Collaborative tagging systems allow users to assign keywords - so called "tags" - to resources. Tags are used for navigation, finding resources and serendipitous browsing and thus provide an immediate benefit for users. These systems usually include tag recommendation mechanisms easing the process of finding good tags for a resource, but also consolidating the tag vocabulary across users. In practice, however, only very basic recommendation strategies are applied. In this paper we evaluate and compare several recommendation algorithms on large-scale real life datasets: an adaptation of user-based collaborative filtering, a graph-based recommender built on top of the FolkRank algorithm, and simple methods based on counting tag occurrences. We show that both FolkRank and collaborative filtering provide better results than non-personalized baseline methods. Moreover, since methods based on counting tag occurrences are computationally cheap, and thus usually preferable for real time scenarios, we discuss simple approaches for improving the performance of such methods. We show, how a simple recommender based on counting tags from users and resources can perform almost as good as the best recommender.
Linear Discriminant Analysis (LDA) is a very common technique for dimensionality reduction problems as a preprocessing step for machine learning and pattern classification applications. At the same time, it is usually used as a black box, but (sometimes) not well understood. The aim of this paper is to build a solid intuition for what is LDA, and how LDA works, thus enabling readers of all levels be able to get a better understanding of the LDA and to know how to apply this technique in different applications. The paper first gave the basic definitions and steps of how LDA technique works supported with visual explanations of these steps. Moreover, the two methods of computing the LDA space, i.e. class-dependent and class-independent methods, were explained in details. Then, in a step-by-step approach, two numerical examples are demonstrated to show how the LDA space can be calculated in case of the class-dependent and class-independent methods. Furthermore, two of the most common LDA problems (i.e. Small Sample Size (SSS) and non-linearity problems) were highlighted and illustrated, and state-of-the-art solutions to these problems were investigated and explained. Finally, a number of experiments was conducted with different datasets to (1) investigate the effect of the eigenvectors that used in the LDA space on the robustness of the extracted feature for the classification accuracy, and (2) to show when the SSS problem occurs and how it can be addressed.
Speaker state recognition is an important issue to understand the human behaviour and to achieve more comprehensive speech interactive systems, and therefore has received much attention in recent years. This work addresses the automatic classification of three types of child emotions in vocalisations: neutral mood, fussing (negative mood) and crying (negative mood). Speech, in a broad sense, contains a lot of para-linguistic information that can be revealed by means of different methods for feature extraction and, in this case, these would be useful for mood detection. Here, several set of features are proposed, combined and compared with state-of-art characteristics used for speech-related tasks, and these are based on spectral information, bio-inspired ear model, auditory sparse representations with dictionaries, optimised wavelet coefficients and optimised filter bank for cepstral representation. All the experiments were performed using the Extreme Learning Machines as classifier because it is a state-of-art classifier and to achieve comparable results. The results show that by means of the proposed feature extraction methods it is possible to improve the performance provided by the baseline features. Also, different combinations of the developed feature sets were studied in order to further exploit their properties.
The problem of Multi-Agent Path Finding (MAPF) is to find paths for a fixed set of agents from their current locations to some desired locations in such a way that the agents do not collide with each other. This problem has been extensively theoretically studied, frequently using an abstract model, that expects uniform durations of moving primitives and perfect synchronization of agents/robots. In this paper we study the question of how the abstract plans generated by existing MAPF algorithms perform in practice when executed on real robots, namely Ozobots. In particular, we use several abstract models of MAPF, including a robust version and a version that assumes turning of a robot, we translate the abstract plans to sequences of motion primitives executable on Ozobots, and we empirically compare the quality of plan execution (real makespan, the number of collisions).
Nowadays, society is in constant evolution, which allows constant production of new knowledge. In this way, citizens are constantly pressured to obtain new qualifications through training/requalification. The need for qualified people has been growing exponentially, which means that resources for education/training are limited to being used more efficiently. In this paper we will focus in the design the user model, so, we propose an innovative approach to design a user model that monitors the user's biometric behaviour by measuring their level of attention during e-learning activities. In addition, a machine learning catego-rization model is presented that oversees user activity during the session. We intend to use non-invasive methods of intelligent tutoring systems, observing the interaction of users during the session. Furthermore, this article highlights the main biometric behavioural variations for each activity and bases the set of attributes relevant to the development of machine learning classifiers to predict users' learning preference. The results show that there are still mechanisms that can be explored and improved to better understand the complex relationship between human behaviour, attention and evaluation that could be used to implement better learning strategies. These results can be decisive in improving ITS in e-learning environments and to predict user behaviour based on their interaction with technology devices.
