Search

The continuous increase in sophistication of threat actors over the years has made the use of actionable threat intelligence a critical part of the defence against them. Such Cyber Threat Intelligence is published daily on several online sources, including vulnerability databases, CERT feeds, and social media, as well as on forums and web pages from the Surface and the Dark Web. Named Entity Recognition (NER) techniques can be used to extract the aforementioned information in an actionable form from such sources. In this paper we investigate how the latest advances in the NER domain, and in particular transformer-based models, can facilitate this process. To this end, the dataset for NER in Threat Intelligence (DNRTI) containing more than 300 pieces of threat intelligence reports from open source threat intelligence websites is used. Our experimental results demonstrate that transformer-based techniques are very effective in extracting cybersecurity-related named entities, by considerably outperforming the previous state- of-the-art approaches tested with DNRTI.

Serverless and the Function-as-a-Service (FaaS) paradigms are seen as two enabling technologies for next-generation computing on the cloud continuum. This article discusses prominent frameworks to deploy and monitor applications that span the cloud continuum. It discusses associated challenges and proposes a novel architecture for a framework that manages intelligently multi-cloud, fog and edge resources in order to cope with the requirements of FaaS-enabled applications and services.

The team-formation problem on social networks asks for a team of individuals that collectively possess the skills to perform a task and have low communication cost, as measured by their distances in the social network. Most related work assumes a flat structure in the team, where team members are all indistinguishable. However, in practice, teams often have complex structures and deep hierarchies, and members with distinct roles in these structures. In this paper, we consider the Template-Driven Team Formation problem, where given a fixed template structure for the team, in the form of a graph, and a designated role for each node in the template, we ask for workers that can fill the roles in the template, while minimizing the communication cost along the template edges. Although the problem is in general NP-hard, there are variants of the problem that can be solved optimally using dynamic programming. For the general case, we provide approximation and heuristic polynomial-time algorithms. We experiment on real data and we demonstrate that our heuristic algorithms perform well in practice while being significantly more efficient. Our case studies highlight the quality of the teams produced by our algorithms.

Deep vein thrombosis (DVT) is a condition, which results from atherosclerosis and the possible formation of a blood clot (thrombus) in a deep vein leading to obstruction of blood flow. DVT is considered a major cardiovascular health risk and its early detection is considered crucial to patient survival and well-being. The present work describes the development and evaluation of two image analysis techniques for the identification of DVT in children. The first technique is based on entropy thresholding and the second on active contours. The robustness and sensitivity of these techniques have been evaluated both quantitatively as well as qualitatively. Towards that purpose error parameters were calculated using phantom and real patient images. The usefulness of our techniques was evaluated by specialist radiologists. The suggested algorithms were developed with the Matlab® simulation environment. Thus the present work has been implemented towards the improvement of pediatric patients' well-being and at the same time achieve early DVT prognosis.

Sports Analytics is an emerging research area with several applications in a variety of fields. These could be, for example, the prediction of an athlete’s or a team’s performance, the estimation of an athlete’s talent and market value, as well as the prediction of a possible injury. Teams and coaches are increasingly willing to embed such “tools” in their training, in order to improve their tactics. This paper reviews the literature on Sports Analytics and proposes a new approach for prediction. We conducted experiments using suitable algorithms mainly on football related data, in order to predict a player’s position in the field. We also accumulated data from past years, to estimate a player’s goal scoring performance in the next season, as well as the number of a player’s shots during each match, known to be correlated with goal scoring probability. Results are very promising, showcasing high accuracy, particularly as the predicted number of goals was very close to the actual one.

Developing an educational institution timetable is a complex problem which requires finding a successful combination of all the parameters involved (courses, professors, students, classrooms, etc.). To address this problem we developed a prototype algorithm that is a hybrid form of the Particle Swarm Optimization (PSO) algorithm. The original PSO algorithm simulates the mode of a bird cluster movement into nature. In particular, as in this case the solution to a problem with discrete values is needed, we developed a hybrid form of this algorithm with local search, in the process of which we incorporated original methods. The main contribution of this paper is how to improve particles based on optimal Gbest (Global best) and Pbest (Particle best) values of the particles. Our work provides also a fully detailed description of the innovate solution on how to update the algorithm particles in each iteration of the optimization process (Local Search). Our algorithm achieves satisfactory results within seconds.

Experience from developing, in-house, an e-governance solution for a Greek public agency is reported here. The article presents the e-governance solution developed, the problems encountered in making the system accepted by the agency’s employees and the barriers that had to be overcome in order to establish basic e-governance procedures. We believe that this article is one of the few in the literature that presents a "report from the trenches" and the problems that the people that tried to implement the change actually encountered.

Existing digital tools used by journalists such as content management systems and search engines focus on helping journalists find relevant information and organize their creative work; they do not provide support for discovering creative angles to investigate. To meet this need in journalism, we have researched new creative search algorithms that manipulate the journalist’s primary medium – written information – and developed an innovative creativity and productivity support tool deliberately for use by journalists. In this paper, we analyse the creative tasks journalists perform when developing new stories, we provide an overview of requirements, which have been addressed by the INJECT digital creativity support tool for journalists. Further, we explain the intended use of the system using a walkthrough scenario and present our early experiences from deploying the tool in journalist’s working environments.

