We live in a world where more choice and information are usually well received. No wonder then that the three-in-one combination represented by RMS (Recognition, Mining and Synthesis) applications—that’s expected to take the load of analytically sifting through excessive information off users—is predicted to storm our world.

The information age has indeed come a long way. From IT users initially craving for information, now a certain fatigue syndrome prevails as a much larger group of users struggle to make sense of the volumes of digital content that they are bombarded with. This scenario is so widespread that in-sync technopreneurs may soon realise (if they haven’t already!) that the next killer application will be one that offers to solve this problem of data explosion for its users.

Does Intel have an answer?

Perhaps they will be pipped to the post by Intel, a company that appears to be many steps ahead in envisaging the future of killer applications. Its RMS taxonomy, for instance, describes a class of emerging applications that bring together three ‘intelligent’ functions a computer was hitherto known to perform singly—recognition, mining and synthesis.

Computer vision has been considerably talked about in recent years, just as data mining has become a focal point. Likewise, graphics as a kind of synthesis application has done many rounds of tech-savvy design circles. So how would it pan out, merging these three separate kinds of applications?

Computer recognition is machine learning that enables a computer to ‘identify’ what it ‘sees’ as it meets a given set of specifications, and subsequently, enables it to model the object or event. Having established recognition, an application can then proceed to sift through huge, complex, static or streaming datasets to mine instances of the model determined to be of interest to the user. However, there may be times when exactly-matching instances of the model are not found. In such cases, the application’s ability to synthesise takes over to create ‘what if’ virtual scenarios, or perform calculations on the model to yield the desired solutions.

RMS at work

Sounds complicated? A real-time example of the use of these component functional technologies in an iRMS (interactive RMS) loop is, say, a virtual dressing room that uses archived apparel images to show you how you would look in them! First the recognition function would be implemented to recognise you (a human object). Then the mining function would hunt out the type of clothes you are looking for—let’s say trouser suits—and the synthesis function would enable you to ‘try on’ various combinations of these clothes, and in every conceivable colour, to boot!

While this example relates to modelling an object (you!), RMS could also model an event. Mathematician and emergent technologist Franz Dill, from Procter & Gamble, explains how he perceives RMS as a possible means to construct systematised models that link the real world with very large-scale data: “I work with retail and supply-chain applications...an example in this domain might be a sensor system that detects consumer traffic in a store, and then based upon that, adjusts inventory management decisions, and databases of sales and inventory. The adjustment would be the event that would be scheduled.”