The paper reviews methods on automatic annotation of texts with Wikipedia entries. The process, called Wikification aims at building references between concepts identified in the text and Wikipedia articles. Wikification finds many applications, especially in text representation, where it enables one to capture the semantic similarity of the documents. Also, it can be considered as automatic tagging of the text. We describe typical approaches to Wikification, and identify their advantages and disadvantages. The main problem for wide usage of the Wikification method is the lack of open-sourced frameworks that enable researchers to work cooperatively on that problem. Also problematic is the lack of a unified platform for evaluation of the results proposed by different approaches.
Time is pervasive of the human way of approaching reality, so that it has been widely studied in many research areas, including AI and relational Temporal Databases (TDB). While temporally imprecise information has been widely studied by the AI community, only few approaches have faced temporal indeterminacy (in particular, "don't know exactly when" indeterminacy) in TDBs. Indeed, as we will show in this paper, the treatment of time in general, and of temporal indeterminacy in particular, involves the introduction of implicit forms of data representation in TDBs. As a consequence, we propose a new AI-style methodology to cope with temporal indeterminacy in TDBs. Specifically, we show that typical AI notions and techniques, such as making explicit the semantics of the representation formalism, and adopting symbolic manipulation techniques based on such a semantics, can be fruitfully exploited in the development of a "principled" treatment of indeterminate time in relational databases.
Many creative methods, such as different types of brainstorming, are based on the collaboration among a set of persons. The collaboration follows some well established workflows, which could be formalized. This would allow the generation of computational models that can be implemented to make some tools that facilitate the enactment of creative processes, or the simulation for the analysis of their characteristics. This work shows how to model this kind of collaborative creative processes as multi-agent systems, by representing the participants as interacting agents in well-defined workflows. This is done with the INGENIAS modeling language and tools, which also support rapid prototyping using the JADE agent platform. A concrete creative method, Symbolic Brainstorming, is used to illustrate and validate the feasibility of the approach.
Multidisciplinary Design Optimization (MDO) is a computational approach for optimizing design of a complex system of systems that require knowledge from multiple disciplines. In a former study, we explored and found that the individual discipline feasible (IDF), a type of MDO design technique, performed well in several benchmark test cases of decentralized Reinforcement Learning (RL) problems, in particular, stabilizing an unknown system. However, the earlier study was not able to resolve as to why the overall system of systems, even with strongly coupled systems, could be stabilized when each agent just focused on stabilizing itself. In this work, we make significant extension in resolving this behavior by conducting a theoretical analysis of the MDO solution of RL problems. Through the analysis, we show that with the proper control law, each MDO agent should be able to bring its state closer to the 0-stable point regardless of how the other agents' states impact the state of the whole system. This is the main reason why the 'selfish' MDO-IDF agents are successful in learning to stabilize the overall system. The simulation results, including benchmark test cases, verify our analysis. Therefore, we propose that the MDO would be a promising solution in many other decentralized RL problems.
Recent advancements in web personalization techniques facilitate enhanced web-based services that allow recommender systems (RSs) to incorporate contextual knowledge about users and items as an additional dimension into recommendation process. Context-awareness is one of the important aspects of ubiquitous computing to support cognitive environment and provide services in various e-commerce recommendation applications. Tracking each user's preferences over various contextual dimensions from their past transactions and providing personalized recommendations to them are the essence of context-aware recommender systems (CARSs). Conventional paradigms for incorporating context in recommendation process cannot fully cover the challenges on several levels of a context-aware system. Our proposed scheme is based on the hybridization of two complementary techniques, collaborative filtering (CF) and reclusive method (RM) to make context valuable at each level of users' preferences and improve predictive capability of CARSs. Further, a fuzzy real-coded genetic algorithm (Fuzzy-RCGA) approach is incorporated for identifying the influential contextual situations and handling the uncertainty of users' preferences under various contextual situations. Furthermore, users' demographic features are utilized for alleviating the problem of data sparsity. The empirical results on two real-world benchmark datasets clearly demonstrate the effectiveness of our proposed schemes for CARS framework.
A significant percentage of urban traffic is caused by the search for parking spots. One possible approach to improve this situation is to guide drivers along routes which are likely to have free parking spots. The task of finding such a route can be modeled as a probabilistic graph problem which is NP-complete. Thus, we propose heuristic approaches for solving this problem and evaluate them experimentally. For this, we use probabilities of finding a parking spot, which are based on publicly available empirical data from TomTom International B.V. Additionally, we propose a heuristic that relies exclusively on conventional road attributes. Our experiments show that this algorithm comes close to the baseline by a factor of 1.3 in our cost measure. Last, we complement our experiments with results from a field study, comparing the success rates of our algorithms against real human drivers.
