Controlling the healthfulness of breeding sows is one of the most critical factors in pig breeding farms since, throughout the production process, the sows show continuous changes in their general condition due to motherhood. However, the existing approaches for estimating sows body condition are mostly manual and of low precision, which limit its automated calculation and subsequent management. To contribute to this area, we propose an IoT Information System for measuring and controlling the degree of sows body reserves based on calculations that use two actual physical parameters simultaneously: Live Weight and Back Fat Thickness of breeding sows, gathered from a smart-scale. The IoT system architecture is composed of three layers: (i) the perception layer conformed by IoT devices connected directly to an End Point; (ii) the network layer in charge of gathering the row data from the perception layer and transmitting them in JSON format to the application layer; and (iii) the application layer in which the data are processed and stored, as the base to develop a variety of services. We describe in detail the architecture of the framework and demonstrate its functionality and suitability with the implementation of a proof-of-concept that includes top services to allow users to visualize and evaluate the progress of the health of breeding sows in terms of Degree of Body Reserves. This experience allows extracting the current limitations of existing solutions and some lessons learned towards an IoT Information System that provides added values to the swine farms sector and serves as

International audience

Distributed systems, spanning pervasive, cloud, and edge computing, require robust scheduling for tasks and data management, facing challenges like hardware diversity, QoS, localization, and energy efficiency. Current solutions often lack a comprehensive approach, failing to achieve balanced load management, QoS, and notably in minimizing application energy use, a key modern concern. We propose a distributed scheduler using multidimensional spaces and data structures aimed at comprehensive load balancing, energy efficiency, and QoS. The scheduler indexes devices within the structure based on diverse criteria such as hardware availability, GPS, and energy features. It then organizes applications into abstract graphs, where nodes are containers and their data items (units of data accessed by the containers), and edges are the transfer rates among them. To migrate containers, the scheduler performs range queries within the structure for optimal devices. To organize data items, it finds the barycenter of the devices executing the containers linked to them. We evaluated our scheduler’s effectiveness using the PISCO simulator, comparing it against other energy-efficient schedulers.

In this paper, a task model based on HAMSTERS-XL notation is proposed to represent activities of daily living (ADL). For an efficient representation of ADL, it is also a need to delimit requirements in time and location for each ADL, as well as the sensor data that define the ADL. In this sense, it is also proposed a procedure consisting of a set of step intending to delimit such time and location requirements, regarding both activities and sensors; alongside the events that need to be identified. The presented approach aims at providing an analysis tool regarding the performance of elderly/handicapped people by comparing data recovered from sensors within a smart home environment simulator to the task models. The feed oneself ADL (as well as its possible variations) is picked in order to evaluate our proposal. After proper analysis was carried out, anomalies detected during ADL performance are pointed out in order to detect possible ADL routine modification in a coherent manner through improved task models.

User-centric applications have recently gained popularity in particular in the tourism domain, in order to satisfy the individual needs of users and to provide personalized information. In turn, there is a need on modeling user profiles, considering different aspects of the information related to the user himself and his context. However, there is a lack of standardization to represent such information. The Semantic Web seems to be a clear solution for the formal representation of this knowledge, due to its capacity for organization and reasoning, in particular through ontologies. There are works that propose ontologies to model the user profile, but only cover partial aspects of the information required and are only applicable to specific applications, they do not propose a generalized user profile model applicable to the domain of tourism. This work proposes the development of GENTE ontology, a GENeral ontology for Tourism Environments, that represents the different dimensions of the information related to users and their context. In addition, techniques to infer characteristics, preferences, interests, and behaviors of users, from their social networks are proposed and developed.

