Despite the importance of the Amazon region due to its biodiversity, ecosystem services and its enormous contribution to reduce global warming, this region is currently facing critical threats and challenges such as deforestation, urban and agricultural expansion, massive forest fires, illegal/non-regulated mining, among others. Given its vast extension, timely monitoring aimed to mitigate these problems represents a complex task. The lack of adequate tools has hindered environmental monitoring and management in this region, highlighting the need to develop advanced techniques to address these issues. This study focuses on the implementation of methods to detect and classify land cover changes, using an portion of the Ecuadorian Amazon as a case study. Our proposed method combines spectral vegetation indices generated from Sentinel-2 satellite image and deep learning techniques. Multitemporal images have been collected and preprocessed, applying the Bitemporal Adapter Network (BAN) for change detection and ResNet152V2 for land cover classification. The BAN is then re-trained with a specific dataset for the Ecuadorian Amazon. Results attain a good level of accuracy (99.36 unchanged and 89.6 changed) showing that these techniques are effective not only for detecting changes, but also for classifying affected land cover types. These findings provide valuable information for the implementation of conservation and management policies in the Ecuadorian Amazon.

This article addresses the challenge of reproducing machine learning (ML) experiments by integrating processes based on MLOps and semantic technologies. The inherent complexity of experimentation in scientific research hinders reproducibility through conventional methods, which has led to the need to automate processes. In this work, a solution has been developed allowing the execution of ML experiments of other researchers and their reproducibility. The use of semantic technologies allows the complete description of the experiment, including the data and resources necessary for its execution. The approach proposed in this work contributes to the automation of the experimentation phases based on MLOps, demonstrating how it can be used to reproduce experiments and offer a solution to the complexity of experimentation in scientific research. The effectiveness of the solution proposed in this work is evaluated by means of a survey-based analysis carried out among researchers who currently use manual processes to perform machine learning experiments. The results indicate that manual processing is prone to errors and not scalable regarding the size and complexity of most experiments. Moreover, the solution proposed in this work, which combines MLOps-based processes and semantic technologies, has been well received by researchers and considered to significantly improve the efficiency, reproducibility, and scalability of machine learning experimentation.

Effective actions to mitigate air pollution require of availability of high-resolution observations. Low-cost sensor technologies have emerged as an affordable solution to cope with this deficiency. However, since low-cost sensors are built with low-cost materials, they are prone to errors, gaps, bias, and noise. These problems need to be solved before data can be used to support research or decision making. Addressing lack of reliability in low-cost sensor data is a complex challenge that is still under research over several lines (e.g. accuracy estimation of low-cost sensor data). Current approaches in this line involve modeling, bias-correction, and more recently, data fusion methods relying on high-resolution air quality computational models. Overall, accuracy estimation can be reduced to a modeling problem. The focus of this work is studying, testing, and comparing suitable approaches for handling point-referenced spatio-temporal sensor data, particularly classical spatial models, spatio-temporal models, and popular machine learning methods. Among these approaches, Bayesian hierarchical models have a special consideration given the attention they have drawn during the last fifteen years. The benchmark supporting this comparison study is a real-life dataset made up of daily ozone observations taken from the USA Environmental Protection Agency (EPA) and meteorological variables extracted from the NCEP/NCAR Reanalysis Project (NNRP). The main contributions of this work are: (1) a systematic comparison of three kinds of models, using a 10-fold cross-validation exercise; and (2) a feature engineering method to create covariates meant to harness spatially correlated observations of point-referenced sensor data.

