- From: Catherine Roussey <catherine.roussey@irstea.fr>
- Date: Tue, 22 Aug 2017 18:17:27 +0200
- To: <public-sdw-wg@w3.org>
- Cc: "'Stephan Bernard'" <stephan.bernard@irstea.fr>
Dear all I do not know if you receive my previous email inform you that we can not do the translation of our weather dataset from ssn to sosa For the 8 of august. About the old version of ssn our datasets is available at http://ontology.irstea.fr/pmwiki.php/Site/WeatherData the sparqlendpoint ishttp://ontology.irstea.fr/weather/snorql/ Now we are in the process of studying the sosa ontology in order to create a new dataset about weather data. We face some trouble related to the cardinality constraint put on hasFeatureOfInteret and hasSample properties. A weather Station is able to study several climatic phenomenon wind, precipitation, air etc... We reuse the climate and forecast vocabulary that defines all features of interest (wind, precipitation) and their associated properties (wind speed, wind direction, etc...). Thanks to this reference ontology all our weather station located at different point in France could be classified automatically depending of the feature of interest observed. Thanks to this weather ontology we could query all our dataset to find all the observation associated to a specific property like (wind speed) If I understand well, we should create a sample that represent the weather station. This sample will established the location of the weather station and also the high of the weather station. For example a weather station in agriculture domain should be as high as a crop culture (the high should be less than 1.5 meter). A weather station for airplane or other domain should be fixed at different high. So we like the precision of the sample. The problem is that the weather station sample is a sample for several features of interets (wind, precipitation etc...) So why do you put a constraint cardinality between sosa:Sample and FeatureOfInterest. In your example of spatial phenomenon the proposed sample is associated to a location. For each sample we have to redefine the property (create a new uri and not reuse a generic uri like climate and forecast windspeed) see example 17 in sosa. It means that we could not query uniformally all our weather station in France to get all the observation related to windSpeed property. What we would like is to be able to associated a each observation (and each sensor) the generic feature of interest (provided by climate and forecast ontology) and also a sample. The associated property of the generic feature of interest and the sample could be linked. For example we could generate automatically the uri of the sample property based on the property of the generic feature of interest. The precision of the sample will be useful for weather expert that have to propose weather forecast model. As mention earlier the high is important for prediction. the generic feature of interest will be useful for farmers and to provide generic query to retrieve data from dataset related to a same domain (climate, crop, soil etc...). Best Regards Catherine and Stephan -- Catherine ROUSSEY Irstea Clermont Ferrand Campus des Cézeaux 9 avenue Blaise Pascal CS 200 85 63178 Aubière tel: 33 (0)4 73 44 06 88 "Imprefection si beauty, madness is genious and it's better to be absolutely ridiculous than aboslutely boring" Maryline Monroe
Received on Tuesday, 22 August 2017 16:17:47 UTC