In robotics, object detection in images or videos, obtained in real-time from sensors of robots can be used to support the implementation of service robot tasks (e.g., navigation, model its social behavior, recognize objects in a specific domain), usually accomplished in indoor environments. However, traditional deep learning based object detection techniques present limitations in such indoor environments, specifically related to the detection of small objects and the management of high density of multiple objects. Coupled with these limitations, for specific domains (e.g., hospitals, museums), it is important that the robot, apart from detecting objects, extracts and knows information of the targeted objects. Ontologies, as a part of the Semantic Web, are presented as a feasible option to formally represent the information related to the objects of a particular domain. In this context, this work proposes an object detection and recognition process based on a Deep Learning algorithm, object descriptors, and an ontology. ODROM, an Object Detection and Recognition algorithm supported by Ontologies and applied to Museums, is an implementation to validate the proposal. Experiments show that the usage of ontologies is a good way of desambiguating the detection, obtained with a and$\mathbf{mAP}{@}0.5=0.88$and a$\mathbf{mAP}{@}[0.5:0.95]=61\%$.

Human emotion recognition from visual expressions is an important research area in computer vision and machine learning owing to its significant scientific and commercial potential. Since visual expressions can be captured from different modalities (e.g., face expressions, body posture, hands pose), multi-modal methods are becoming popular for analyzing human reactions. In contexts in which human emotion detection is performed to associate emotions to certain events or objects to support decision making or for further analysis, it is useful to keep this information in semantic repositories, which offers a wide range of possibilities for implementing smart applications. We propose a multi-modal method for human emotion recognition and an ontology-based approach to store the classification results in EMONTO, an extensible ontology to model emotions. The multi-modal method analyzes facial expressions, body gestures, and features from the body and the environment to determine an emotional state; this processes each modality with a specialized deep learning model and applying a fusion method. Our fusion method, called EmbraceNet+, consists of a branched architecture that integrates the EmbraceNet fusion method with other ones. We experimentally evaluate our multi-modal method on an adaptation of the EMOTIC dataset. Results show that our method outperforms the single-modal methods.

With the robotics development, social robots interact with people, demanding they model the human being behavior to increase social navigation, considering proxemic spaces. However, human proxemic preferences can change in function of different social restrictions (e.g., culture, gender, local, the environment). Thus, robots should consider all these aspects to tailor their navigation. Towards an adaptable social navigation, in this article we develop the GProxemic Navigation system that allows identifying the robot localization in a geo-referenced map, with semantic annotations related to social restrictions, in function of which they chose the correct proxemic spaces they most respect in their autonomous navigation process. Results show that the GProxemic Navigation system efficiently obeys the proxemic space through the semantic annotations received.

Best Student Paper Award

Smart environments are currently overpowering traditional Human-Computer Interaction (HCI) approaches for people's everyday life applications. Proxemic interaction is an emerging area for improving HCI experiences in such environments, causing the so-called proxemic environments to emerge. Proxemic interaction describes how the five dimensions (i.e., Distance, Identity, Location, Movement, and Orientation - DILMO) can be used to define interactions among people and digital devices. Current studies in this area are focused on developing proxemic applications with specific functionalities, based on a toolkit that allows developers to obtain DILMO information from sensors. However, there exists a notable lack of general approaches able to support the whole implementation process, starting from the modelling of proxemic environments to represent general proxemics behaviours, and finalizing with the development of specific applications. In order to facilitate the integration of proxemic capabilities in HCI, we propose an approach for modelling proxemic environments based on a graphical Domain-Specific Language (DSL). The DSL allows non-specialist designers to express proxemic interactions for modelling proxemic environments and supports the development process. In order to show its suitability and functionality, we have implemented a prototype to design proxemic behaviours in different proxemic environments.

The widespread diffusion of smart and mobile devices continuously connected to the Internet has facilitated the users' contact and interaction with other people, devices, and with their physical surroundings.Proxemic interaction, derived from proxemics theory, focuses on how Human-Computer Interaction (HCI) works with smart devices, using the five proxemic dimensions: Distance, Identity, Location, Movement, and Orientation (DILMO).The current tools for developing proxemic applications require fixed devices that make it difficult to build proxemic mobile apps.In this work, we propose a framework to manage all components in a proxemic environment (i.e., interaction objects and DILMO dimensions that govern the HCI).We demonstrate and evaluate the effectiveness and suitability of our framework, through the development of two proxemic mobile applications, as proof-of-concept.

