Tweets using the TtT syntax be put into Chile.Ushahidi so we will contact Ushahidi about this. Simon has created a website showing some tweets using the TtT syntax but they are not filtered entirely for TtT here: http://chile.tweetneed.org/

We would like to have more tools up that will use Tweets in the TtT syntax so that people can see examples of its usefulness.

[Twigger, Simon (MCW)]

Hi Sophia,

Thanks for the update on what's going on. If there is a call today can you let me know and I will try and join you all - you mentioned 1pm, what time zone is that? I have made some more progress on my end, Im trying to incorporate an additional database structure under the hood using the AllegroGraph triple store, this will allow me to do something locally with the ontologies and RDF I'm generating which hopefully will result in making some more useful data available to people.

You mentioned that chile.tweetneed.org isn't filtered entirely for TtT - Would it be helpful to have a way to view 'official' TtT tweets, ones that follow the syntax more closely? I think I am grabbing a superset of these tweets so if you can define how you want these filtered I can most likely set that up. 

For the #sebusca tweets and pushing to Ushahidi, I've been improving the 'name recognition' parsing code so its doing a better job of finding names in #sebusca tagged tweets. Im still working on this code (but I could keep tweaking it forever) would it be helpful to make these tweets available in some reasonable format (online, downloadable CSV, etc?) so you can add them into the google spreadsheet more easily? I'd like to take the growing number of #sebusca_* tweets (eg #sebusca_concepcion) that have a city name at the end and use those too, adding in the location info from the hashtag into the database.

Ushahidi has a great API so if we can figure out what tweets we might like to push over to their chile site (the 'official' TtT tweets mentioned above, or something similar?) then one option would be for me to add that code in to chile.tweetneed.org so we can get the tweets heading over there.

Simon.

--

Simon N. Twigger, Ph.D.

Assistant Professor, Department of Physiology

Medical College of Wisconsin

8701 Watertown Plank Road, 

Milwaukee, WI, USA

tel: 414-456-8802

fax: 414-456-6516

AIM/iChat: simontatmcw

[Catharine E. Starbird]

Simon,

There seem to be two big sets of TtT data - one where people have adopted the syntax, and one where they are naturally using something close to it. Both could be extremely valuable. I've been picking them up and storing them in different sets. For the strict #TtT syntax, I have code that parses all of them into records, removes duplicates, and throws out a spreadsheet file: http://tinyurl.com/ttt-sebusca. I run the scripts to generate a new spreadsheet every few hours - and manually update/upload the spreadsheet file. This is a patch solution - ideally we would get this information to a live webpage - or get it integrated into other sources (like People Finder / Ushahidi).

To pick strict #TtT up, I've had to go to the presence of third tag (#sitio, #contacto, or #info a this point is all I'm checking for). That's provided a pretty noiseless sample. I also had to adjust my location parsing to accept the next word, even if it's tagged - because so many are (smartly) tagging the city in their tweets.

As for the non-TtT tweets, I tried to write some parsing stuff to pick up the names, locations and contact info. I had a hard time removing duplicate records, because there were so many variations in the ways that names were presented. And then I ran out of cycles. My to do list has on it to do something with those #sebusca_ location tweets. But I won't get to it. From what I can tell, a decent parser could turn about 80% of those messages into records of some kind. They won't be super clean, but they'll be effective.

It would be great if you could pump over both types of tweets in a record format that Ushahidi can/will accept - both strict #TtT tweets and less strict ones. I find splitting them up makes for one very pretty set and one very-mildly noisy set.

And we are still pushing out the syntax. The initial push for #chile was pretty significant - now it goes out in waves. The best chance for deployment is to find an official information source to carry the prescriptive tweets on their Twitter account. We're still brainstorming ways of getting that done - though we have seen smaller sources in Chile helping to distribute.

The chile.tweetneed.org site looks good.

I hope you had an excellent conference last week.

Kate

Kate Starbird

University of Colorado

ATLAS PhD Student

catharine...@colorado.edu