Organizing Tweetdeck Columns during Weather Events
Articles Blog

Organizing Tweetdeck Columns during Weather Events


Hello, I’m Trevor Boucher from WFO Austin/San Antonio. I’m the social media program lead here at the office, and as part of
the series on social media data mining I’m going to be covering TweetDeck and
how to use it as a situational awareness tool for data mining. Now you may already
have TweetDeck and use it regularly in your office, but just want to give an
example of how we do it at our office that helps us for data mining both
during routine and non-routine weather events. Now I’m not going to talk a whole
lot about the features that TweetDeck has because you’ve already heard most of
that in other lessons prior to this, but this is more of a methodology of how to
maybe set this up in a way that you may or may not have already considered in
the past. So a couple of things that you probably have seen before. We have
multiple columns and you probably do, too. We have about 20 of these here.
But starting from the left, everybody probably has their own content as a list,
or a column rather, that they’re monitoring so that you know the
rest of the office knows what’s been tweeted, what questions
have been answered like this here, or anything like that. So obviously, you need
to have that, probably, at a base minimum. But over here on the next column is
another one that probably everybody has. Notifications – what’s been tweeted to you –
keep track of questions, comments, things like that. We don’t have every
notification type available turned on. Obviously, we don’t want to be notified
for every little thing that happens – every little like that comes our way. But
we do have the two that would potentially be asking us a question,
mentions and quoted tweets, there. And then we also have sounds turned on for
notifications so that the office can be aware of anything that comes our way.
So those are pretty standard, probably, for all offices. Now beginning on column
three here, we have a special hashtag for our CWA for submitting reports. Now,
this is shared through all of our spotter trainings and things like that,
so this is fairly used heavily in our CWA. So during severe weather scenarios
we will get EWX spotter reports. So it’s very important to have this
sort of front and center for monitoring when it comes to, you know, getting
reports and in getting quality LSRs. Now starting over here on row four
through six, we have twitter lists. Now I won’t talk about creating a list because
you’ve probably seen that in a previous lesson. But we have our
Twitter list set up here a little bit differently maybe than some others. Where you can break a list down into maybe emergency managers or broadcasters
or something like that, we’ve broken ours down into areas. So this actually mimics the
Austin viewing area, the San Antonio broadcast viewing area, and then the Rio
Grande Valley which doesn’t have a viewing area but they this is all the
counties that are outside of those two on the right. And also, its more
rural than those locations. Not as much population. So we grouped all
those together as well. So pay attention to the titles here Rio Grande Valley,
Austin Hill Country, in San Antonio/ Victoria, because that is how we’ve
titled these columns. So how we broke these lists down is that instead of
it being, say, emergency managers, something like that, we’ve taken anything and
anyone that would potentially be tweeting about the weather in these
respective locations. So we have media here. We have San Antonio Fire,
so fire departments. We have newspapers. We have utility companies, schools, even
city organizations and city governments that we have followed in all the
counties that were color-coded in the previous picture you’ve seen. So that’s
how we’ve broken down the lists and that’s why we have these based off areas.
So if we have a storm that is moving into the Austin area we can take a look
at this list here, and then and try to find reports that we’re looking for. If
it’s in the San Antonio area we can go into the content look for, say, any
tweets that might be matching, say, rain. And so we’ll pull out a tweet that
mentions the word rain in it, for instance. So that’s how we would extract information
from our various lists. Now our lists are pretty substantial in our first two. Austin,
San Antonio and – For Austin and San Antonio, rather. You’ll see
almost two hundred members in that one. The Rio Grande
doesn’t have nearly as many because it’s not as much. But we do also use this one
as well. As you can see, we even have some information from what’s going on over
the border. So this can be really useful for breaking this into the different
regions for obvious reasons. And then if you go a little bit farther to the right
here you’ll see that we have our virtual operations support teams (VOST). So we have the
Travis and Bayer counties that have community public service groups that are
