Ana Roxin

Automatic ontology-based user profile learning from heterogeneous web resources in a big data context

The Web has developed to the biggest source of information and entertainment in the world. By its size, its adaptability and flexibility, it challenged our current paradigms on information sharing in several areas. By offering everybody the opportunity to release own contents in a fast and cheap way, the Web already led to a revolution of the traditional publishing world and just now, it commences to change the perspective on advertisements. With the possibility to adapt the contents displayed on a page dynamically based on the viewers context, campaigns launched to target rough customer groups will become an element of the past. However, this new ecosystem, that relates advertisements with the user, heavily relies on the quality of the underlying user profile. This profile has to be able to model any combination of user characteristics, the relations between its composing elements and the uncertainty that stems from the automated processing of real-world data. The work at hand describes the beginnings of a PhD project that aims to tackle those issues using a combination of data analysis, ontology engineering and processing of big data resources provided by an industrial partner. The final goal is to automatically construct and populate a profile ontology for each user identified by the system. This allows to associate these users to high-value audience segments in order to drive digital marketing.

SWRL rule-selection methodology for ontology interoperability

Data interoperability represents a great challenge for todays enterprises. Indeed, they use various information systems, each relying on several different models for data representation. Ontologies and notably ontology matching have been recognized as interesting approaches for solving the data interoperability problem. In this paper, we focus on improving the performance of queries addressed over ontology alignments expressed through SWRL rules. Indeed, when considering the context of executing queries over complex and numerous alignments, the number of SWRL rules highly impacts the query execution time. Moreover, when hybrid or backward-chaining reasoning is applied, the query execution time may grow exponentially. Still, the reasoners involved deliver performant results (in terms of execution time) when applied over reduced and simpler rule sets. Based on this statement, and to address the issue of improving the query execution time, we describe a novel approach that allows, for a given query, to ignore unnecessary rules. The proposed Rule Selector (RS) is a middleware between the considered systems and the reasoner present on the triple store side. Through the benchmarks realized we prove that our approach allows considerably minimizing query execution time.

A performance benchmark over semantic rule checking approaches in construction industry

As more and more architectural design and construction data is represented using the Resource Description Framework (RDF) data model, it makes sense to take advantage of the logical basis of RDF and implement a semantic rule checking process as it is currently not available in the architectural design and construction industry. The argument for such a semantic rule checking process has been made a number of times by now. However, there are a number of strategies and approaches that can be followed regarding the realization of such a rule checking process, even when limiting to the use of semantic web technologies. In this article, we compare three reference rule checking approaches that have been reported earlier for semantic rule checking in the domain of architecture, engineering and construction (AEC). Each of these approaches has its advantages and disadvantages. A criterion that is tremendously important to allow adoption and uptake of such semantic rule checking approaches, is performance. Hence, this article provides an overview of our collaborative test results in order to obtain a performance benchmark for these approaches. In addition to the benchmark, a documentation of the actual rule checking approaches is discussed. Furthermore, we give an indication of the main features and decisions that impact performance for each of these three approaches, so that system developers in the construction industry can make an informed choice when deciding for one of the documented rule checking approaches.

FOWLA, A Federated Architecture for Ontologies

The progress of information and communication technologies has greatly increased the quantity of data to process. Thus, managing data heterogeneity is a problem nowadays. In the 1980s, the concept of a Federated Database Architecture (FDBA) was introduced as a collection of components to unite loosely coupled federation. Semantic web technologies mitigate the data heterogeneity problem, however due to the data structure heterogeneity the integration of several ontologies is still a complex task. For tackling this problem, we propose a loosely coupled federated ontology architecture (FOWLA). Our approach allows the coexistence of various ontologies sharing common data dynamically at query execution through logical rules. We have illustrated the advantages of adopting our approach through several examples and benchmarks. We also compare our approach with other existing initiatives.

Dynamic, Behavior-Based User Profiling Using Semantic Web Technologies in a Big Data Context

The success of shaping the e-society is crucially dependent on how well technology adapts to the needs of each single user. A thorough understanding of one’s personality, interests, and social connections facilitate the integration of ICT solutions into one’s everyday life. The MindMinings project aims to build an advanced user profile, based on the automatic processing of a user’s navigation traces on the Web. Given the various needs underpinned by our goal (e.g. integration of heterogeneous sources and automatic content extraction), we have selected Semantic Web technologies for their capacity to deliver machine-processable information. Indeed, we have to deal with web-based information known to be highly heterogeneous. Using descriptive languages such as OWL for managing the information contained in Web documents, we allow an automatic analysis, processing and exploitation of the related knowledge. Moreover, we use semantic technology in addition to machine learning techniques, in order to build a very expressive user profile model, including not only isolated “drops” of information, but inter-connected and machine-interpretable information. All developed methods are applied to a concrete industrial need: the analysis of user navigation on the Web to deduct patterns for content recommendation.