Representation of the knowledge related to any domain with flexible and well-defined models, such as ontologies, provides the base to develop efficient and interoperable solutions. Hence, a proliferation of ontologies in many domains is unleashed. It is necessary to define how to compare such ontologies to decide which one is the most suitable for specific needs of users/developers. Since the emerging developing of ontologies, several studies have proposed criteria to evaluate them. Nevertheless, there is still a lack of practical and reproducible guidelines to drive a comparative evaluation of ontologies as a systematic process. In this paper, we propose a methodological process to qualitatively and quantitatively compare ontologies at Lexical, Structural, and Domain Knowledge levels, considering Correctness and Quality perspectives. Since the evaluation methods of our proposal are based in a golden-standard, it can be customized to compare ontologies in any domain. To show the suitability of our proposal, we apply our methodological approach to conduct a comparative study of ontologies in the robotic domain, in particularly for the Simultaneous Localization and Mapping (SLAM) problem. With this study case, we demonstrate that with this methodological comparative process, we are able to identify the strengths and weaknesses of ontologies, as well as the gaps still needed to fill in the target domain (SLAM for our study case).

Twitter is one of the most popular information source available on the Web. Thus, there exist many studies focused on analyzing the credibility of the shared information. Most proposals use either Twitter API or web scraping to extract the data to perform such analysis. Both extraction techniques have advantages and disadvantages. In this work, we present a study to evaluate their performance and behavior. The motivation for this research comes from the necessity to know ways to extract online information in order to analyze in real-time the credibility of the content posted on the Web. To do so, we develop a framework which offers both alternatives of data extraction and implements a previously proposed credibility model. Our framework is implemented as a Google Chrome extension able to analyze tweets in real-time. Results report that both methods produce identical credibility values, when a robust normalization process is applied to the text (i.e., tweet). Moreover, concerning the time performance, web scraping is faster than Twitter API, and it is more flexible in terms of obtaining data; however, web scraping is very sensitive to website changes.

Currently, mobile technologies are present in our daily day in different activities and their use keeps increasing. In this context, proxemic interaction, derived from proxemic theory, is becoming an influential approach to implement Mobile Human-Computer Interaction (MobileHCI) in smart environments, based on five proxemic dimensions: Distance, Identity, Location, Movement, and Orientation (DILMO). The existing tools for implementing proxemic applications require fixed devices that make it difficult to built mobile proxemic apps. This work aims to propose a framework for the development of proxemic applications for smart environments comprised by entities, whose interactions are supported by MobileHCI defined in terms of DILMO dimensions. Our proposed framework allows defining and managing all components in a smart proxemic environment. The framework also provides an API, that allows developers to simplify the process of proxemic information sensing (i.e., detection of DILMO dimensions) from mobile phones and wearable sensors. We demonstrate and evaluate the effectiveness and suitability of our framework, through the proof-of-concept which describes the implementation of two proxemic mobile applications built in a context-based infrastructure for smart proxemic environments based on mobile devices.

Currently, people's daily lives are affected by the pandemic produced by the COVID-19. One of the main problem is the quickly and easy spreading of the virus. Healthcare workers are affected by nosocomial infections (also called as hospital-acquired infections) that exist in workplaces and more specifically, from health care equipment. In practice, the use of technology is quite common in health care settings. However, due to the touchability of mobile digital devices, their use contributes to nosocomial infections, according to several studies. Some applications based on tracking people have been implemented in order to facilitate Human-Computer Interaction (HCI) and preventing contamination of surfaces by people's hands. Notwithstanding, their use still presents limitations related to implementing applications to be used in some hospital environments, such as care rooms, laboratories, clinical workrooms. To overcome these limitations, we propose the use of interpersonal distances and proxemic dimensions (i.e., Distance, Identity, Location, Movement, and orientation- DILMO) for implementing HCI with mobile devices that reduces their touchability. The aim is to facilitate the development of mobile apps with proxemic HCI, supported in a proposed architecture, to stop spreading of nosocomial infection of COVID-19 and others. To show the usability and suitability of our proposal, we present two prototypes of apps for mobile devices as proof-of concept, using several combination of proxemic DILMO dimensions to model proxemic HCI that allow flexibility in interpersonal and devices people interactions.

