Networks encode dependencies between entities (people, computers, proteins) and allow us to study phenomena across social, technological, and biological domains. These networks naturally evolve over time by the addition, deletion, and changing of links, nodes, and attributes. Despite the importance of modeling these dynamics, existing work in relational machine learning has ignored relational time series data. Relational time series learning lies at the intersection of traditional time series analysis and statistical relational learning, and bridges the gap between these two fundamentally important problems. This paper formulates the relational time series learning problem, and a general framework and taxonomy for representation discovery tasks of both nodes and links including predicting their existence, label, and weight (importance), as well as systematically constructing features. We also reinterpret the prediction task leading to the proposal of two important relational time series forecasting tasks consisting of (i) relational time series classification (predicts a future class or label of an entity), and (ii) relational time series regression (predicts a future real-valued attribute or weight). Relational time series models are designed to leverage both relational and temporal dependencies to minimize forecasting error for both relational time series classification and regression. Finally, we discuss challenges and open problems that remain to be addressed.

With the growth of the Internet, crowdsourcing has become a popular way to perform intelligence tasks that hitherto would be either performed internally within an organization or not undertaken due to prohibitive costs and the lack of an appropriate communications infrastructure. In crowdsourcing systems, whereby multiple agents are not under the direct control of a system designer, it cannot be assumed that agents will act in a manner that is consistent with the objectives of the system designer or principal agent. In situations whereby agents’ goals are to maximize their return in crowdsourcing systems that offer financial or other rewards, strategies will be adopted by agents to game the system if appropriate mitigating measures are not put in place. The motivational and incentivization research space is quite large; it incorporates diverse techniques from a variety of different disciplines including behavioural economics, incentive theory, and game theory. This paper specifically focusses on game theoretic approaches to the problem in the crowdsourcing domain and places it in the context of the wider research landscape. It provides a survey of incentive engineering techniques that enable the creation of apt incentive structures in a range of different scenarios.

The International Planning Competition (IPC) is a prominent event of the artificial intelligence planning community that has been organized since 1998; it aims at fostering the development and comparison of planning approaches, assessing the state-of-the-art in planning and identifying new challenging benchmarks. IPC has a strong impact also outside the planning community, by providing a large number of ready-to-use planning engines and testing pioneering applications of planning techniques.

This paper focusses on the deterministic part of IPC 2014, and describes format, participants, benchmarks as well as a thorough analysis of the results. Generally, results of the competition indicates some significant progress, but they also highlight issues and challenges that the planning community will have to face in the future.

Recent proliferation of surveillance systems is mostly attributed to advances in both image-processing techniques and hardware enhancement of smart cameras, as well as the ubiquity of sensor-driven architectures. Owing to these capabilities, new aspects are coming to the forefront. This paper addresses the current state-of-the-art and provides researchers with an overview of existing surveillance solutions, analyzing their properties as a system and drawing attention to relevant challenges when developing, deploying and managing them. Also, some of the more prominent application domains are highlighted here. In an effort to understand the development of the advanced solutions, based on their most distinctive characteristics, we propose a taxonomy for surveillance systems to help classify them and reveal gaps in existing research. We conclude by identifying promising future research lines.

Selecting pivot features that connect a source domain to a target domain is an important first step in unsupervised domain adaptation (UDA). Although different strategies such as the frequency of a feature in a domain, mutual (or pointwise mutual) information have been proposed in prior work in domain adaptation (DA) for selecting pivots, a comparative study into (a) how the pivots selected using existing strategies differ, and (b) how the pivot selection strategy affects the performance of a target DA task remain unknown. In this paper, we perform a comparative study covering different strategies that use both labelled (available for the source domain only) as well as unlabelled (available for both the source and target domains) data for selecting pivots for UDA. Our experiments show that in most cases pivot selection strategies that use labelled data outperform their unlabelled counterparts, emphasising the importance of the source domain labelled data for UDA. Moreover, pointwise mutual information and frequency-based pivot selection strategies obtain the best performances in two state-of-the-art UDA methods.