Traditionally most of the steps involved in a Systematic Literature Review (SLR) process are manually executed, causing inconvenience of time and effort, given the massive amount of primary studies available online. This has motivated a lot of research focused on automating the process. Current state-of-the-art methods combine active learning methods and manual selection of primary studies from a smaller set so they can maximize the finding of relevant papers while at the same time minimizing the number of manually reviewed papers. In this work, we propose a novel strategy to further improve these methods whose early success heavily depends on an effective selection of initial papers to be read by researchers using a PCAbased method which combines different document representation and similarity metric approaches to cluster and rank the content within the corpus related to an enriched representation of research questions within the SLR protocol. Validation was carried out over four publicly available data sets corresponding to SLR studies from the Software Engineering domain. The proposed model proved to be more efficient than a BM25 baseline model as a mechanism to select the initial set of relevant primary studies within the top 100 rank, which makes it a promising method to bootstrap an active learning cycle.

We present a study of research trends for the area of Computer Sciences in Ecuador in recent years. This analysis was performed through a new method that leverages on semantic web technologies and external knowledge bases (i.e. DBpedia and UNESCO nomenclature) for identifying research topics within articles' metadata. This information takes into account the documents' publication date in order to construct time series which are analyzed and interpreted looking for trends. Concretely, we focused our study on the REDI (Semantic Repository of Ecuadorian Researchers) knowledge base which compiles most of the scholarly assets produced in Ecuador and more specifically on the Computer Science subset of publications. This study found that most of the research topics have shown an steady growth in the volume of publications over time, whereas the Semantic Web and E-Government research topics had a great impact initially and now have been slightly reducing its share in favor of new topics such as Information Integration, Machine Learning and Data Mining.

The indexing services are an important element of researching process because they make publicly available research results and articles for the community. Furthermore, regional indices such Latindex have contributed to spreading scientific works and incentivizing research in the Latin-American context. However, the un-centralized publication approach that its member journals follow has made impossible to form a unified view of Latindex-indexed articles and its corresponding journals. This drawback has limited activities such biblio-metric studies and integration from the perspective of information systems. In this paper, a linking mechanism between journals and publications is outlined, which aims to identify explicitly whether or not an arbitrary article belongs to Latindex. The proposed approach leverages on the Linked Data principles and takes advantage of web search engines to validate its results. This proposal has successfully been evaluated on an Ecuadorian publications dataset obtaining a 0.91 f-score respect to a manually classified sample.

Currently, we are witnessing an exponential growth in the amount of data being generated and captured at multiple locations. This trend will continue over the next years. Hence, we have envisioned a scenario in which many objects will be referencing to or generating location information. Thus, the need for appropriately managing geospatial data is evident. In this paper, we present our vision for an integral Geo Linked Data platform; pointing out the current limitations and challenges in the GeoRDFization, Storage, Query Federation, and Visualization of data with an inherent spatial context.

Data ambiguity from various sources remains as a complex problem that affects services provided by digital libraries. From the point of view of integration of information from different sources, the challenge of author ambiguity is one of the most important, and there are numerous methods proposed to deal with this issue using different approaches. They generally work for some scenarios but they have important limitations, specially when dealing with heterogeneous sources. In this work, we review a group of existing methods and then propose a technique that combines some of them, also incorporating a measure of distance using semantic technologies to solve the ambiguity of authors while integrating bibliographic data from various sources. This technique has been successfully tested in disambiguating Ecuadorian authors from both internal sources (institutional repositories) and external digital libraries.

This paper proposes a general framework that allows to identify a video in real time using perceptual image hashing algorithms. In order to evaluate the versatility and performance of the framework, it was coupled for a use case about ads tv monitoring. Four Perceptual Image Hashing (PIH) algorithms were subject to a benchmarking process in order to identify the best one for the use case. This process was focused on analyze differences in terms of discriminability (D), robustness (R), time processing (Tp) and efficiency (E). A truth table was used to obtain information about discriminability and robustness, while processing time was directly measured. An efficiency metric based on time processing and identification capacity was proposed. In general terms, DHASH and PHASH algorithms have higher identification capacities than AHASH and WHASH in order to identify a video using only one frame. Moreover, a progressive decrease in robustness with the increment of the Hamming distance is observed in all cases. However, in a specific case of tv monitoring where speed is critical, the processing time becomes the most discriminatory parameter for the selection of the algorithm. So, for this case, a particular type of PIH (Average Hash) is highlighted as the most efficient one among other techniques, reaching an accuracy of 100% and frame rates on processing average of 108 fps with a Hamming Distance of 1. At the end, the proposed framework has remarkable identification skills, and presents an efficient search. Furthermore, presents the steps to select the best algorithm and its more adequate parameters, according to the requirements of each particular case.