Recent improvements in deep learning techniques show that deep models can extract more meaningful data directly from raw signals than conventional parametrization techniques, making it possible to avoid specific feature extraction in the area of pattern recognition, especially for Computer Vision or Speech tasks. In this work, we directly use raw text line images by feeding them to Convolutional Neural Networks and deep Multilayer Perceptrons for feature extraction in a Handwriting Recognition system. The proposed recognition system, based on Hidden Markov Models that are hybridized with Neural Networks, has been tested with the IAM Database, achieving a considerable improvement.
The rapidly increasing deployment of AI raises societal issues about its safety, reliability, robustness, fairness and moral integrity. This paper reports on a declaration intended as a code of conduct for AI researchers and application developers. It came out of a workshop held in Barcelona in 2017 and was discussed further in various follow up meetings, workshops, and AI schools. The present publication is a matter of historical record and a way to publicize the declaration so that more AI researchers and developers can get to know it and that policy makers and industry leaders can use it as input for governance. It also discusses the rationale behind the declaration in order to stimulate further debates.
The extraction of the relevant and debated opinions from online social media and commercial websites is an emerging task in the opinion mining research field. Its growing relevance is mainly due to the impact of exploiting such techniques in different application domains from social science analysis to personal advertising. In this paper, we present SMACk, our opinion summary system built on top of an argumentation framework with the aim to exchange, communicate and resolve possibly conflicting viewpoints. SMACk allows the user to extract debated opinions from a set of documents containing user-generated content from online commercial websites, and to automatically identify the mostly debated positive aspects of the issue of the debate, as well as the mostly debated negative ones. The key advantage of such a framework is the combination of different methods, i.e., formal argumentation theory and natural language processing, to support users in making more informed decisions, e.g., in the context of online purchases.
We present the interactive assistant ROBERT that provides situation-adaptive support in the realisation of do-it-yourself (DIY) home improvement projects. ROBERT assists its users by providing comprehensive step-by-step instructions for completing the DIY project. Each instruction is illustrated with detailed graphics, written and spoken text, as well as with videos. They explain how the steps of the project have to be prepared and assembled and give precise instructions on how to operate the required electric devices. The step-by-step instructions are generated by a hierarchical planner, which enables ROBERT to adapt to a multitude of environments easily. Parts of the underlying model are derived from an ontology storing information about the available devices and resources. A dialogue manager capable of natural language interaction is responsible for hands-free interaction. We explain the required background technology and present preliminary results of an empirical evaluation.
Minimally Unsatisfiable Subformulas (MUS) find a wide range of practical applications, including product configuration, knowledge-based validation, and hardware and software design and verification. MUSes also find application in recent Maximum Satisfiability algorithms and in CNF formula redundancy removal. Besides direct applications in Propositional Logic, algorithms for MUS extraction have been applied to more expressive logics. This paper proposes two algorithms for computation of MUSes of propositional formulas in Conjunctive Normal Form (CNF). The first algorithm is optimal in its class, meaning that it requires the smallest number of calls to a SAT solver. The second algorithm extends earlier work, but implements a number of new techniques. Among these, this paper analyzes in detail the technique of recursive model rotation, which provides significant performance gains in practice. Experimental results, obtained on representative practical benchmarks, indicate that the new algorithms achieve significant performance gains with respect to state of the art MUS extraction algorithms.
We present a new collaborative visual storytelling platform, Aesop, for direction and animation. Our system operates in two main modes, common sense grounding (annotation) and conversation. The Aesop system senses the human state and input using a natural language parser and human gesture monitoring for natural interactions. The interface consists of a 3D animation software and a web controller to interact with the internal state of the system. For knowledge representation, we formulate novel knowledge graphs which enable spatio-temporal event representation. Aesop thus enables 3D spatial and temporal reasoning which are both essential for storytelling. Finally, the system utilizes a dialog manager to track the conversation state and manage goals. Aesop provides a rich platform that enables research in language, gestures, vision, and planning in the context of storytelling.
Personalization is pervasive in the online space as it leads to higher efficiency for the user and higher revenue for the platform by individualizing the most relevant content for each user. However, recent studies suggest that such personalization can learn and propagate systemic biases and polarize opinions; this has led to calls for regulatory mechanisms and algorithms that are constrained to combat bias and the resulting echo-chamber effect. We present our balanced news feed via a demo that displays a dashboard through which users can view the political leaning of their news consumption and set their polarization constraints. The balanced feed, as generated by the user-defined constraints, is then showcased side-by-side with the unconstrained (polarized) feed for comparison.