Approximate Semantic Matching Over Linked Data Streams

In the Internet of Things (IoT),data can be generated by all kinds of smart things. In such context, enabling machines to process and understand such data is critical. Semantic Web technologies, such as Linked Data, provide an effective and machine-understandable way to represent IoT data for further processing. It is a challenging issue to match Linked Data streams semantically based on text similarity as text similarity computation is time consuming. In this paper, we present a hashing-based approximate approach to efficiently match Linked Data streams with users’ needs. We use the Resource Description Framework (RDF) to represent IoT data and adopt triple patterns as user queries to describe users’ data needs. We then apply locality-sensitive hashing techniques to transform semantic data into numerical values to support efficient matching between data and user queries. We design a modified k nearest neighbors (kNN) algorithm to speedup the matching process. The experimental results show that our approach is up to five times faster than the traditional methods and can achieve high precisions and recalls.