• Blog
  • Viralheat’s Twitter sentiment analysis plug-in for Chrome: my feedback

  • Viralheat’s Twitter sentiment analysis plug-in for Chrome: my feedback

    Viralheat’s Twitter sentiment analysis plug-in for Chrome: my feedback

    Automated sentiment analysis is one of the most criticized features within Social media analytics / monitoring family.

    From my experience I have an idea that out of all comments generated by any tool, be it Radian6, Alterian SM2, or any other tool, there is an accuracy gap. I believe that the inaccuracy of measuring sentiment by any tool is about 70%. Most of these gaps occur when the sentiment is marked neutral.  Another reason of gap is these tools are not able to identify local slangs, words or phrases translated or written in English, sarcasm and mood of author.

    When I recently read the news that Viralheat, a social media monitoring tool, has released free Chrome plug-in that adds sentiment analysis to Twitter timeline, I decided to give it a shot.

    About Viralheat Sentiment Analysis Plug-in for Twitter:

    Viralheat’s this new plugin can be found easily on Chrome web store and it takes couple of minutes to install. Once the plug-in is installed, and you open your Twitter page on Chrome, the plug-in adds a bar on top of Twitter page and displays aggregated sentiment for displayed tweets on that page. When user searches for any brand i.e. ‘iPhone’, it will provide sentiment for all relevant tweets.

    What’s New and Good?

    My first thought was that the tool is pretty handy if I am on my Twitter page. It lets me see the sentiment of the post and get overall idea of what’s happening around the brand, product or service I search for. One other feature I found interesting was that it lets you change the sentiment on Twitter page itself.

    This tool can prove handy to see the sentiment around a live event or a hashtag (#).

    What’s Bad?

    I thought to run a small test on the accuracy of the tool. I took a sample of 50 tweets and compared the sentiment provided by the tool with manually analysed sentiments for same tweets.

    I noticed, the tool analysed multiple posts on same topic with different sentiments. For example, a news article which was shared by many users, “Samsung Galaxy SII sales hit 10 Million, excluding United States” was analysed incorrectly 85% times. The overall accuracy was observed to 30%.

    My Rating


    Disclaimer: Views of authors are personal and do not represent the views of Blogworks, or any of its clients.

  • 5782060475547725833610559520754756486758466107716472447574868586361277314725475848705869612874847264759487459126129754472747605411594961302478047284761546959536134277874734476254905982617011079447354763550559916203113182347364764561260136204114183247494765565660196205115196847504767570060326206143200047514768570160476209144200747524769571160816262150202047534770571960906273370
    Notice: Undefined offset: -1 in /var/www/html/ on line 207
    » «

Also Check out

Go to top