How to Analyse Sentiment and Benefit from the Insight it Provides

With just two weeks until Monitoring Social Media Bootcamp, we thought we’d give you a taster of the event with a series of posts about the workshops we’ll...

sentiment analysisWith just two weeks until Monitoring Social Media Bootcamp, we thought we’d give you a taster of the event with a series of posts about the workshops we’ll be having. First up is Marshall Sponder’s session called “How to Monitor Sentiment and Benefit from the Insight this Provides”.

Marshall has spent much of the last ten years trying out various social media monitoring solutions and sentiment is one of his favourite topics. In this workshop he aims to explain, in layman’s terms, how best to use the sentiment analysis features of social media monitoring tools, how to make sense of the results they produce and how to create value from this knowledge.

First off, Marshall will analyse the different approaches to sentiment from some of the leading monitoring solutions on the market, including Brandwatch, Scoutlabs, Radian6, Sysomos, Crimson Hexagon and Alterian (SM2). He will also demonstrate the differences in sentiment analysis results that these solutions can produce from essentially the same data. Scary stuff if you’re paying good money for comprehensive results!

One of the other key questions Marshall will be addressing in this session is: when is sentiment analysis useful and when isn’t it? He will explore which aspects of social media are best analysed numerically and identify those where sentiment can offer genuine insight and value, citing examples of how – right now – businesses are benefiting from each approach.

Marshall will also look at the accuracy of sentiment analysis. In other words, are the results produced related to the topics we’re interested in? And to what extent can results be improved by filtering out noise? He will demonstrate how to remove non-relevant search results and how, using your social media monitoring tools, you can construct queries that produce accurate results.

One of the hottest issues in sentiment analysis is always the “human” or “machine” intervention question. The question being, should we employ humans to analyse results and rate the sentiment, or should we develop sophisticated reading technology to rate sentiment for us. Marshall will offer his view on which works best (in various case studies) and suggest which option is best for what situations and how accurate you can expect the data from either option to be.

Finally, Marshall will prophesise what we should expect for sentiment analysis over the coming few years. As one of the world’s most experienced social media monitoring analysts – I would expect his opinion to be around 89% accurate 😉

In this article

Join the Conversation

1 comment

  1. The Key Issues in Social Media Monitoring Today « The Cube Reply

    […] Sentiment detection is one of the most interesting aspects of social media monitoring. Using a combination of machine learning and natural language processing, brands can now track positive and negative mentions to around 70% accuracy. Be aware though, that these tools require teaching – you can’t get those results from a “black box” solution. Equally, with sentiment context is king; what’s positive for your product team may be inconsequential to your marketing department. […]