Changes, planned or unexpected, are common during the execution of real-life processes. Detecting these changes is a must for optimizing the performance of organizations running such processes. Most of the algorithms present in the state-of-the-art focus on the detection of sudden changes, leaving aside other types of changes. In this article, we will focus on the automatic detection of gradual drifts, a special type of change, in which the cases of two models overlap during a period of time. The proposed algorithm relies on conformance checking metrics to carry out the automatic detection of the changes, performing also a fully automatic classification of these changes into sudden or gradual. The approach has been validated with a synthetic dataset consisting of 120 logs with different distributions of changes, getting better results in terms of detection and classification accuracy, delay, and change region overlapping than the main state-of-the-art algorithms.

Predictive monitoring of business processes is a subfield of process mining that aims to predict, among other things, the characteristics of the next event or the sequence of the next events. Although multiple approaches based on deep learning have been proposed, mainly recurrent neural networks and convolutional neural networks, none of them really exploit the structural information available in process models. This paper proposes an approach that simultaneously learns spatio-temporal information from both the event log and the process model by combining recurrent neural networks with graph convolutional networks. Thus, common patterns from process models, such as loops or parallels, can be learned while avoiding overwriting information during the encoding phase. An experimental evaluation of real-life event logs shows that our approach is more consistent and outperforms the current state-of-the-art approaches.

Process mining techniques need to know not only the activity that is recorded at each moment but also its execution context, i.e., how the activity is related to the rest of process activities. This information is very relevant for addressing many process mining problems, such as predictive monitoring, especially when applying deep learning techniques. However, the representation of this execution context has not received much attention in the literature. In this paper, we present a novel approach based on an autoencoder for the generation and training of embeddings that properly capture the contextual information related to the execution of an activity. This deep learning-based architecture allows to obtain representations at the level of individual activity, roles, or any other categorical attributes. To evaluate this proposal, an experimentation has been conducted that shows how the embeddings generated with the proposed autoencoder allow to improve the accuracy of the next activity prediction with three of the main state-of-the-art predictive models. Particularly noteworthy are the results on complex processes, where the contextual information provided by our embeddings can significantly improve the state-of-the-art results.

Real life business processes change over time, in both planned and unexpected ways. The detection of these changes is crucial for organizations to ensure that the expected and the real behavior are as similar as possible. These changes over time are called concept drifts and its detection is a big challenge in process mining since the inherent complexity of the data makes difficult distinguishing between a change and an anomalous execution. In this article, we presentC2D2(Conformance Checking-based Drift Detection), a new approach to detect sudden control-flow changes in the process models from event traces.C2D2combines discovery techniques with conformance checking methods to perform an offline detection. Our approach has been validated with a synthetic benchmarking dataset formed by 68 logs, showing an improvement in the accuracy while maintaining a very low delay in the drift detection.

Predictive monitoring of business processes is concerned with the prediction of ongoing cases on a business process. Lately, the popularity of deep learning techniques has propitiated an ever-growing set of approaches focused on predictive monitoring based on these techniques. However, the high disparity of process logs and experimental setups used to evaluate these approaches makes it especially difficult to make a fair comparison. Furthermore, it also difficults the selection of the most suitable approach to solve a specific problem. In this paper, we provide both a systematic literature review of approaches that use deep learning to tackle the predictive monitoring tasks. In addition, we performed an exhaustive experimental evaluation of 10 different approaches over 12 publicly available process logs.

Learning Analytics (LA) is currently the most effective way of achieving better information and in-depth insights of the learning processes. Specifications like Experience API (xAPI) have been defined as part of LA initiatives to add interoperability among LA-aware applications. Tools that validate the conformance of data to these specifications are key components to assure the interoperability among applications. In this paper, a comparative evaluation of relevant validation tools of the xAPI specification is presented. This comparison focus specially on the structural and semantic features of the xAPI specification, revealing that most of the currently available tools do not support the semantic constraints of the specification.

In this paper, we present a big data software architecture that uses an ontology, based on the Experience API specification, to semantically represent the data streams generated by the learners when they undertake the learning activities of a course, e.g., in a course. These data are stored in a RDF database to provide a high performance access so learning analytics services can process the large amount of data generated in a virtual learning environment. These services provide valuable information to teachers and instructors such as predict the learner's performance, discover the real learning paths, extract the learner's behavior patterns and so on. The proposed architecture has been validated in the Educational Technology undergraduate course of the Degree in Pedagogy at the Faculty of Education of the University of Santiago de Compostela.

Capturing and storing learners' data is the first step to implement a learning analytics architecture. The Experience API (xAPI) specification is a de facto standard that describes (i) a REST-based API to store and retrieve the learners' activity data, and (ii) an RDF-based data model where the restrictions among data are specified in natural language. In this paper, we present an ontology that formally represents both the data model and the restrictions of the xAPI specification, facilitating the conformance testing of the data coming from the learning management system to guarantee the compliance with the specification.

SUMMARY The Universia repository is composed of more than 15 million of educational resources. The lack of metadata describing these resources complicates their classification, search and recovery. To overcome this drawback, it was decided to semantically annotate the available educational resources using the ADEGA algorithm. For this objective, we selected the DBpedia, a cross‐domain linked data composed of more than 3.77 million ‘things’ with 400 million ‘facts’, in order to make sure that the wide range of Universia topics are covered by the ontology. However, this kind of process is extremely expensive from a computational point of view: more than 1600 years of CPU time was estimated to achieve it. In this paper, parallel programming techniques and distributed computing paradigms are combined in order to achieve this semantic annotation in a reasonable time. The cornerstone of this proposal is a resource management and execution framework able to integrate heterogeneous computing resources at our disposal (grid, cluster and cloud resources). As a result, the problem was solved in less than 180 days, demonstrating that it is perfectly feasible to exploit the advantages of these computing models in the field of linked data. Copyright © 2014 John Wiley & Sons, Ltd.

