Abstract
Traditional Collaborative Filtering algorithms for recommendation are designed for stationary data. Likewise, conventional evaluation methodologies are only applicable in offline experiments, where data and models are static. However, in real world systems, user feedback is continuously being generated, at unpredictable rates. One way to deal with this data stream is to perform online model updates as new data points become available. This requires algorithms able to process data at least as fast as it is generated. One other issue is how to evaluate algorithms in such a streaming data environment. In this paper we introduce a simple but fast incremental Matrix Factorization algorithm for positive-only feedback. We also contribute with a prequential evaluation protocol for recommender systems, suitable for streaming data environments. Using this evaluation methodology, we compare our algorithm with other state-of-the-art proposals. Our experiments reveal that despite its simplicity, our algorithm has competitive accuracy, while being significantly faster.

Abstract
The ISTTOK tokamak was upgraded with a plasma control system based on the Advanced Telecommunications Computing Architecture (ATCA) standard. This control system was designed to improve the discharge stability and to extend the operational space to the alternate plasma current (AC) discharges as part of the ISTTOK scientific program. In order to accomplish these objectives all ISTTOK diagnostics and actuators relevant for real-time operation were integrated in the control system. The control system was programmed in C++ over the Multi-threaded Application Real-Time executor (MARTe) which provides, among other features, a real-time scheduler, an interrupt handler, an intercommunications interface between code blocks and a clearly bounded interface with the external devices. As a complement to the MARTe framework, the BaseLib2 library provides the foundations for the data, code introspection and also a Hypertext Transfer Protocol (HTTP) server service. Taking advantage of the modular nature of MARTe, the algorithms of each diagnostic data processing, discharge timing, context switch, control and actuators output reference generation, run on well-defined blocks of code named Generic Application Module (GAM). This approach allows reusability of the code, simplified simulation, replacement or editing without changing the remaining GAMs. The ISTTOK control system GAMs run sequentially each 100 µs cycle on an Intel® Q8200 4-core processor running at 2.33 GHz located in the ATCA crate. Two boards (inside the ATCA crate) with 32 analog-to-digital converters (ADCs) were used for acquiring the diagnostics data. Each ADC operates at 2 Msample/s but (for real-time operation) the acquired data is decimated in real-time on the board's Field-programmable gate array (FPGA) to a frequency defined by the control cycle time. This paper presents the ISTTOK real-time architecture and the human-machine Interface (HMI) for simplified AC discharge programming. © 2014 Instituto de Plasmas e Fusão Nuclear.

Abstract
The incorrect requirements elicitation, requirements changes and evolution during the project lifetime are the main causes pointed out for the failure of software projects. The requirements in the context of Software as a Service are in constant change and evolution which makes even more critical the attention given to Requirements Engineering (RE). The dynamic context evolution due to new stakeholders needs brings additional challenges to the RE such as the need to review the prioritization of requirements and manage their changes related to their baseline. It is important to apply methodologies and techniques for requirements change management to allow a flexible development of SaaS and to ensure their timely adaptation to change. However, the existing techniques and solutions can take a long time to be implemented so that they become ineffective. In this work, a new methodology to manage functional requirements is proposed. This new methodology is based on collecting and analysis of information about the usage of the service to extract pages visited, execution traces and functionalities more used. The analysis performed will allow review the existing requirements, propose recommendations based on quality concerns and improve service usability with the ultimate goal of increasing the software lifetime. Copyright © 2014 SCITEPRESS - Science and Technology Publications.

Abstract

Abstract
The present study surveys and maps the existing methodological approaches for measuring creative employment. Based on a unique matched employer-employee data-set which encompasses over three million Portuguese workers, we found that the magnitude of the creative class varies considerably between approaches, ranging from 2.5%, using the conventional industry-based taxonomy and 30.8%, using Florida's occupational proposal. The disparities are justified on the basis of the departure definition of what creative employment is and operationalization issues regarding which industries and occupations should be included. Interestingly, when we focus on core creative employment, the figures conveyed by the distinct approaches are strikingly similar (around 6%), suggesting that, at least where core creative employment is concerned, the distinct approaches converge. The diversity of approaches and measurements are not necessarily a bad thing in itself, but has to be adequately acknowledged in order to accomplish adequate public-policy guidance.

Abstract
The conventional assignment-based first/best fit decreasing algorithms (FFD/BFD) are not polynomial in the one-dimensional cutting stock input size in its most common format. Therefore, even for small instances with large demands, it is difficult to compute FFD/BFD solutions. We present pattern-based methods that overcome the main problems of conventional heuristics in cutting stock problems by representing the solution in a much more compact format Using our pattern-based heuristics, FFD/BFD solutions for extremely large cutting stock instances, with billions of items, can be found in a very short amount of time.