search

For the last couple of years, one of the most important columns in my TweetDeck setup was the one tasked with presenting tweets that mention Refresh Austin. The search itself is pretty straightforward, though it includes several variants to account for all of the possible ways that people might reference our group: refreshaustin OR austinrefresh OR “refresh austin” OR “austin refresh” OR @refreshaustin. This worked beautifully for a long time, but a little while back (I don’t know when exactly) I noticed that the feed included many tweets that have nothing to do with our group. A significant portion of these are written in languages other than English, so it’s been hard to detect a pattern.

Until today.

The photo shows five recent tweets, three of which do not apply to our group at all. The key detail is that all three link to bit.ly/ra, which I set up a long time back to point to the Refresh Austin site. But in these cases, the problem appears to be due to the fact that a longer URL, which starts with ‘ra’ was cut off when the users retweeted or simply posted a tweet greater than 140 characters.

Twitter’s search now follows links within tweets to determine that they are both valid (non-spam, no malware) and also to provide additional context. So, in this case as bit.ly/ra resolves to RefreshAustin.org, the tweets appear in my search feed even though they have absolutely nothing to do with us.

So now I understand why this happens, but I do not have a solution for it. While I do not want to filter out tweets that contain the bit.ly/ra link as it is a valid link, I’d love to reduce the overall noise. This is a bit frustrating, but seems solvable with a bit of time and effort. Should I figure something out, I’ll post it.