Wednesday, May 24, 2017

SearchResearch Challenge (5/24/17): Finding tweets from a particular place


When big events happen, people tweet about it.  

Big events can be pure fun (like Red Bull's Flugtag "flying" event), they can be tragic (such as the recent terrorist bombing in Manchester, England), and they be technological (Google's I/O developer conference last week).  

But when big events happen, people share on social media--Snapchat, Facebook, Twitter, etc.  

Often the real-time event is tracked on Twitter.  So it's occasionally useful to use Twitter to figure out what's going on.  


For example:  Is that my friend's house on fire??? 


Luckily, not my friend's house.

I recently used my ability to search the tweet-stream to find out why there was a huge plume of smoke from a house fire near my home.  In just about 1 minute, I went from finding the tweets in my area, searching them for mentions of "fire" and then discovering that a local television station was showing real-time video.  That was a fantastic search experience because I learned from the live-stream that it was next door to my friend's house, and not actually their house that was ablaze!  (They did suffer a little heat damage, but the fire department did a great job.)  

This makes me think that this is a great SearchResearch skill to know.  So it prompted today's Challenge.  See if you can figure them out!  


1.  When the Google I/O event happened last week, I wanted to see what kinds of things were being tweeted about.  Unfortunately, not everyone adds the #GoogleIO hash tag to their tweet, and sometimes people add the hashtag when they weren't really there.  Can you find tweets that were posted from INSIDE of Google I/O 2017?  
2.  Obviously, you'd like to be able to restrict your tweet lookup by time and date.  Can you find those INSIDE tweets from Google I/O that were posted only during the days of the event?  (May 17 - 19, 2017)  
3.  Since you can find the location of a tweet, is it possible to make a map of tweets that are posted from the city of San Francisco during a single day? How would you do this?  (I'll post the best maps in the blog next week.)  
4.  For completeness, it's useful to know how many tweets come with a geocode. Can you estimate the fraction of tweets that are geocoded?  (I ask because you need to know what fraction of tweets you're NOT seeing when you do a search for tweets posted from a particular location.  If you're only picking up 5%, then that's a very different story than if you're seeing 95% of all tweets.)  

Be sure to let us know how you did this!  

Search on!  

8 comments:

  1. Hello Dr. Russell and everyone

    1. When the Google I/O event happened last week, I wanted to see what kinds of things were being tweeted about. Unfortunately, not everyone adds the #GoogleIO hash tag to their tweet, and sometimes people add the hashtag when they weren't really there. Can you find tweets that were posted from INSIDE of Google I/O 2017?

    [find tweet with location] also searching with tools past year

    Search Advanced Twitter

    From Quora question followed this link:
    The one million tweet map

    And

    Twitter Location Search: A complete guide

    For Q1 (Inside) Q2 (date) and Q3 (San Francisco)

    Advanced search gives options to dates so that is easy.

    Searched [site of Google I/O 2017]: Shoreline Amphitheatre Mountain View, California. To get the coordinates in Google Maps (37.426475, -122.081221)

    With those data went to Advanced Twitter Search and got this :” near:"37.426475, -122.081221" within:1mi since:2017-05-17 until:2017-05-18 ) I changed 15 for just 1 mile. “ Then, returned to advanced and added #GoogleIO And I got this: “ #GoogleIO near:"37.426475, -122.081221" within:1mi since:2017-05-17 until:2017-05-18 “

    I think if we remove within we have inside tweets and for San Francisco, still need to find if we need a distance or if it is possible to add a city

    That gave me tweets but not numbers of how many tweets. So

    [number of tweets with certain query]

    The Search API by Twitter About Geolocalization mentions: “ When conducting geo searches, the search API will first attempt to find Tweets which have lat/long within the queried geocode, and in case of not having success, it will attempt to find Tweets created by users whose profile location can be reverse geocoded into a lat/long within the queried geocode, meaning that is possible to receive Tweets which do not include lat/long information.”

    [count twitter advanced search]

    Is there an easy way to count the number of tweets posted with a specific hashtag? Tried with Keyhole and others but others need account or pay

    ReplyDelete
    Replies
    1. I tried again

      MapD with #GoogleIO I think I still don't understand because it only shows 772 tweets and only 19 from San Francisco. At this time #EuropaLeague (Ajax Vs Manchester United) has according to Twitter 11.2 k tweets. MapD says only 68. Choropleth looks much better to understand.

      [tweetmap numbers meaning] Found by luck possible answer to Q4: Can you estimate the fraction of tweets that are geocoded? ( Mar 08, 2016) from November 2014 to February 2015. Geocoded tweets make up approximately 1-2% of the entire Twitter feed...so this demo showcases less than 0.2% of all tweets sent during the time period

      Once again, I am sure the numbers in the Map express another thing not the actual, total number of tweets made

      Delete
    2. Searching in Spanish

      [como hacer mapa tuits]

      Crean un mapa de Twitter en tiempo real (tweetping) (
      MapBox: un espectacular mapa de tweets

      Some of possible tools are not longer available or require signing. Btw, how we can know when is safe to sign those extensions? Therefore, nothing new to add to find the best solution to this SRS

      Delete
  2. Quick and very incomplete answer.

