How it works
The “Recommender Systems” are systems that allow you to suggest personalized products to the users.
For each user, the system, generates a list of recommended products with relative rating( to be used ,for instance, in a website) or for each product a list of users interested in products with relative rating (to be used , for instance, in the launch of a new product).
- Suades can be used in every business context as it is out from the classical definition of product: for Suades a product is whatever interests the customer such as a business product, a company communication, some rules, a picture.
-Suades needs input data, mainly information about the user and his behaviors (purchases, assistance requests,web navigation….) from which the profiles of each user will come out.
- The System creates the array "User-Item(Users-products)" and, with its predictive algorithms for each user profile, provides the user's future behavior , showing the user only the products (communications , rules news, pictures) of his interest.
The Suades Recommender System uses both consolidated and the most used algorithms by the leaders of the specific field and the innovative algorithms developed by the Conquist and Ingenium team (which is a spin off Bari Polytechnic institute).
The consolidated algorithms allow you to:
- Recommend products that other customers liked, either similar to the connected customer or with whom you want to communicate (collaborative user-based or item-based): the algorithm creates a customer’s profile based on his preferences and feedbacks with reference to the products (e.g. purchase, observation,appeal) then identifies similar profiles belonging to other customers and foreshadows the customer’s preferences about products he has not seen yet.
- Recommend products with characteristics similar to those that have already enjoyed the customer (Content Based). The algorithm creates a customer profile based on the characteristics of the products to which the client has shown a feedback (eg purchase, observation, appeal), and identifies products with characteristics which are fit for the customer’s preferences.
The Innovative algorithms:
- Context-aware. The context is a variable that influences the user’s purchasing behavior and his products assessment. The algorithm allows the assessment of the contest in which the user has given his product assessment. This allows the creation of a user profile different for each context and therefore different recommendations consistent with each context.
- Profit-control. What to do if recommended products are only those with a smaller margin ? How to offer really attractive products to customers by monitoring the profit expectations ? The algorithm will recommend the products, also considering the company’s constraints.