FreGraPaD: Frequent RDF graph patterns detection for semantic data streams

Belghaouti, Fethi; Bouzeghoub, Amel; Kazi-Aoul, Zakia; Chiky, Raja

doi:10.1109/rcis.2016.7549333

Cited by 7 publications

(8 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, as the above-described approach relies on traditional data mining techniques for performance pattern discovery and semantic modelling for representation of the results together with the available building data, we compare it to a direct semantic data mining approach using a frequent RDF graph pattern analysis method (Belghaouti et al 2016).…”

Section: Methodsmentioning

confidence: 99%

“…That requires the selection of relevant ontologies, defining an appropriate mapping language for conversion, selection of continuous query languages and choosing relevant datasets to link to (Llanes et al 2016) (Fig.10). To demonstrate the principle of pattern recognition within the RDF graph structure, we employ a method for frequent RDF graph pattern detection in semantic data streams, which relies on the graph predicates (Belghaouti et al 2016).…”

Section: Rdf Stream Processing and Rdf Graph Pattern Recognitionmentioning

confidence: 99%

“…The proposed method relies on the fact that the streams are represented according to particular ontologies, which means that most streams will be relatively uniform and expose a very frequent RDF graph structure. Figure 11: A directed star graph as represented in a RDF stream and the corresponding adjacency matrix based on the graph predicates, based on Belghaouti et al (2016) According to Belghaouti et al (2016), RDF graphs could then be represented using adjacency matrices, however, such an approach would not be fully efficient and is not suitable for RDF graphs as it will result in a very sparse matrix (right in Fig.11). Therefore, it is proposed to reduce the associated adjacency matrix to a bit vector (Fig.12).…”

Section: Rdf Stream Processing and Rdf Graph Pattern Recognitionmentioning

confidence: 99%

See 2 more Smart Citations

Semantic Enrichment of Association Rules Discovered in Operational Building Data for Reuse of Building Performance Patterns

Petrova¹,

Pauwels

2020

CIB W78 Proceedings

View full text Add to dashboard Cite

The advancements in Building Information Modelling, Building Monitoring Systems and machine learning have made the discovery of hidden insights and performance patterns in operational building data possible and highly accurate. Semantic web technologies play a fundamental role in terms of knowledge representation and provide the necessary infrastructure for reuse of the discovered insights. Such knowledge can be of particular importance to decision-making for building performance improvement, however, this requires patterns discovered with traditional data mining techniques to be attributed with semantics, so that they can be machine-interpretable and reusable. Using linked data-based crowdsourcing techniques for interpretation of building performance patterns enables the creation of knowledge graphs of building data, enriched with contextualized building performance insights. This paper presents a crowdsourcing mechanism that allows the semantic enrichment of building performance patterns through semantic annotation and classification. We discuss the results and the potential of linked building data graphs enriched with building performance insights.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Rdf Stream Processing and Rdf Graph Pattern Recognitionmentioning

confidence: 99%

Section: Rdf Stream Processing and Rdf Graph Pattern Recognitionmentioning

confidence: 99%

See 1 more Smart Citation

Semantic Enrichment of Association Rules Discovered in Operational Building Data for Reuse of Building Performance Patterns

Petrova¹,

Pauwels

2020

CIB W78 Proceedings

View full text Add to dashboard Cite

show abstract

“…The following sections explain this architecture focusing on (1) how patterns are discovered and added to the graph, (2) how user profiles can be built and benefit from the system, including feedback, and (3) how recommendations can be generated. We present an example for RDF pattern discovery in a semantic data stream by implementing a method suggested by Belghaouti et al (2016) and discuss its potential feasibility. Finally, we demonstrate an initial implementation of a linked data-based recommender system by applying the concept of Linked Data Semantic Distances proposed by Passant (2010).…”

Section: Conceptual Frameworkmentioning

confidence: 99%

“…subclassOf relations). To analyze how RDF stream processing would affect the recommendation concept, we employ the method described by Belghaouti et al (2016), who identify frequent RDF patterns in RDF streams by mapping the graphs to adjacency matrices based on the graph predicates. Using this method, one is able to construct bit vectors, which describe the graph structure.…”

Section: Listing 1: Namespaces and Prefixes Used In The Following Examentioning

confidence: 99%

Semantic data mining and linked data for a recommender system in the AEC industry

Petrova¹,

Pauwels

Svidt³

et al. 2019

Proceedings of the 2019 European Conference on Computing in Construction

View full text Add to dashboard Cite

Even though it can provide design teams with valuable performance insights and enhance decision-making, monitored building data is rarely reused in an effective feedback loop from operation to design. Data mining allows users to obtain such insights from the large datasets generated throughout the building life cycle. Furthermore, semantic web technologies allow to formally represent the built environment and retrieve knowledge in response to domain-specific requirements. Both approaches have independently established themselves as powerful aids in decision-making. Combining them can enrich data mining processes with domain knowledge and facilitate knowledge discovery, representation and reuse. In this article, we look into the available data mining techniques and investigate to what extent they can be fused with semantic web technologies to provide recommendations to the end user in performance-oriented design. We demonstrate an initial implementation of a linked data-based system for generation of recommendations.

show abstract