Interactive advertising based on multiple devices opens new possibilities for mobile applications, where users can search, select, or expand the information provided in advertising commercials by incorporating interactivity-friendly companion devices such as smart phones and tablets. In this paper, we derive the basic requirements for a flexible infrastructure that can support interactive ad applications. The infrastructure comprises a set of components and their externally visible properties, and the relationships between them. The main contributions of this work are first, establishing a desirable set of requirements for a suitable working scenario on which the different interactivity supporting systems people use in a regular basis can be easily integrated, and second, the definition of a platform which considers the different stages and requirements identified from different works in related areas.

In the domain of Digital Television (DTV) broadcasting technology, the enhancement of signals features over classic analog signal transmission allows increasing the amount of content available for TV viewers. Recommender Systems (RS) arose as a suitable choice to assist users in the overwhelming task of selecting audiovisual content, however, the cold-start problem normally associated to the lack of information in early RS stages, causes that user stereotyping approaches are employed meanwhile the lack of information in user profiles is overcome. This paper presents an experimental approach aimed to determine the best conditions for which users who were categorized within a determined stereotype during the cold-start stage, could migrate to a new state in which they receive personalized recommendations. Experimental results show that the best condition under the selected demographic stereotyping scheme for this transition is directly related to the number of TV programs that a user has rated while making use of the system.

Currently, institutions store the information that they produce in digital repositories, especially those within the academic scope, such as universities and libraries. However, the lack of agreement in defining protocols, meta-data descriptions and technologies for publication has caused issues while trying to integrate repositories. Numerous alternatives have been proposed to tackle this problem, however, most of them are based on syntactical technologies, which are exclusive to the ambit of digital repositories, hampering a higher level integration(Web). In addition, much of the research done so far uses centralized approaches, which are hardly scalable and can threaten to inter-institutional independence. In this work a new distributed integration architecture is proposed, which is based on semantic technologies and federated queries. This architecture has been successfully tested to integrate a group of institutional repositories belonging to universities, demonstrating its applicability within this context.

Electronic Program Guides (EPGs) describe broadcast programming information provided by TV stations. However, users may obtain more information when these guides have been enriched. The main contribution of this work is to present an automation process for EPG's information enrichment through the use of semantic technologies and external resources. Among the several resources involved in the enrichment process, the following can be mentioned : ontologies, web services, semantic repositories and natural language processing techniques.

In this paper the authors present an approach for the semantic annotation of RESTful services in the geospatial domain. Their approach automates some stages of the annotation process, by using a combination of resources and services: a cross-domain knowledge base like DBpedia, two domain ontologies like GeoNames and the WGS84 vocabulary, and suggestion and synonym services. The authors’ approach has been successfully evaluated with a set of geospatial RESTful services obtained from ProgrammableWeb.com, where geospatial services account for a third of the total amount of services available in this registry.

In this paper we present the process that has been followed for the development of an application that makes use of several heterogeneous Spanish public datasets that are related to three themes of INSPIRE Directive, specifically Administrative Units, Hydrography, and Statistical Units. Our application aims at analysing existing relations between the Spanish coastal area and different statistical variables such as population, unemployment, dwelling, industry, and building trade. Besides providing methodological guidelines for the generation, publishing and exploitation of Linked Data from such datasets, we provide an important innovation with respect to other similar processes followed in other initiatives by dealing with the geometrical information of features.