Trust and reputation allow agents to make informed decisions about potential interactions. Trust in an agent is derived from direct experience with that agent, while reputation is determined by the experiences reported by other witness agents with potentially differing viewpoints. These experiences are typically aggregated in a trust and reputation model, of which there are several types that focus on different aspects. Such aspects include handling subjective perspectives of witnesses, dishonesty, or assessing the reputation of new agents. In this paper, we distil reputation systems into their fundamental aspects, discussing first how trust and reputation information is represented and second how it is disseminated among agents. Based on these discussions, a unifying abstraction is presented for trust and reputation systems, which is demonstrated by instantiating it with a broad range of reputation systems found in the literature. The abstraction is then instantiated to combine the range of capabilities of existing reputation systems in the Machine Learning Reputation System, which is evaluated using a marketplace simulation.

During the software system development lifecycle, different types of Unified Modeling Language model are developed to represent different views of the system. Sequence diagrams represent one of the most important types of model. They show how objects in the system interact to offer the functionality manifested in the form of use cases. As the complexity of the system being modeled increases, creating the sequence diagrams manually becomes harder. However, the problem of automatic sequence diagrams generation has not caught enough researchers yet. This paper presents an approach and a tool to automatically generate sequence diagrams from use cases and class diagrams. The problem of determining the sequence of message passing is treated as an Artificial Intelligence action planning problem and solved as such. In doing so, Design by Contract concepts are enforced in the specification of the given use cases and class diagrams. A special message-passing planner, Communiqué, was developed and implemented accordingly. The overall approach was empirically evaluated against sets of manually created sequence diagrams available in the literature. Communiqué was able to create sequence diagrams comparable to the manually developed ones. The paper also sheds the light on some directions for future work to advance the applicability of the approach.

Automated machine classification will play a vital role in the machine learning and data mining. It is probable that each classifier will work well on some data sets and not so well in others, increasing the evaluation significance. The performance of the learning models will intensely rely on upon the characteristics of the data sets. The previous outcomes recommend that overlapping between classes and the presence of noise has the most grounded impact on the performance of learning algorithm. The class overlap problem is a critical problem in which data samples appear as valid instances of more than one class which may be responsible for the presence of noise in data sets.

The objective of this paper is to comprehend better the data used as a part of machine learning problems so as to learn issues and to analyze the instances that are profoundly covered by utilizing new proposed overlap measures. The proposed overlap measures are Nearest Enemy Ratio, SubConcept Ratio, Likelihood Ratio and Soft Margin Ratio. To perform this experiment, we have created 438 binary classification data sets from real-world problems and computed the value of 12 data complexity metrics to find highly overlapped data sets. After that we apply measures to identify the overlapped instances and four noise filters to find the noisy instances. From results, we found that 60–80% overlapped instances are noisy instances in data sets by using four noise filters. We found that class overlap is a principal contributor to introduce class noise in data sets.

Several Apriori algorithm implementations for mining association rules have been proposed in the literature using the Hadoop-MapReduce framework and, more recently, Spark. However, none of the works have made a detailed assessment of its performance, for example, comparing it with other implementations in various characteristics of data sets. In this work, we present a review of the main algorithms proposed for Hadoop-MapReduce and compared their implementations in a single environment under several different situations. Moreover, these algorithms had their implementations adapted to Spark, and also compared under the same circumstances. Based on the results of the experiments, we present a framework for recommending the Apriori implementation most appropriate for solving a given problem, according to the data set characteristics and minimum required support. The results show that Spark implementations overcome Hadoop-MapReduce implementations at runtime in most experiments. However, there is no single implementation that is the best in all the evaluated situations.

