Tracking Location in Public Social Media Posts

[Policy Wonk]

I'm working on a project that among other things seeks to document that there are signs of human life in and around a location and immediate area on the periphery of downtown.

Something we would like to capture is social media posts the originate in that area, we have someone searching public twitter and instagram posts, but it is extremely labourious. An hour's work might yield a dozen relevant posts, most of which are pictures of food.

We talked to a "social media expert" and the only positive thing to come out of the meeting was none of us stabbed her within the first ten minutes. She completely didn't comprehend that we're trying to do, but spewed more jargon in a short time than I had ever heard before in my life.

Does anyone know of a way to relatively automate this, there are various websites that search by location. But are either latching on to places identified by google, or facebook, or whatever. Many of which are just wrong, rather than a geographic or neighbourhood location that is none-the-less embedded in many of the posts.

The ultimate goal is to say look at all these cool hipsters who already frequent the area.

Does anyone have any idea how to tackle this without human eyes glazing over at endless pictures of pilaf?

[Fuzz]

Put the most hipstery thing possible in the location, put a twitter tag on it and some text to the effect of "tweet this totally radical doohicky to show how unique a flower you are" then follow the tweets... I dunno, I don't even have a cell phone, but I imagine that might work. People are narcisists, play to that.

I would think any technical solution would require scraping all sorts of networks. You could also setup a motion sensing camera to snap a pic of everyone that walks by.

[Policy Wonk]

The end goal, if this actually works is to make a presentation board like the displays at the Vancouver airport that has tonnes of people's tweets about travel and YVR.

It's as much a theoretical exercise to see how much useful data we can pull out of thin air. But I feel like a complete creep at the same time since we can predict within five minutes when somebody with a specific Samsung cell phone is going to walk by in the morning and afternoon. We can even extrapolate which bus they take.

[Fuzz]

It's amazing how little privacy people actually have, and that so many of them just give it away. Knowingly or unknowingly.

[sim]

Originally Posted by Policy Wonk

I'm working on a project that among other things seeks to document that there are signs of human life in and around a location and immediate area on the periphery of downtown.

Something we would like to capture is social media posts the originate in that area, we have someone searching public twitter and instagram posts, but it is extremely labourious. An hour's work might yield a dozen relevant posts, most of which are pictures of food.

We talked to a "social media expert" and the only positive thing to come out of the meeting was none of us stabbed her within the first ten minutes. She completely didn't comprehend that we're trying to do, but spewed more jargon in a short time than I had ever heard before in my life.

Does anyone know of a way to relatively automate this, there are various websites that search by location. But are either latching on to places identified by google, or facebook, or whatever. Many of which are just wrong, rather than a geographic or neighbourhood location that is none-the-less embedded in many of the posts.

The ultimate goal is to say look at all these cool hipsters who already frequent the area.

Does anyone have any idea how to tackle this without human eyes glazing over at endless pictures of pilaf?

http://nesit.org/how-to-search-twitt...eocode-or-gps/

I tried it and it seems like it works. You could probably take the url and with a bit of programming skill just ping the relevant Twitter API every so often - it will have some query limits. Not sure what the data type is, because I'm not really a programmer, but using something like R, you can parse most data structures and kinda do with it what you want.

[Policy Wonk]

Thanks, I think that is one of the techniques she is using for the manual searches.

apparently we're getting a demo of something called hootsuite next week that can automate the monitoring.