One of the most challenging issues in learning analytics is the development of techniques and tools that facilitate the evaluation of the learning activities carried out by learners. In this paper, we faced this issue through a process mining-based platform, called Soft Learn, that is able to discover complete, precise and simple learning paths from event logs. This platform has a graphical interface that allows teachers to better understand the real learning paths undertaken by learners.

In this paper an approach for taking advantage of the information available in the social network tools of Learning Management Systems is presented. Specifically, our objective is to determine domain experts from their participation in the social networks. Our approach first integrates the different sources of information through an ontology based on the Dublin Core Metadata, and then applies Semantic Web technology to filter and classify each one of the social resources, so providing the basis for a better precision in the information retrieval process. The solution has been implemented following the paradigm of service oriented architecture.

Learning objects have arisen in response to the need of high-quality and reusable instructional materials. The repositories that hold learning objects allow educators to create and share their instructional contents in an organized infrastructure and where information should be easily searchable. Therefore, a key point of these repositories is the way learning objects are categorized. In this paper, we present an approach for classifying semantically learning objects whose metadata are described in Dublin Core. Specifically, our objective is to annotate automatically the subject of the learning object with instances of the DBpedia ontology, that is, to annotate the learning objects repository with linked open data. The syntactic and semantic analysis of the learning object will drive the selection of the most appropriate categories in the linked data repository.

Nowadays a greater number of domains require an improvement in the flexibility of its e-learning systems to so facilitate the adaptation of their course to their user needs. Some adaptive learning systems have been developed in recent years to cover part of these needs, and provide the ability to redirect the student learning flow based on the knowledge base (mainly rules) defined by the teacher. However, these systems, most of them based on the IMS Learning Design specification, have a major restriction: once the course has been published, it can not be changed, regardless that the circumstances in which the course is taught have changed. To remedy this lack, in this paper we present a solution to adapt the structure of units of learning at runtime. This solution extends a proposal for modelling IMS Learning Design through workflows with new capabilities that facilitate the replacement of plays, events and activities on the fly.

In this paper we present an approach for improving a specific class of semantic annotation, that relates a term of the document with a (sub)tree of the ontology, instead of linking a term with a single concept of the ontology. An important part of this class of annotation is filtering the relevant (sub)nodes and relations, because the returned graph should only contain relevant information, that is, nodes that are truly related with the topics of the document. In addition, we consider that the relevance of nodes vary depending on if the node is a branch or a leaf, that is, if the node has links to other nodes or it is a text-based description. This paper focuses on the relevance of branch nodes, which is calculated from the relevance of its links, since leaf nodes relevance is usually estimated by similarity metrics. Specifically, our approach incises in learning (through a genetic algorithm) and assigning the most appropriate weights to these links in order to reduce the precision/recall curve of the annotation process. The results show that our solution is viable and outperforms the state of the art approaches.

This paper describes a service-oriented architecture that facilitates the execution of learning units, based on the IMS Learning Design specification (IMS LD), in a metaverse platform like Second Life or OpenSim. This architecture ex-ternalizes the functionalities of the IMS LD engine through web services that are invoked by metaverse platform scripts. A software module that translates the requests of the virtual platform has been developed in order to facilitate this invocation and a virtual teaching scenario has been designed to test the feasibility of this architecture.

In this paper an ontology-based Petri net engine to execute units of learning based on the IMS Learning Design (IMS LD) specification is presented. This engine has a layered architecture that enables the execution of IMS LD learning flows by means of high-level Petri nets representing the semantics of the IMS LD dynamic elements. Even if changes in the specification were made, the engine would not have to be re-implemented, and only the introduction of new rules and/or axioms in the ontology would required. Moreover, a service-oriented architecture has been implemented for allowing web-based learning management systems to access to the IMS LD engine.

Machine scheduling is a critical problem in industries where products are custom-designed. The wide range of products, the lack of previous experiences in manufacturing, and the several conflicting criteria used to evaluate the quality of the schedules define a huge search space. Furthermore, production complexity and human influence in each manufacturing step make time estimations difficult to obtain thus reducing accuracy of schedules. The solution described in this paper combines evolutionary computing and neural networks to reduce the impact of (i) the huge search space that the multi-objective optimization must deal with and (ii) the inherent problem of computing the processing times in a domain like custom manufacturing. Our hybrid approach obtains near optimal schedules through the Non-dominated Sorting Genetic Algorithm II (NSGA-II) combined with time estimations based on multilayer perceptron networks.

This work describes the design and implementation of a knowledge based system that automatizes the furniture budgeting task. The aim of this task is to estimate the production cost that would suppose a furniture order manufacturing. Fulfillment of this aim requires the management of the company resources, information and knowledge in order to determine workload and financial impacts of client request manufacturing. Task modelling has been done through the workflows formalism approach and describes working activities and its relationships with the company resources. The system software architecture is based on workflow modeling and is an adjustment of a conceptual architecture proposed to be a standard for workflows systems.