Ebook: Smart Digital Futures 2014

This volume contains the proceedings of three KES conferences: Intelligent Decision Technologies (KES-IDT-14), Intelligent Interactive Multimedia Systems and Services (KES-IIMSS-14), and Smart Technology-based Education and Training (KES-STET-14). These events were held as part of the Smart Digital Futures 2014 (SDF-14) multi-theme conference in Chania, on the island of Crete in Greece, on 18–20 June 2014.

Smart Digital Futures was organised as part of the KES International conference portfolio. For over a decade the mission of KES International has been to provide a professional community, networking and publication opportunities for all those who work in knowledge-intensive subjects. At KES we are passionate about the dissemination, transfer, sharing and brokerage of knowledge. The KES community consists of several thousand experts, scientists, academics, engineers, students and practitioners who participate in KES activities. KES runs conferences in different countries of the world on leading edge topics including Intelligent Systems, Sustainability in Energy and Buildings, Innovation, Knowledge Transfer and Enterprise, and Digital Media. KES provides routes to publication, such as journals, book series, and online publications, both through its own media and in conjunction with major publishers. KES also provides knowledge transfer services and consultancy through the Institute of Knowledge Transfer (IKT).

The aim of Smart Digital Futures was to bring together researchers working at the leading edge of developments in smart systems and intelligent technology theory and applications.

This volume contains approximately 80 papers selected from a much greater number after full peer review by members of the International Programme Committees of the three conferences.

We thank the four keynote speakers for providing informative accounts of the latest research in this area: Prof Vladimir Tikhomirov, Moscow State University of Economics, Statistics and Informatics, Russia; Prof George A. Tsihrintzis, University of Piraeus, Greece; Prof Nikos Tsourveloudis, Technical University of Crete, Greece; and Prof Maria Virvou, University of Piraeus, Greece.

We appreciate the efforts of those organising and chairing special sessions, members of the International Programme Committee and reviewers and we thank them. We are grateful to authors for the papers they have contributed to the conference, and to delegates for their attendance and we give them our thanks.

Finally we thank the conference administration for their organisational efforts, and the people of Chania for welcoming us to their city.

R.J. Howlett and L.C. Jain

Various generalizations of fuzzy reasoning are frequently used in decision making. While in many application areas it is natural to assume that truth degrees of a property and its complement sum up to 1, such an assumption appears problematic, e.g., in modeling ignorance. Therefore, in some generalizations of fuzzy sets, degrees of membership in a set and in its complement are separated and are no longer required to sum up to 1. In frequent cases, this separation of positive and negative evidences for concept membership is more natural.

As we discuss in the current paper, symbolic explanations of results of such forms of reasoning provide additional important information. In the present paper we address two related questions: (i) given generalized fuzzy connectives and a finite set of truth values τ, find a finitely-valued logic over τ, explaining fuzzy reasoning, and (ii) given a finitely-valued logic, find a fuzzy semantics, explained by the given logic. We also show examples illustrating usefulness of the approach.

As Facebook becomes a quite relevant tool for companies marketing and sales it is important to analyze and understand posting activity. Propagation of relevant episodes in Facebook is quite fast and companies must not only plan, monitor and control the posting activities in their own Facebook page but also understand what is happening in their competitors Facebook. This paper presents a model and algorithm that allows the implementation of automated monitoring of Facebook posting activity, identifying normal and outliers in their activity, and hence enhancing companies' Facebook competitive intelligence. The model is validated with a data sample of 27924 public publications from the 550 companies Facebook pages.

Determinization of finite automata is performed by the Subset Construction algorithm (SC). However, some application domains, including monitoring and diagnosis of active systems in artificial intelligence, and model-based mutation testing in software engineering, require determinization to be performed incrementally, in real time. Making incremental determinization by means of SC is bound to poor performances. To this end, an algorithm called Incremental Subset Construction (ISC) was proposed a few years ago. Disturbingly, this algorithm was recently discovered to be incorrect is some instance problems. The incorrect behavior of ISC originates when the redirection of a transition causes a portion of the DFA to be disconnected from the initial state. This paper provides a revisited version of ISC, called Revised Incremental Subset Construction (RISC), operating in either busy or lazy mode. In busy mode, the disconnection is always avoided. In lazy mode, since no check of disconnection is performed, a garbage collector is eventually required.

Due to the scarcity of raw materials, the recycling of end-of-life vehicles is becoming increasingly important. The essential decision for the cycle of materials is made in the dismantling company which disassembles the vehicle parts and determines the further recycling path: reuse as a replacement part, specific substantial exploitation or exploitation by shredding. This decision is the central aspect of this paper, taking into account the economic aspects, the uncertainties of the market and the applicability of the method. Therefore a detailed cost analysis model is presented, including a method for prediction of the replacement part market. In addition an IT concept is presented to visualize the result of the analysis and support the decision making.