Various ontology visualization tools using different visualization methods exist and new ones are being developed every year. The goal of this paper is to follow up on previous surveys with an updated classification of ontology visualization methods and a comprehensive survey of available tools. The tools are analyzed for the used visualization methods, interaction techniques and supported ontology constructs. It shows that most of the tools apply two-dimensional node-link visualizations with a focus on class hierarchies. Color and shape are used with little variation, support for constructs introduced with version 2 of the OWL Web Ontology Language is limited, and it often remains vague what tasks and use cases are supported by the visualizations. Major challenges are the limited maturity and usability of many of the tools as well as providing an overview of large ontologies while also showing details on demand. We see a high demand for a universal ontology visualization framework implementing a core set of visual and interactive features that can be extended and customized to respective use cases.

The World Wide Web has had a notable impact on a variety of epistemically relevant activities, many of which lie at the heart of the discipline of knowledge engineering. Systems like Wikipedia, for example, have altered our views regarding the acquisition of knowledge, while citizen science systems such as Galaxy Zoo have arguably transformed our approach to knowledge discovery. Other Web-based systems have highlighted the ways in which the human social environment can be used to support the development of intelligent systems, either by contributing to the provision of epistemic resources or by helping to shape the profile of machine learning. In the present paper, such systems are referred to as knowledge machines. In addition to providing an overview of the knowledge machine concept, the present paper reviews a number of issues that are associated with the scientific and philosophical study of knowledge machines. These include the potential impact of knowledge machines for the theory and practice of knowledge engineering, the role of social participation in the realization of knowledge-based processes, and the role of standardized, semantically enriched data formats in supporting the ad hoc assembly of special-purpose knowledge systems and knowledge processing pipelines.

Ontologies are constantly evolving as new requirements daily occur and the modeling choices of the past should be updated or adapted. Exploring this evolution will enhance the understanding, augmenting the exploitation potential of the available ontologies. However, recent research focuses mostly on detecting changes between ontology versions, overloading end-users with hundreds or even thousands of changes between ontology versions, making it impossible to explore this evolution. To this direction, in this paper, we present EvoRDF, a novel framework for exploring ontology evolution using provenance queries. Our approach uses a high-level language of changes and effectively answers queries about when a specific resource was introduced and how—by which change operations. Even more, why queries can identify the sequence of changes that led to the creation of a specific resource in the latest ontology version or track the evolution of a specific resource from a past ontology version. The evaluation performed shows the feasibility of our solution and the great advantages gained.

Nowadays travellers can benefit from the computing capabilities, collection of on board sensors and ubiquitous Internet access provided by mobile devices. These are the three pillars of any tourist support system since they provide the power, means and data to establish the local user context, to access remote services and to provide value-added user-centred context-aware applications. However, making sense of the user context data is not straightforward, as it requires dedicated knowledge acquisition and knowledge representation solutions. Besides, the range and diversity of available data sources is huge, requiring appropriate knowledge processing techniques to provide addequated tourism services. This article presents an updated review, and a comparison of recent context-aware tourism applications (CATA), including supporting technologies; and considering four possible dimensions: knowledge acquisition, knowledge representation, knowledge processing and knowledge-based services. We propose and apply a CATA analysis framework, contemplating these four dimensions to the applications found in the literature. This survey constitutes, not only, a state of the art review on tourism mobile applications, but, also, anticipates the latest development trends in tourism-related applications.

Reinforcement learning (RL) can be extremely effective in solving complex, real-world problems. However, injecting human knowledge into an RL agent may require extensive effort and expertise on the human designer’s part. To date, human factors are generally not considered in the development and evaluation of possible RL approaches. In this article, we set out to investigate how different methods for injecting human knowledge are applied, in practice, by human designers of varying levels of knowledge and skill. We perform the first empirical evaluation of several methods, including a newly proposed method named State Action Similarity Solutions (SASS) which is based on the notion of similarities in the agent’s state–action space. Through this human study, consisting of 51 human participants, we shed new light on the human factors that play a key role in RL. We find that the classical reward shaping technique seems to be the most natural method for most designers, both expert and non-expert, to speed up RL. However, we further find that our proposed method SASS can be effectively and efficiently combined with reward shaping, and provides a beneficial alternative to using only a single-speedup method with minimal human designer effort overhead.