Adapting numerical representations of lung contours using Case-Based Reasoning and Artificial Neural Networks

In case of a radiological emergency situation involving accidental human exposure, a dosimetry evaluation must be established as soon as possible. In most cases, this evaluation is based on numerical representations and models of subjects. Unfortunately, personalised and realistic human representations are often unavailable for the exposed subjects. However, accuracy of treatment depends on the similarity of the phantom to the subject. The EquiVox platform (Research of Equivalent Voxel phantom) developed in this study uses Case-Based Reasoning principles to retrieve and adapt, from among a set of existing phantoms, the one to represent the subject. This paper introduces the EquiVox platform and Artificial Neural Networks developed to interpolate the subject’s 3D lung contours. The results obtained for the choice and construction of the contours are presented and discussed.

Querying and reasoning over large scale building data sets: an outline of a performance benchmark

The architectural design and construction domains work on a daily basis with massive amounts of data. Properly managing, exchanging and exploiting these data is an ever ongoing challenge in this domain. This has resulted in large semantic RDF graphs that are to be combined with a significant number of other data sets (building product catalogues, regulation data, geometric point cloud data, simulation data, sensor data), thus making an already huge dataset even larger. Making these big data available at high performance rates and speeds and into the correct (intuitive) formats is therefore an incredibly high challenge in this domain. Yet, hardly any benchmark is available for this industry that (1) gives an overview of the kind of data typically handled in this domain; and (2) that lists the query and reasoning performance results in handling these data. In this article, we therefore present a set of available sample data that explicates the scale of the situation, and we additionally perform a query and reasoning performance benchmark. This results not only in an initial set of quantitative performance results, but also in recommendations in implementing a web-based system relying heavily on large semantic data. As such, we propose an initial benchmark through which new upcoming data management proposals in the architectural design and construction domains can be measured.

Automatic Integration of Spatial Data into the Semantic Web.

For several years, many researchers tried to semantically integrate geospatial datasets into the semantic web. Although, there are many general means of integrating interconnected relational datasets (e.g. R2RML), importing schema-less relational geospatial data remains a major challenge in the semantic web community. In our project SemGIS we face significant importation challenges of schema-less geodatasets, in various data formats without relations to the semantic web. We therefore developed an automatic process of semantification for aforementioned data using among others the geometry of spatial objects. We combine Natural Language processing with geographic and semantic tools in order to extract semantic information of spatial data into a local ontology linked to existing semantic web resources. For our experiments, we used LinkedGeoData and Geonames ontologies to link semantic spatial information and compared links with DBpedia and Wikidata for other types of information. The aim of our experiments presented in this paper, is to examine the feasibility and limits of an automated integration of spatial data into a semantic knowledge base and to assess its correctness according to different open datasets. Other ways to link these open datasets have been applied and we used the different results for evaluating our

COBieOWL, an OWL Ontology Based on COBie Standard

Building Information Modelling BIM standards have been recognized by construction and political actors as a highly promising tool for resolving issues of data dematerialization, notably in the field of Architecture, Engineering, Construction and Facility Management. Among worldwide-adopted BIM standards, there is COBie. COBies goals are to dematerialize building data and to enhance building information interoperability. Nevertheless, COBie data is available in static formats such as STEP or spreadsheet templates. Such formats lack logical formalisms and semantic features provided by languages such as OWL. Because of this, in this paper, we propose a method for semi-automatically conceiving an OWL ontology for the COBie standard starting from a COBie spreadsheet template. We call this ontology COBieOWL and we populate it directly from COBie spreadsheet data files as used by building actors. We also discuss various benefits of adopting our approach, for example: it reduces semantic heterogeneity of the COBie model. Besides, we discuss how COBieOWL can be linked to other LOD datasets such as FOAF.

Ontologies and Knowledge Aggregation in the Active Semantic Learning System

The construction of semantic-based learning systems depends on the development of ontologies and the capacity to integrate and exploit knowledge using semantic technologies, notably RDF and ontologies. In this paper we present some of the concepts and ontologies defined in the Active Semantic Learning System (Active SLS). The purpose is to obtain a learning system that is capable of aggregating knowledge from different sources from the web and exploiting that knowledge for the benefit of the learner.