Summary The ACID transaction model has played a cornerstone role in service composition to guarantee that composite services (CSs) have transactional support and consistent outcomes even in presence of failures. However, the classical ACID properties are too restrictive for independent and multi‐proprietors services running on distributed environments, such clouds. Transactional properties allow a relaxed atomicity and isolation, providing an “all‐or‐(almost)nothing” model. In previous works, we proposed a model to relax atomicity, calledfuzzyatomicity, according user requirements (acceptable fuzzy atomicity). In this article, we extend that model and propose an approach to measure thefuzzyatomicity. Our model allows to self‐adapt the CS execution, taking into account the state of the CS execution and user preferences (ie, theacceptable fuzzy atomicityexpressed in the user requirements). Thisfuzzyatomicity measure is applied in a self‐adaptive CS execution model based on the transactional properties (pivot, compensatable, and retriable) of its component services and on a checkpointing mechanism. It is suitable to cloud computing, which offers cloud services on demand and in which CS execution should preserve the self‐organizing and self‐adaptivity properties of such environment, specially in the presence of failures. We show how it is possible to relax the retriable property in CSs execution, based on ourfuzzyatomicity model. Additionally, in this work, we present a comparative analysis of the most recent works in the context of ACID properties relaxation for CS.

In this paper, we propose a framework to study the AGGIR (Autonomy Gerontology Iso-Resources Groups) grid model, in order to evaluate the level of independency of elderly people, according to their capabilities of performing activities and interact with their environments over the time. To model the Activities of Daily Living (ADL), we also extend a previously proposed Domain Specific Language (DSL), in order to employ operators to deal with constraints related to time and location of activities, and event recognition. Our framework aims at providing an analysis tool regarding the performance of elderly/handicapped people within a home environment by means of data recovered from sensors using the iCASA simulator. To evaluate our approach, we pick three of the AGGIR variables (i.e., dressing, toileting, and transfers) and evaluate their testability in many scenarios, by means of records representing the occurrence of activities of the elderly. Results demonstrate the accuracy of our framework to manage the obtained records correctly and thus generate the appropriate event information.

In the context of RDF document matching/integration, the datatype information, which is related to literal objects, is an important aspect to be analyzed in order to better determine similar RDF documents. In this paper, we present an R DF D atatype in F erring F ramework, called RDF-F, which provides two independent datatype inference processes: 1) a four-step process consisting of (i) a predicate information analysis (i.e., deduce the datatype from existing range property), (ii) an analysis of the object value itself by a pattern-matching process (i.e., recognize the object lexical space), (iii) a semantic analysis of the predicate name and its context, and (iv) generalization of Numeric and Binary datatypes to ensure the integration; and 2) a non-ambiguous lexical-space-matching process, where literal values are inferred by the modification of their representation, following new lexical spaces. We evaluated the performance and the accuracy of both processes with datasets from DBpedia. Results show that the execution time of both indicators is linear and their accuracy can increase up to 97.10 and 99.30%, respectively.

Multimedia Sensor Networks (MSNs) have gained increasing attention in recent years from both academic and industrial sectors. Unlike scalar sensor networks, the data collected from MSNs are enriched with multimedia data which can be used for defining and detecting complex and more application-meaningful events. However, to do so, there are several processing tasks need to be executed such as multimedia data decoding, translating semantic information from multimedia data, and integrating multimedia data from several sensors. Combining these tasks into one single generic framework so to process and detect complex events in MSNs is of great interest. However, developing such a framework is challenging due to the infrastructure of MSNs (which includes heterogeneous sensors) and types of multimedia data (which are diverse). Also, events in MSNs are needed to be detected in a near real-time manner. In this study, we propose an ontology-based framework to support complex event modeling and detecting in MSNs. Our framework helps users model MSNs infrastructure, complex events, and data collected from MSNs. It is also able of translating semantic formation, detecting, and reporting the events in a near real-time manner. Our framework is validated by means of prototyping and simulation. The results show that it can detect complex multimedia events in a high-work load scenario with average detection latency for less than 625 milliseconds.