    1. I happen to receive regular emails with tweets from a specific area. I use this IFTTT recipe/applet by a police officer (with stalking abilities…) named kev: Get an email digest of Tweets from a specific area. After finding where Google I/O 2017 took place (the Shoreline Amphitheater at 1 Amphitheatre Pkwy, Mountain View, CA 94043, USA) you may use that address with zoom 16 or 17 to guarantee that tweets are only from people actually there. This solution is good for daily tweets of last day events, not for last week's tweets.

    2. Advanced Search within Twitter allows searching for tweets near a location and from a specified date range, but only if: a) you allow geolocation yourself; b) in general areas, like Mountain, CA. So this is not exactly what is asked on number 2.

    4. If you google search [ fraction of tweets that are geocoded ], one of the search suggestions is [ percent of tweets with geolocation ]. Interesting results there. Really, "less than 2% of tweets are geotagged" (Laylavi et alii, 2016), but there's a lot of articles (on Scholar and off Scholar) about geolocation prediction / inference, claiming that 80% of twitter users can be located (not particular tweets though, if I "read" the articles correctly).

    ReplyDelete
  3. Quick and very incomplete answer.

    I happen to receive regular emails with tweets from a specific area. I use this IFTTT recipe/applet by a police officer (with stalking abilities…) named kev: Get an email digest of Tweets from a specific area. After finding where Google I/O 2017 took place (the Shoreline Amphitheater at 1 Amphitheatre Pkwy, Mountain View, CA 94043, USA) you may use that address with zoom 16 or 17 to guarantee that tweets are only from people actually there. This solution is good for daily tweets of last day events, not for last week's tweets.

    Advanced Search within Twitter allows searching for tweets near a location and from a specified date range, but only if: a) you allow geolocation yourself; b) in general areas, like Mountain, CA. So this is not exactly what is asked on number 2.

    If you google search [ fraction of tweets that are geocoded ], one of the search suggestions is [ percent of tweets with geolocation ]. Interesting results there. Really, "less than 2% of tweets are geotagged" (Laylavi et alii, 2016), but there's a lot of articles (on Scholar and off Scholar) about geolocation prediction / inference, claiming that 80% of twitter users can be located (not particular tweets though, if I "read" the articles correctly).


    Quick and very incomplete answer.

    1. I happen to receive regular emails with tweets from a specific area. I use this IFTTT recipe/applet by a police officer (with stalking abilities…) named kev: Get an email digest of Tweets from a specific area. After finding where Google I/O 2017 took place (the Shoreline Amphitheater at 1 Amphitheatre Pkwy, Mountain View, CA 94043, USA) you may use that address with zoom 16 or 17 to guarantee that tweets are only from people actually there. This solution is good for daily tweets of last day events, not for last week's tweets.

    2. Advanced Search within Twitter allows searching for tweets near a location and from a specified date range, but only if: a) you allow geolocation yourself; b) in general areas, like Mountain, CA. So this is not exactly what is asked on number 2.

    4. If you google search [ fraction of tweets that are geocoded ], one of the search suggestions is [ percent of tweets with geolocation ]. Interesting results there. Really, "less than 2% of tweets are geotagged" (Laylavi et alii, 2016), but there's a lot of articles (on Scholar and off Scholar) about geolocation prediction / inference, claiming that 80% of twitter users can be located (not particular tweets though, if I "read" the articles correctly).

    ReplyDelete
  4. Questions 1 & 2)- Had no idea there was an advanced search feature on twitter. Did a search on how to find tweets from a specific venue and learned that you can narrow down search results in twitter by both location and by date range. So by putting in mountain view, CA into the location it narrowed down results to that location- although I suppose that someone could have been in that location but not at the conference and still tweeting about the conference.
    For question 4- did a search for how many tweets are geocoded and came up with about 1%. One result was from Quora where someone who did research on this posted their result. Will try to find more results which are more definitive. Did find this article which says researchers from IBM figured out a way to discover a persons location from their tweets without having geotagging set - https://arstechnica.com/business/2014/03/lack-of-twitter-geotags-cant-stop-researchers-from-getting-location/ - Well did some more searching and the name Leetaru came up as having done some research on tweets using geotagging. So searched for Leetaru twitter geocode and got this result- http://firstmonday.org/article/view/4366/3654 in this paper he concludes that just over 3% of all tweets are geocoded. So whether it's 1 or 3% just a small number.
    Will work on map later. This was a very interesting challenge most esp. because I use twitter but never realized there was an advanced feature that allowed one to narrow tweet searches.

    ReplyDelete
  5. 1. One way to do this is to use geolocation using the Advance Search on Twitter. To do this, you need to search the coordinates of the Shoreline Ampitheater, where the event was held, along with a radius that covers the area of the amphitheater. Then you filter the results by date range, according to when the Google I/O event happened. To figure this out, I searched something like [geotagging tweets] and found a how-to article that explains the method I described: http://www.hrmarketer.com/blog/2016/03/how-to-search-tweets-by-location-twittertips/

    2. Just combine the above method with the advanced search feature on Twitter. I figured this out just through prior knowledge of advanced search.

    3. It appears this widget may be able to do the trick, https://www.tweettomap.com/. I found this by looking through a forum post I found after googling [mapped geolocated tweets free].

    4. Apparently, roughly 20% of all tweets are geocoded. This I gathered from a study from USC: https://pressroom.usc.edu/twitter-and-privacy-nearly-one-in-five-tweets-divulge-user-location-through-geotagging-or-metadata/ This article was found on the USC website through the search [geocoded tweets percentage].

    ReplyDelete