funneling reports and actually doing some tweeting themselves. So we have Bexar County severe weather and Travis County severe weather that we are
monitoring. So it’s important for us to keep track of them. And then also their
public accounts are there – I’m sorry, rather, their personal accounts –
we keep a track of as well and make sure we know what they’re tweeting and let
them know if there’s anything we need to address or anything like that. Over here
in column ten, we have popular hashtags. Now, every state has, as I’m
sure you know, every state has a hashtag. Usually hashtag whatever your state is
WX. So we’re monitoring that [#txwx], but we have a number of others that we’ve come, over
the years, to find out that are somewhat helpful. Like #atxwx. A lot of our
broadcasters use that. #satwx in San Antonio. Texas fire [#txfire] is often used when we
have fire situations going on. So we follow those hashtags as well as another
means of following along. But then as we go farther to the right we have these
columns that are entitled #atxwx and #satwx and whatnot here. They’re
titled that way on purpose because they are geocoded searches. Now in an earlier
lesson you had learned how to do a geocoded search. But what we did is we took
that and we broke our whole CWA into these various regions here. We have six
different regions. Some smaller, some medium, and one large, where we
basically have 99% of our CWA covered in these geocoded search radiuses. Now
this only pulls about 25% of the amount of tweet activity from these areas, but
that these are all with geocoding turned on. Now this is helpful
because this really helps us pin down where the information is
coming from. If we find something in the Kerrville column, for instance, then we
know that it’s coming from this location here, and if there’s any storms in this
location then we can keep track of the reports are coming from that
location. So we have these broken down here. And in addition to that we are
looking at keywords. So the operator ‘OR’ allows us to look for these terms, “OR
wind OR windy” or any tweet that has any of these keywords in them. And this
geocode syntax here, as you’ve learned in a previous lesson, this is how we do
the actual geocoded searching. So we’re constantly looking for these keywords,
and you can add as many or as few as you’d like. But we now can have a
constant search all the time. Now most of this content will not be useful. It’ll
just be random tweets that, for instance, if they use the word fire somebody
might be saying, like it’s, you know, “Just made a fire in my house,” for instance, and
that’s not something we need to know. But occasionally you might end up having
some good nuggets of information being extracted from here. Now there’s
some caveats by doing these type of searches like this. One of which is that
you’ll sometimes find obscene pictures. You might see some profanity. This
is just Twitter in its rawest form, so keep that in mind when you’re doing these
searches. But many times you’ll end up finding some very useful
information. In fact, just glancing at this on these columns I’m actually
seeing @KENS5 talking about a house fire. Now if this was like a wildfire
this might be really – something that we just pulled up right off the bat. So this
can be really useful. We often will drag the bar over to this area when we can’t
seem to find anything from the official sources and we need to be diving into
the general public and seeing their tweet activity. So that’s why they’re
broken up into these various columns here, as again they are broken into
different tweet searches for our whole CWA. Now the very last two columns
here is scheduled tweets and then messages, direct messages sent directly
to us. So those are far less used, so that’s why they’re on the far right here.
But anytime that we don’t get the content we’re looking for in
our lists that we have set up, which are usually a majority of what we get it
from over here in these four columns, we can’t seem to find anything,
we’ll come over here and take a look at what the general public is tweeting in
and can really pull out some good nuggets for that. So if you have any – This
is just one example of one way to do it here in our area.
You can do it however we’d like to. If you have for instance DSS that’s going
on and you’d like to monitor a large outdoor event – in our area we have South
by Southwest that’s #sxsw – and we’ll add a column for that.
(And you’ve already learned how to do added columns and doing searches and
things like that.) So this is how we would set it up for the majority of our
situations. And we can always add more columns if we need to for different
events. So, that’s pretty much all I have for this topic. I hope you check out some
of our other data mining lessons in this series, and lots of really good content
from various program leads throughout the country here. So if
you have any questions, feel free to ask. Otherwise, happy data mining!

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top