Multi-Agent Systems (MASs) have received much attention in recent years because of their advantages on modeling complex distributed systems, such Digital Ecosystems (DESs). Many existing modeling languages that support the design of such systems are based on ontologies to assist the representation of agents knowledge. However, in the context of DESs, there is still a need for more general conceptual models to represent the specific characteristics of DESs in terms of win-win interaction, engagement, equilibrium, and self-organization. Then, concepts such behavior, roles, rules, and environment are needed. This paper describes an ontology-based approach by proposing MAS2DES-Onto, as the conceptual model, which considers the essential static and dynamic aspects of MASs by a clear representation of their concepts and relationships to support the design and development of DESs. To validate and conduct experimental tests, we integrate MAS2DES-Onto into a framework to automatically generate MAS-based DESs. Results show the efficiency and effectiveness of our approach.

Analytical data management applications, affected by the explosion of the amount of generated data in the context of Big Data, are shifting away their analytical databases towards a vast landscape of architectural solutions combining storage techniques, programming models, languages, and tools. To support users in the hard task of deciding which Big Data solution is the most appropriate according to their specific requirements, we propose a generic architecture to classify analytical approaches. We also establish a classification of the existing query languages, based on the facilities provided to access the big data architectures. Moreover, to evaluate different solutions, we propose a set of criteria of comparison, such as OLAP support, scalability, and fault tolerance support. We classify different existing Big Data analytics solutions according to our proposed generic architecture and qualitatively evaluate them in terms of the criteria of comparison. We illustrate how ou r proposed generic architecture can be used to decide which Big Data analytic approach is suitable in the context of several use cases.

Nowadays, most of the information exchanged in Internet (particularly through social networks) is in the form of multimedia data. Each actor in Internet (individuals, enterprises, territorial communities, etc.) becomes producer and consumer of contents. Development and increasing demand for multimedia data offer producers/consumers more choices and opportunities for sharing in collaborative environments. Nevertheless, social network platforms and other traditional collaborative environments present limitations regarding content selection and collection, categorization and classification, aggregation, linking and interoperability, usage control and privacy. In this context, this ongoing work proposes a MultiMedia-Oriented Digital EcoSystem (MMDES) as a new environment of collaboration, sharing, and computing of multimedia content and applications ensuring the benefits of all its participants. In this paper, we formally define MMDES components, objectives and rules through a dedicated ontology. We also present here a P2P architectural design for MMDES.

Similarity search in metric spaces refers to searching elements in data repositories that are similar to an element supplied by the user (query example). Similarity functions are used to determine which elements in the data repositories are similar to the query example and indexing mechanisms are used to improve the efficiency in the search. Classic indexation mechanisms such as LSH, M-Index, and M-Tree behave different according to the dimensionality in the metric space, volume of data repositories, and query strategies. In this paper, we describe SimSearch, a modular and flexible framework for similarity search in metric spaces, which allows to use, analyse, compare, and add several indexation mechanisms, search approaches, and query strategies. SimSearch allows doing queries given one or more example elements to obtain the set of elements more similar to the query examples, using query composition and Skyline. We show the variability of performance of several indexation mechanisms, including LSH-ML (our proposed variant of LSH), with experimental study in the domain of images represented by a feature vector in a high dimensionality metric space and Web Services represented by a vector with the values of Quality of Service (QoS) parameters.

In this work, we propose the implementation of a Distributed Platform for Massively Emails distribution, called OPTIMUS (OPTImized Mail distribUtion System). It allows efficient workload distribution among multiples email servers, optimizes the use of available resources, and provides fault tolerance in a robust distribution platform. OPTIMUS architecture extends the Postfix server implementation to incorporate data base management, workload balance, and fault tolerance. It is composed by several extended Postfix servers, connected in a Peer-to-Peer platform. Experimental results show that with a few amount of emails, Postfix (with its original centralized implementation) and OPTIMUS have similar performance. However, with huge amount of emails, OPTIMUS highly overcomes the performance of the original centralized Postfix version.