Current topics in e-Learning Management Systems encourage innovative technologies to use digital libraries to exploit Learning Objects metadata in order to facilitate dissemination and re-process of their content. Digital repositories are mostly implemented as distributed computing systems architectures dealing with major technological and modeling issues that hinder interoperability between heterogeneous databases impeding data accessibility and reusability. Learning Object metadata are lack of interactivity and address operational deficiencies. Though, a content-structure and systematic approach needs to be established. A promising method to employ learning object metadata adaptation is to use semantic Web technologies. Expressing semantics in Learning material can heal interoperability issues and improve information retrieval by providing knowledge representation inherent in xml syntax and applying inference rules. This work develops an ontology-web model for learning object metadata. Detecting semantic relations on educational material aims to assess in an abstract level its tutoring content to improve the learning procedure.

The emergence of the Internet of Things paves the way for enhancing the monitoring capabilities of enterprises by means of extensive use of physical and virtual sensors generating a multitude of data. The generated real-time data streams provide the basis for anticipating future undesired events and thus enable enterprises to decide and act ahead of time, i.e. in a proactive manner. Proactive information systems can benefit from the use of context-awareness in order to provide more reliable recommendations. In this paper we propose a probabilistic model for context-aware proactive event-driven decision making that deals with the uncertainty that is inherent in the contextual elements' values. We validated our approach in an oil and gas industry scenario and we conducted a comparative analysis.

We implemented a recommendation system that presents persuasive messages to commuters using a route planning application. Our aim is to nudge users towards following routes that are environmentally friendly with lower CO2 emissions than those they usually take. Our implementation leverages contextual information and transportation user profiles while the messages are defined based on a set of persuasive strategies. The recommender is integrated in a route planning application for android based smartphones and was used in a trial in the cities of Vienna and Dublin. Users' qualitative feedback showed that the feature was found interesting and provided insights for future improvements.

Risk analysis has become recently an important activity in many industrial domains: mining, off-shore drilling, health services etc. For short, risk analysis studies the relationship between causes of an accident and the severity of its consequences. It comes as no surprise that Risk Management Information Systems (RMIS) for tracking and reporting the costs of preventing accidents and mitigating the consequences of critical events have attracted a lot of research and development efforts. The amount and especially the complexity of data in a typical RMIS make such systems very difficult to design. In this paper we present a popular model for risk management data representation, the bowtie model, which is almost the de facto model in risk management systems. The bowtie model we present is inherently a directed graph. Then we show how we define and use an extension of the XQuery language, called TreXQuery, to perform intricate searching of graph models in risk management systems.

The rapid evolution of technology in Management Information Systems has enabled easy access to a huge amount of complex and specialized business information. The vast amount of business information has created the need for investigating ways to improve the efficiency and effectiveness of information retrieval techniques for querying business information. Among the different types of business information, information about business processes has been leveraged with dedicated languages, methods and tools. Business Process Management Notation (BPMN) is an international standard for modeling business processes. In this paper, we first develop the appropriate business process models of a complex real-life scenario using BPMN and we then represent process models in the form of an ontology business process model. We, next, query the two models with the same queries: the BPMN model using a query language for object-oriented databases and the semantic model using the SPARQL ontology-based search language. We compare the two query methods, discuss the results and explain the benefits of using semantics-based information retrieval techniques when dealing with business process information.

Risk management is a critical part of every industrial activity. In recent years, due to the complexity, importance and potential impacts of industrial processes, risk analysis has taken advantage of increasing computational availability to deliver tools at every level of the risk management spectrum. Bow-tie models, so called because of their generalized shape, are used at various levels in the risk analysis process, fundamentally because their presentation aggregates complex, real-world scenarios in consistent, reducible and communicable ways. As a result, vendors have sought to map additional capabilities onto the Bow-tie extending their applicability into qualitative, semi-quantitative, and quantitative risk assessment modalities. In this paper we present a generalized computational framework for risk analysis using Bow-tie models that simultaneously perform all such analytical modes.

In order to understand how users interact with PV-powered products, the behavior of 100 people was observed and analyzed, while interacting with 6 different photovoltaic-powered products in their daily life. The sample of users to be observed consisted of 20 groups, each one formed by 5 students of the Industrial Design Engineering Department of TU Delft in the Netherlands. The tested PV products were: the IKEA Sunnan lamp, the Waka Waka light, the Waka Waka Power (charger and light), the Little Sun light, the Logitech solar keyboard and the Beurer kitchen weight scale. In this study the design of the 6 tested PV products was analyzed, users' expectations were outlined and the users' opinion of the products' performance after use was addressed. Main results show that the users need more reliable PV products, made with nice materials, that have an interesting design, and perform sufficiently. Consumers are willing to invest money and buy a PV product if it is useful and works properly.

Public Sector has a wide range of information to share. However, there is lack of interoperability in the information exchange between organizations due to the different terminologies and techniques adopted by business entities. Combining Semantic technologies with Management Information Systems is a growing research area. Business Process Management Notation is a good practice for examining these fields. In this work, we develop a business process model of a real-life scenario in a semantic representation and define the generic semantic concepts using the Business Process Management Notation.

Effective management of the supply chain is becoming increasingly important for modern enterprises. A supply chain consists of several sub-processes, internal and external to the enterprise, which are nevertheless rarely executed as planned because deviations due to disruptive events are very common. In this paper we present an event-based framework which is designed to sense events generated by external and internal resources which can lead to undesired deviations in the supply chain. Moreover, it re-synchronizes the whole supply chain process so as to minimize the influence of undesired situations. We implemented and tested this framework with a service based application designed with BPEL 2.0.