Whoever seeks will find, says an old proverb. But it doesn't consider the time you need to do it and mainly how much the outcomes really fit your preferences. Nowadays we live in a digital environment where we leave daily evidence of our presence and preferences, commenting on social networks, “googling” information or checking-in into a specific place. This is not noise: on the contrary, it represents an extremely rich context where tools like recommendation systems still have room to improve accuracy and appeal of their issues.
In this sense, Tykli Forward can provide a powerful tool for profiling users and personalizing suggestions for them, adding this enormous number of external inputs to information that traditionally helps understand users' behaviour and exploring and ranking relationships that lie among them.
Recommendation systems work, for instance, behind user interfaces of on demand tv channels, music web players, news sites and e-commerces, and no matter what the user looks for, this kind of activity falls into what we define as complex systems, as we have:
- many subjects who act as interconnected parts, i.e. thousands of surfers who, through their computers, search, listen and recommend to other users the latest rock songs, even outside the borders of the specific website where they do that;
- a continuous flow of information moves from users to content providers and back, but also among users and even, as told before, in less predictable directions thanks to the digital interactions we do everyday in a number of different channels;
- a big amount of data to deal with that is generated in this way.
Until now, recommendation systems process different sorts of information about consumers: who and how many they are, which kind of subscription they have chosen, which types of items they prefer, when they make use of them during the day, how much time they spend examining each one. The same goes for products: price, features, internal rating. But now they are not enough: relationships among these elements and all those that come from the surrounding contexts are fundamental.
Tykli Forward, beyond collecting this sort of information, connect them, considering what happens in all environments where users act. So, not only the fact that one client buys a movie, sees it once at evening time, but also that two days later he comments on Twitter about the main actor who is making a new film in his own city and then several people join the discussion, commenting and suggesting another movie. Giving a different value for each of these connections, it enhances the skill to better predict what users like; at the same time, it works in a flexible way because the relevance of every relationship discovered varies according to the starting point of researches.
Don't hesitate to contact us to know more.