Å·±¦ÓéÀÖ

Try to think of the last time you contacted the call center of a company, where an automated system probably answered your call. Replacing the human factor presents many competitive advantages in terms of cost and availability. However, these systems do not fully incorporate the communicative behaviours humans use and hence are limited in reaching their full potential.

The more content that is available online, the less easy it is to discover and consume the most important information efficiently. Automatically extracting the gist of longer texts into an accurate summary and thus eliminating irrelevant content is urgently needed. Once again, machines can undertake this role.

Systems of this kind try to identify similarities between customers based on past behaviours and people with similar purchase habits can recommend products to each other. The benefit, in this case, is that customers are exposed to items in which they have never expressed any explicit interest.

A handy rule of thumb: if your feature set is to be useful, you need at least five training examples for every feature.

Thank you Nikos! Just started...