I can say with 100 per cent level of accuracy that when it comes to monitoring social media, sentiment analysis is one aspect of monitoring that we could talk on all day.
How accurate is yours – do you have stats on this? Are you 83 per cent accurate or 95 per cent? Lest we forget, one person’s positive is another person’s negative and so on.
Sentiment analysis is subjective and can often only be as good as the source dictionary, NLP list or human that makes the call. Other pitfalls are associated with accuracy across long form and short form content. A tweet by its very virtue of being only 140 characters is likely to render a higher level of accuracy in assessing sentiment than a long-form blog post of complex sentences, where multiple expressions of sentiment may be expressed within one sentence, paragraph or article unit.
This isn’t to say that sentiment analysis is wholly redundant. Alterian SM2 utilises a proprietary set of technologies including word parsing, weighting, proximity and Natural Language Processing to enable the most accurate and customisable sentiment analysis. It also allows you to develop and define customisable dictionaries, manually review and override sentiment, conduct mining in C# script and select multiple language sentiment. All that said, sentiment analysis should be judged with a pinch of salt.
Sentiment accuracy can be improved through data cleansing, combined machine and human analysis and by using custom dictionaries that are specific to the linguistics of the area of study that you are conducting. The automated analysis of sentiment may not always be something you would bet your house on but it can give you great guidance as to areas that need further human analysis and exploration.
In the same way that social media monitoring isnt just seeking out negative comments about your brand and responding to ‘neutralise a threat’, social media monitoring is made up of many parts, of which, sentiment is one – a useful one but not THE one. Never forget the business need you are actually looking to address with your monitoring, be that looking to instigate a sentiment shift or otherwise.
Do we get too hung up on but one part of what social media monitoring is about? How important a feature of a social media monitoring tool is sentiment analysis for you?
Take a look at our Social Media ROI series