This paper presents a new ‘Language Independent Recommender Agent’ (LIRA), using information distributed over any text-source pair on the Web about candidate items. While existing review-based recommendation systems learn the features of candidate items and users’ preferences, they do not handle varying perspectives of users on those features. LIRA constructs agents for each user, which run regression algorithms on texts from different sources and builds trust relations. The key advantages of LIRA can be listed as: LIRA does not require reviews from target users, LIRA calculates trust values based on prediction accuracy instead of social connections or rating similarity, LIRA does not require the reviews to come from the same community or peer user group. Since ratings of the reviewers are not necessary for LIRA, we can collect and use reviews from different sources (web pages, professional critiques), as long as we know the corresponding item and source of that text. Since LIRA does not combine text from different sources, texts from different sources are not required to be in the same language. LIRA can utilize text from multiple languages, as long as sources are consistent with their language usage.

The development of complex systems ensembles that operate in uncertain environments is a major challenge. The reason for this is that system designers are not able to fully specify the system during specification and development and before it is being deployed. Natural swarm systems enjoy similar characteristics, yet, being self-adaptive and being able to self-organize, these systems show beneficial emergent behaviour. Similar concepts can be extremely helpful for artificial systems, especially when it comes to multi-robot scenarios, which require such solution in order to be applicable to highly uncertain real world application. In this article, we present a comprehensive overview over state-of-the-art solutions in emergent systems, self-organization, self-adaptation, and robotics. We discuss these approaches in the light of a framework for multi-robot systems and identify similarities, differences missing links and open gaps that have to be addressed in order to make this framework possible.

Nowadays the composition and formation of effective teams is highly important for both companies to assure their competitiveness and for a wide range of emerging applications exploiting multiagent collaboration (e.g. crowdsourcing, human-agent collaborations). The aim of this article is to provide an integrative perspective on team composition, team formation, and their relationship with team performance. Thus, we review the contributions in both the computer science literature and the organizational psychology literature dealing with these topics. Our purpose is twofold. First, we aim at identifying the strengths and weaknesses of the contributions made by these two diverse bodies of research. Second, we aim at identifying cross-fertilization opportunities that help both disciplines benefit from one another. Given the volume of existing literature, our review is not intended to be exhaustive. Instead, we have preferred to focus on the most significant contributions in both fields together with recent contributions that break new ground to spur innovative research.

In this paper, we discuss the potential role of answer set programming (ASP) in the context of approaches to the development of agents and multi-agent systems especially in the realm of Computational Logic. After shortly recalling the main (computational-logic-based) agent-oriented frameworks, we introduce ASP; then, we discuss the usefulness of a potential integration of the two paradigms in a modular heterogeneous framework, and the feasibility of such integration. This also in the more general view of improving and empowering flexibility of agent-oriented frameworks. Relevant literature will be mentioned and discussed. Possible future directions and potential developments will be outlined.

Automated planning has been a continuous field of study since the 1960s, since the notion of accomplishing a task using an ordered set of actions resonates with almost every known activity domain. However, as we move from toy domains closer to the complex real world, these actions become increasingly difficult to codify. The reasons range from intense laborious effort, to intricacies so barely identifiable, that programming them is a challenge that presents itself much later in the process. In such domains, planners now leverage recent advancements in machine learning to learn action models, that is, blueprints of all the actions whose execution effectuates transitions in the system. This learning provides an opportunity for the evolution of the model toward a version more consistent and adapted to its environment, augmenting the probability of success of the plans. It is also a conscious effort to decrease laborious manual coding and increase quality. This paper presents a survey of the machine learning techniques applied for learning planning action models. It first describes the characteristics of learning systems. It then details the learning techniques that have been used in the literature during the past decades, and finally presents some open issues.