The concept of automated sentiment analysis has pretty poor reputation. Not least because expectations have been raised in the past by vendors only to be dashed on the hard rocks of failed promises.
Glide Intelligence – launched this morning to group of 50+ senior comms professionals at the CIPR HQ in Russell Square – thus enters the market with a hefty hurdle of cynicism to overcome.
However, having been involved in the beta testing of the product over the last 12 months – and having sampled many rival attempts at sentiment analysis in the past – I’m very optimistic that Glide Intelligence really does take a major step towards the holy grail of genuine, real time automated sentiment analysis.
So what sets Glide Intelligence apart from rival sentiment analysis systems?
- The product hasn’t been knocked together in five minutes. As Glide CEO Sam Phillips said this morning, the project started nearly 5 years ago and has seen a 7 figure investment in its development.
- One of the key brains behind the project is Keith Woods-Holder, who, if anyone, is entitled to the moniker of godfather of automated sentiment analysis. He began his career 25 years ago creating advanced mathematical models for the UK Government’s Advanced Planning Unit, followed by 3 years as Research Director at Saatchis. He was then recruited by IBM to set up KWHR, on of the first ever firms to build a commercial sentiment analysis model which was subsequently adopted by brands such as Kodak, Dell, Sony and NASDAQ (Keith does a good line in Michael Dell anecdotes). The man has form.
- The technology is based on 4th generation advanced NLP sentiment analysis. It is also context-based, rather than keyword or dictionary based. This means it gets over one of the major traditional objections, namely, that automated sentiment analysis can’t handle irony, sarcasm or slang.
- The breadth of sources. Glide Intelligence will monitor broadcast, print, online and social media all at once – and in real time. For example, you could have a real time, minute by minute, monitor of brand sentiment – and be able to spot where comms issues are developing in real time (just think what Peter Morgan at Rolls Royce could have done with this). An example was given this morning about how the tool could be used to monitor reaction to tube strikes – and where the sentiment is developing and how that is translating across media platforms. And what comms action could be taken – in real time.
- It can also be used to trace how a story developed eg if a particular article generates reaction across Twitter, broadcast, etc – and which could provide a blueprint for dealing with a similar issue in the future.
- Glide Intelligence provides multiple perspectives – in other words, not only can you view sentiment for your own organisation, but you can see how the world looks through your competitors eyes. The implications of this kind of analysis for comms professionals is obvious.
- Full transparency – you can pretty much drill down as far as you want to an original Tweet or article.
- The reporting capabilities are immense. More charts and tables than you can shake a stick at.
As you can guess, I’m very impressed with what I’ve seen so far. If Glide can deliver what’s on the tin, then perhaps the long awaited promise of automated sentiment analysis may finally have arrived.
Form an orderly queue for your demo now.
One reply on “Automated sentiment analysis? Yes, it is possible. And it’s here: Glide Intelligence”
[…] Automated sentiment analysis? Yes, it is possible. And it’s here: Glide Intelligence « In Front O… – not so sure myself – where is the great leap forward in machine learning? The NLP aspect to it doesn’t inspire confidence. Even if it was based on a giant database of lexeme clusters a la the work of Professor David Crystal at AND and Crystal Semantics I would understand, but from what I’ve heard so far I am skeptical […]