Abstract

Abstract
Industrial Internet and Industrial Internet of Things are emerging concepts that concern the use of Internet technologies on industrial environments. The main objective of such architectural visions is allowing a tight and seamless integration between all the functional units and layers that compose industrial processes, from the lowest levels (e.g. field level devices such as sensors and actuators) to the higher layers, including management, logistics and maintenance. This kind of architecture promises, among other advantages, improving efficiency and flexibility, reduce installation and maintenance costs and reduce unplanned downtime. However, industrial processes often encompass functionalities like closed-loop control of physical processes that are highly critical and have strict timeliness requirements. These requirements are not satisfied by normal Ethernet-based systems. Standards such as IEEE AVB and TSN are addressing this problem, enhancing the real-time properties of Ethernet. However, considering the information presently available, such standards still present some limitations and inefficiencies. This paper reports the extension of HaRTES, an Ethernet-based real-time architecture originally developed for use at the lower layers of industrial scenarios, with MAC Bridge standard functionalities, to make it capable of being integrated on Industrial Internet of Things frameworks. The paper also presents preliminary results obtained with a prototype realization of the extended HaRTES switch.

Abstract
Many online communities and services continuously generate data that can be used by recommender systems. When explicit ratings are not available, rating prediction algorithms are not directly applicable. Instead, data consists of positive-only user-item interactions, and the task is therefore not to predict ratings, but rather to predict good items to recommend - item prediction. One particular challenge of positive-only data is how to interpret absent user-item interactions. These can either be seen as negative or as unknown preferences. In this paper, we propose a recency-based scheme to perform negative preference imputation in an incremental matrix factorization algorithm designed for streaming data. Our results show that this approach substantially improves the accuracy of the baseline method, outperforming both classic and state-of-the-art algorithms.

Abstract

Abstract
Numerous stream mining algorithms are equipped with forgetting mechanisms, such as sliding windows or fading factors, to make them adaptive to changes. In recommender systems those techniques have not been investigated thoroughly despite the very volatile nature of users' preferences that they deal with. We developed five new forgetting techniques for incremental matrix factorization in recommender systems. We show on eight datasets that our techniques improve the predictive power of recommender systems. Experiments with both explicit rating feedback and positive-only feedback confirm our findings showing that forgetting information is beneficial despite the extreme data sparsity that recommender systems struggle with. Improvement through forgetting also proves that users' preferences are subject to concept drift.

Abstract