A Framework for Acquiring Semantic Sensor Descriptions (Short Paper) Luka Bradeško, Alexandra Moraru, Blaž Fortuna, Carolina Fortuna, Dunja Mladenić Jožef Stefan Institute, Ljubljana Slovenia {luka.bradesko, alexandra.moraru, blaz.fortuna, carolina.fortuna, dunja.mladenic}@ijs.si Abstract. There has been great effort in developing ontologies for modeling sensor networks, describing various types of sensors and their context. Howev- er, when faced with a large scale deployment, the process of acquiring and managing semantic sensor metadata is challenging. This paper focuses on ac- quiring contextual metadata of sensors, such as location and surrounding envi- ronment, as opposed to technical metadata which can be derived from sensor’s firmware. More specifically, the paper proposes a framework for collecting contextual metadata information with help of the mobile devices, which allows usage on the deployment site and as such lowers the cost. Keywords. Semantic Sensor Web, Semantic Sensor Networks, Knowledge Ac- quisition, Mobile Applications 1 Introduction With the rapid development and increasing number of real world applications of large scale sensor networks it became obvious that there is a need for software solu- tions supporting communication, sensor data retrieval and storage. In parallel to that, there is a need for an infrastructure to cover sensor descriptions, deployment and maintenance data. This paper focuses on the infrastructure supporting sensor meta- data acquisition and management based on semantic technologies. Semantic technologies have been identified as one of the key enabling technologies for sensor networks [1], contributing to understanding and managing of the sensors and measurements. One of the advantages of applying semantic technologies to sen- sor networks is the interoperability support, which in terms of comparison and data merging of different sensor networks, enables new solutions in solving problems. Existing systems that semantically annotate data require it to be inserted manually via xml configuration files or wiki. Attempts to use meta-data freely inserted by users haven’t proved too successful [3]. More recent approaches use custom network proto- cols such as the Device Identification Protocol (DIP) to automatically obtain the tech- nical meta-data [4], or they manually annotate a small number of sensors and then, based on the similarity of measurements they label other sensors [5]. The existing applications mostly focus on the measurements and insufficient attention has been