Sentiment analysis well worth further investigation as part of performance management armory, says Socitm briefing
Sentiment analysis, a way of using technology to aggregate many individual opinions, attitudes and emotions to gain insight into opinions of whole communities, may be something today’s head of ICT would do well to investigate further.
This is the view put forward in Sentiment analysis: a useful technique? which suggests that, with many senior ICT managers now responsible for customer service and digital access channels, the topic will be of interest as much from a service development standpoint as from a technical standpoint.
While it is unlikely that public sector organisations will be thinking of setting up their own sentiment analysis facilities, they might well wish to buy into one of the many commercial services on offer.
The briefing, therefore, aims to provide an insight into current practice with a view to helping readers to understand the potential of sentiment analysis, consider its relevance to their local needs, and to formulate searching questions for use in discussions with possible providers.
It begins by describing how sentiment analysis is done and its key tools: data mining (including big data), linguistics and natural language processing (NLP). It explains how the rise of social media, in particular Twitter and Facebook, is delivering huge volumes of ‘opinionated text in digital form’ which is particularly suitable for processing using the two latter techniques.
However, there are many pitfalls that lie in the path of sentiment analysis, many of them to do with potential misinterpretation by computer systems of the way words and phrases are used in real life.
But these challenges should not stop public sector managers giving serious consideration to sentiment analysis. Like algorithm-based search engines, sentiment analysis can be employed as a ‘black box’
solution, providing those buying into it are able to have an informed and critical conversations with potential solution providers and identify those that have addressed these issues with machine learning and other techniques.
The briefing goes on to describe how sentiment analysis has been used to predict election results and box office and stock market movements, while its most common application is in commerce to test customer opinion of goods and services.
In the public sector, its use may be as a means of better assessing reactions to, for example, new digital services. Initially, organisations might use sentiment analysis simply to validate what they are doing, but experience of the technique matures, the next step would be to use
it as a learning tool, applying information gleaned from customer reaction to development of future service designs.
As ICT is moves out from the back office into the mainstream of public service delivery, with many heads of ICT taking on customer services responsibilities, the question of the performance management of digital services looms large.
Services will want feedback on how their outputs are received by the community. How easy are services to use and what factors inhibit take- up? Sentiment analysis is a powerful way of gathering evidence, and being able to track changes over time.
It is a technique that Heads of ICT may conclude is well worth further investigation.
Vicky Sargent, Socitm Press Office
Tel: 07726 601 139 email: email@example.com
Martin Greenwood, Programme Manager, Socitm Insight
Tel: 01926 498703 e-mail: firstname.lastname@example.org