Multi-objective optimization is a way to manage multiple objectives in analytical decision support systems. However, for real-life problems, different types of uncertainty often become prominent when defining the model. In this paper, we analyze these different types of uncertainties and suggest a suitable typology for a decision process based upon multi-objective optimization models. Uncertainty analysis can be performed based on the proposed typology; therefore, this analysis provides the necessary support for a decision maker in the identification the crucial uncertainty in the decision process.

Forensic inferential reasoning is a “fact-finding” journey for crime investigation and evidence presentation. In complex legal practices involving various forms of evidence, conventional decision making processes based on human intuition and piece-to-piece evidence explanation often fail to reconstruct meaningful and convincing legal hypothesis. It is necessary to develop logical system for evidence management and relationship evaluations. In this paper, a forensic application-oriented inferential reasoning model has been devised base on Bayesian Networks. It provides an effective approach to identify and evaluate possible relationships among different evidence. The model has been developed into an adaptive framework than can be further extended to support information visualisation and interaction. Based on the system experiments, the model has been successfully used in verifying the logical relationships between DNA testing results and confessions acquired from the suspect in a simulated criminal investigation, which provided a firm foundation for the future developments.

The subject of this paper is developed in the scope of decision support to reduce energy consumption in industrial processes. It starts from the observation that context under production units are producing might have a strong influence on the specific energy consumption. In this line, this paper proposes an approach to model energy consumption taking the value of context variables in consideration. The approach is based on the estimation of multiple models, using RLS and valid in identified context regions, collecting the necessary knowledge to support future decision making processes. An example using experimental data from a cement factory is used to illustrate the proposed methodology.

Increasing use of commercial off-the-shelf Mini-Micro Unmanned Aerial Vehicle (MAV) systems with enhanced intelligence methodologies can potentially be a threat, if this technology falls into the wrong hands. In this study, we investigate the level of threat imposed on critical infrastructure using different MAV swarm artificial intelligence traits and coordination methodologies. The critical infrastructure in consideration is a moving commercial land vehicle that may be transporting for example an important civil servant or politician. Non-dimensional fitness functions used for measuring MAV mission effectiveness have been established for the case studies considered in this paper. The findings indicated that increased in intelligent and coordination level elevate teams' efficiency, therefore poses a higher degree of threat to targeted land vehicle. Observations from the study have suggested that memory-based cooperative technique provides a consistent efficiency compared to other methods for the mission objectives considered in this paper.

The position tracking of laser beams or laser spots at a distance is often achieved with the aid of a four quadrant detector. Such systems are used to detect and track pulsed laser radiation. The tracking system is capable of determining the position error of a received pulse and then to realign the quadrant detector centre. The purpose of this study was to test a position predicting algorithm, which used a neural network to track a signal in the presence of an additional signal and noise. The model was constructed using both simulated and experimentally obtained data from a quadrant detector. The results show that automated tracking of a single spot was possible but tracking of the spot in the presence of an additional signal required re-evaluation of weights in the neural network model.

In this paper we present our experience in developing a fuzzy-logic based negotiation system capable of achieving a mutually beneficial deal for the seller and buyer in uncertain situations. Fuzzy utility in our system allows users who are often unsure about their utility function to express their preferences in fuzzy terms such as low, middle and high. The system evaluates offers based on this fuzzy utility and feeds utility score along with remaining negotiation time to a fuzzy inference system to compute conceding rate of its next counter offer. The experimental results have shown that that the system concedes less when negotiated item matches its preferences less and concedes more when the negotiated item matches its preferences closely and when negotiation is ending. Our concluding remarks and future research are presented.

This study analyzes consumer purchasing behavior by introducing voluminous product information into a consumer decision-making model. We focus on the case where consumers and firms have insufficient preferences for product attributes. This situation can apply to the case where a new product or product with new features is introduced into a market. We analyze consumer dysfunction in the presence of excessive alternatives and investigate how many products and product attributes are offered when selling a product. Following the ideal point model, we construct a new model and investigate it. Furthermore, by relaxing the assumptions made in the theoretical model, we numerically investigate and verify the robustness of our theoretical results. The results reveal that consumers may not be able to choose the best product because of confusion and dysfunction of their preferences for product attributes. This confusion causes consumers to choose each product with equal probability.

Understanding mutual relation between informal meaning in human mind and formalised semantics is crucial for formalizing human processes of thinking when performing them on computers. It is fictious to suppose machines are able to process informal meaning in the way as it is in human mind. On the other side, reflecting the requirements for rapid, flexible and automated changes in real world a new quality of informal processing with help of machines is needed. That is why binding an informal meaning and formal semantics is in the centre of many applications that have attributes of machine intelligence. Moreover, we are strongly interested in automating the production of these applications. By other words, we need rather to evolve them in an automated manner than to construct them mannually. This paper is devoted to mutual binding of metasemantics via semantics to informal meaning, with abstract interpretation as a central point, introducing simple example, how it works. We define mutual binding of informal meaning to abstract symbols of regular expression. We define the semantics of operations in terms of metaoperations inherited from EBNF metalevel, and finally, we derive the minimum-state deterministic finite state automaton, which reflects active bindings to informal meaning in its states.

