According to Forbes analysts, one of the key trends in 2013 will be new solutions in the field of visualization and analysis of big data. A Big Data epoch is coming steadily, and now it is not collection, but analysis and visualization of a rising information wave that is coming to forefront. Speed, visualization and functionality are key performance indicators of analytic systems.
Large volumes of information require new visualization techniques, which would show deeper and more detailed interconnections, hierarchy elements, correlations among various objects etc. As noticed by one expert, David McCandless, visualization is one of the forms of data compression. Thus, a good visualization can greatly simplify analysis and processing of Big Data.
However most mass media and social media monitoring and analysis solutions today offer a severely poor and out-of-date arsenal – e.g. for sentiment monitoring.
The only result is an initial analysis of data, overall statistics, which will neither resolve, nor bring us closer to solving business tasks or assessing a market situation. In other words, although we have analyzed messages (marking each message as positive or negative), the only result is a general picture which provokes more questions rather than give answers.
Now let’s get back to the diagram. If you take a look at it, you as a company representative may have at least the following questions:
- What companies or persons were mentioned as positive and negative?
- Who spoke negative about me and positive about my competitors?
- Who speaks positive about me and positive about my competitors?
- Were there any bipolar messages (when I was spoken about both positive and negative)?
- What was I praised for and what was I scolded for?
Can one find answers to the above questions by a simple look at a certain graph? Yes. Moreover, you can get all answers by taking a look at a single diagram. And we are happy to offer you such opportunity.
In our work we try to stick to a principle that any visualization, any graph or diagram should answer business-relevant questions rather than serve as a nice picture – a form without content. The same also relates to more complicated variants of data display.
A SemanticForce platform utilizes an object approach to sentiment determination: a sentiment is determined for each object in a message, rather than for the entire message. For instance, there may exist comparisons with competitors – here it is significant to detect a sentiment not only to your brand, but also to the foreign one. People often say ambiguously about a product – some features suit them, while others displease them. It is impossible to interpret such messages using a unipolar sentiment, therefore an object approach and a bipolar sentiment is an essential tool of modern information analysis. When such object analysis of messages is complete, you can see an overall statistics – the number of opinions about you, what competitors were mentioned, what kind of tonality was used for them. But our goal was to make such statistics as visual and easy to assess at one glance as possible.
We had examined a lot of visualization options for that data (the above is one of them, where an overall statistics by mentions of competitors is seen), till we found a mechanism for effective and interactive reflection of object sentiment.
Here is our next innovation in data visualization techniques – an interactive diagram of object sentiment. It allows you to determine immediately, how often you were compared to your competitors, what companies were mentioned in the context with you and in what sentiment, as well as what exactly was said.
You can see a habitual circular diagram, where a sector with positive feedback is coloured in green, and the one with negative feedback is coloured in red. Additional circles on the sectors correspond to other objects, which were mentioned with the brand under research, with both nature and volume of mentions by competitors being clear at once. We recommend you look at green circles on the red sector – they are positive feedbacks about your competitors in the context of negative mentions of you. You can also easily understand a volume of bipolar mentions (which contain positive and negative at the same time) – a corresponding circle is shaded and related to your brand. More examples of diagrams by brands from different industries are given below.
By clicking on the object of interest you will get a more detailed statistics by mentions, their subject matter and volume, and be able to move directly to a list of those messages. Thus, just a few clicks will give an answer to almost any business question about your competitors. The diagram is realized on HTML5, so it can be run on iOS, Android and other platforms and devices.
The diagram is updated automatically in real time, therefore when the analysis of messages is done using an automatic tool (Blueberry text analysis platform that will be described very soon in a separate article), you can see an enthralling sight of changing indices on the fly.
We suppose that such visualization will be particularly useful for Social Media Control Centres, which we discussed in our blog – it will help orient oneself in high flow of input information and carry out a detailed analysis.
We plan to actively develop this kind of visualization, particularly adding other types of diagrams (e.g. emotions in messages, factor analysis etc.). We hope that the tool will be a valuable aid in analyzing market situation and customers’ behaviour and become a significant landmark in decision-making. We would also like to note that we see the semantic navigator as an entire platform for Big Data visualization and are open for cooperation with mass media, colleagues, suppliers of analytic systems etc.
We will be glad to answer questions in comments!
Подпишитесь и следите за нашими новостями на Facebook!