Client profile prediction using convolutional neural networks for efficient recommendation systems in the context of smart factories.

In this work, we exploit a classifier based on convolutional neural network to predict users' interests regarding products and services, according to their publications in social media. We aim at designing a customizable recommendation system to target social networks' users as potential buyers. We evaluate the performance of training algorithms that use adaptive learning. The evaluation is performed in terms of the achieved accuracy rate to predict user's interest. The selection result is backed up by the multi-criteria decision making. Two metrics are used to identify the real user interest profile of users based on publicly available photos. [ABSTRACT FROM AUTHOR]