We propose an exam scheduling approach to deal with problems that may appear in some oral exams, such as the cases when student turnout is considerably above or below expectation. As opposed to similar approaches, we focus on predicting the number of students applying for an exam by performing data mining on student records. Our predictive model considers previous student scores, attendance records, and past exam attempts. We evaluate the prediction segment of this approach on a real-world data set containing university records for a pair of database-related courses.

We have proposed a super pairwise comparison matrix to express all pairwise comparisons in the evaluation process of the dominant AHP (analytic hierarchy process) or the multiple dominant AHP as a single pairwise comparison matrix. This paper shows the treatment of hierarchical criteria in the dominant AHP with super pairwise comparison matrix.

In this paper, we describe methods which synthesize scores of AHP with its alternatives categorized. First, we proposed simple way to synthesize the scores, and we show that this approach occurs a rank reversal problem. Then we improve the method by combining Dominant AHP to overcome the problem.

In this study, for the Analytic Hierarchy Process (AHP) and the Analytic Network Process (ANP), the improving normalization method is considered. In the traditional AHP and ANP, the sum of weights of alternatives is normalized to 1. Using this method, some problems, for example rank reversal, were pointed out. Based on the numerical results, one of the causes of these problems is the inappropriate normalizing process. Therefore, for the AHP and the ANP, at first, in this paper, two kind of normalization method are described. One is normalizing to 1 to the sum of the column vector of alternatives for each criterion, and other is normalizing to 1 to the maximum element of each column vector of alternatives. At second, some problems by using two kind of method are shown. At third, by applying proposed method to the verification example, it shows that the correct result is obtained. Finally, proposed normalization method is discussed and concluded.

This paper reviews the role of Consistency Index of Analytic Hierarchy Process (AHP). On the hypothesis that the index is able to discriminate the consistency of judgments, discriminating power of the index is verified. The verification is carried out through survey questionnaire consisting of questions asking the same issue in different ways; one using so called ranking method, the other requires respondents to conduct pairwise comparisons. The result implies that the Consistency Index may not be able to distinguish the consistency of judgment.

Existing collaboration support systems cope poorly with today's working environments. Although they offer functionalities to overcome cognitive overhead, these prove to be insufficient in data-intensive situations. In this paper, we identify the main causes of cognitive complexity in contemporary collaboration support systems and present the approach adopted in the Dicode project to address such concerns. We discuss how Dicode mitigates the effects of cognitive overload issues through an adaptive collaboration support infrastructure that allows organizational aspects of the discourse to change according to the user's needs, interests and specific tasks during the evolution of the collaboration.

The development of argumentation support systems for different types of groups and application areas has been receiving growing interest in the last twenty years. Such systems address the needs of a user to interpret and reason about knowledge during a discourse, and demonstrate diverse human and machine reasoning functionalities. However, methodologies to check whether the reasoning mechanisms of such systems adhere to broadly accepted argumentation theories are missing. Provision of such methodologies is of much value, especially in data intensive contexts. The approach described in this paper is a first step towards this direction. Specifically, we formally assess a specific argumentation support system, namely HERMES, against Dungs argumentation theory and prove its correctness as far as the acceptability of arguments is concerned.

This paper presents a trial study to gain insight into collaborative sensemaking behaviour. We use a generic argumentation tool to capture interaction data, which is then analyzed by exploiting collaboration-specific semantic types. A Community Behaviour Analytics Tool, CommBAT, was developed. Further studies were conducted to visualize the different patterns in the collaborative behaviour. Initial findings show the power of semantics in helping to detect a series of sensemaking behaviour patterns.

In this paper, for solving imbalanced classification problem, more attention is placed on data points in the boundary area between two classes. The fuzzy k-nearest neighbors algorithm, which has good performance in conventional classification problems, is adapted here to solve imbalanced classification problems, where G-mean accuracy is used to evaluate our proposal method and compare it with other approaches.

In this paper, a Radial Basis Function Network (RBFN) trained with the Dynamic Decay Adjustment (DDA) algorithm (i.e., RBFNDDA) is deployed as an incremental learning model for tackling transfer learning problems. An online learning strategy is exploited to allow the RBFNDDA model to transfer knowledge from one domain and applied to classification tasks in a different yet related domain. An experimental study is carried out to evaluate the effectiveness of the online RBFNDDA model using a benchmark data set obtained from a public domain. The results are analyzed and compared with those from other methods. The outcomes positively reveal the potentials of the online RBFNDDA model in handling transfer learning tasks.