Dynamic load balancing on a cluster computing consists on evenly dividing at any moment of time the workload to be distributed among the nodes, in order to avoid load unbalance. Load unbalance can be resolved if nodes can migrate some of their work. Using a stability criterion to allow the analysis of stability in the cluster is a way to decide the migration tasks. The Routh-Hurwitz criterion applied to the equation feature allows knowing whether a system is stable or not. The aim of this work is to develop a mathematical model based on linear differential equations representing the load, execution, and migration of task on a cluster computing. With the model, it is possible to determine the stability of the cluster. If the workload on the nodes tends to their equilibrium, new migration tasks do not need to check the stability of the nodes to be executed.

During the execution of a Composite Web Service (CWS), different faults may occur and cause a Web Service (WS) to fail. To repair failures some strategies can be applied, such as WS retry or substitution, compensation of the performed execution, roll-back, replication, or take checkpoints to later restart the execution. Each strategy has advantages and disadvantages on different execution scenarios and can produce different impact on the CWS QoS, depending on the execution environment and execution state at the moment of the failure. In this paper we propose a model and show experimental results to dynamically decide which recovery strategy is the best choice in terms of the impact on the CWS QoS.

Web Services (WSs) that provide transactional properties are useful to guarantee reliable Composite WSs (CWSs) execution. In this paper, we propose a framework for efficient, fault tolerant, and correct distributed execution of Transactional CWSs (TCWSs). Our framework relies on WSs replacement and on a compensation protocol to support forward and backward recovery. We represent a TCWS and its corresponding backward recovery process by Colored Petri-Nets (CPNs) and, to ensure correct execution and compensation flows, unfolding processes of the CPNs are followed. We formalize the TCWS execution and recovery processes based on CPN properties. We also present the framework architecture and execution and recovery algorithms.

QoS parameters are used to describe services in terms of their behavior and can be used to rank services according to non-functional criteria. To provide an accurate characterization of the quality of a service, we propose a sampling-based technique. The proposed technique uses Adaptive and Sequential Sampling strategies to estimate the QoS parameters that satisfy the required confidence levels while the size of the sample remains small. QoS estimates are used by a hybrid composer, named PT-SAM, to identify the service compositions that satisfy a functional condition and best meet non-functional criteria of a user query. PT-SAM adapts a Petri-Net unfolding algorithm to find a desired marking from an initial state by using a utility function defined on QoS estimates and functional properties of the available services. PT-SAM uses a QoS-based utility function to guide the search into portions of good quality service compositions; thus, PT-SAM is able to scale up to large-scale search spaces of services. We report on the quality of the sampling techniques and the performance of the composer. First, we show correlation between the estimates and the real values of the QoS parameters; then, we report on the benefits of using these estimates to traverse large search spaces of service compositions (e.g., in the range of 1,000 to 100,000 services). Our experiments show that the quality of the compositions identified by our algorithm is close to the optimal solution produced by the exhaustive algorithm.

Distributed genetic algorithm (DGA) is one of the most promising choices among the optimization methods. In this paper we describe DGAFrame, alpha flexible framework for evolutionary computation, written in Java. DGAFrame executes GAs across a range of machines communicating through RMI network technology, allowing the implementation of portable, flexible GAs that use the island model approach. Each island can be configured independently from others providing the implementation of heterogeneous DGAs. To evaluate the performance of DGAFrame, we implemented the protein structure prediction problem and compare the DGA execution to its sequential counterpart through quality of solution. We also measure the computation to communication ratio and results show that the proposals consistently outperform equivalent sequential GAs.

We define a framework for remote service installation, called SIMG (service installation metaservice on grids). Its main goal is to allow grid users to define and safely install their own data filters on remote data servers. Once installed, these filters may be used like the predefined services and may be available for a broader community of users. SIMG is implemented in SUMA/G, a Globus-enabled platform for remote execution of Java bytecode. We also present our experiences using SIMG for defining and remotely installing two complementary services in a content-based image retrieval system.