My Photo

« Search User Interfaces: Marti Hearst | Main | Steve Irwin, Michael Jackson »

June 26, 2009


Jan Berkel

I completely agree. It's not too hard to improve the results by using the frequency of a trend to filter out noise. #iranelection is trending almost every day, which makes it completely uninteresting, but it was interesting when it trended for the first time. Then there are trends like Spain or Brazil which trend regularly (but not daily), usually when a football match is on, which might be of interest.

I think the most interesting trends are the ones which have never trended before (= "breaking news" on twitter).

It also shouldn't be too hard to determine that all the RIP, MJ, Michael Jackson trends are basically the same by using some statistical analysis on the content.


Also, once a topic has been on the "trending" list for more than an hour or so, the signal-to-noise ratio in the search results for it plummets as people tweet things like:

1. "Huh? What is #[topic name]? Why is it trending?"

2. "#[topic 1] #[topic 2] #[topic 3] please follow me follow me!"

3. "We did it, guys! #[topic] is now a trending topic!!!!!!"

A lot of this could be solved by having the trending topics on a separate page.


My above point made better with a graphic:

Mark Roddy

I totally agree with the sentiment that they're doing poorly in distinguishing between topics and phrases. The same thing happened when the Iran election started trending (multiple 'topics' on iran all refering to the same thing).

The comments to this entry are closed.

Twitter Updates

    follow me on Twitter

    March 2016

    Sun Mon Tue Wed Thu Fri Sat
        1 2 3 4 5
    6 7 8 9 10 11 12
    13 14 15 16 17 18 19
    20 21 22 23 24 25 26
    27 28 29 30 31    


    Blog powered by Typepad