- From: Xiang Su <xiangsu@ee.oulu.fi>
- Date: Tue, 4 Aug 2015 15:40:07 +0300
- To: public-rsp@w3.org, haroonr@iiitd.ac.in, fadirra@gmail.com, jean-paul.calbimonte@epfl.ch, Xiang Su <xiang.su@ee.oulu.fi>
Hi, We have some publications about utilizing different semantic representation for IoT data. http://onlinelibrary.wiley.com/doi/10.1002/cpe.3203/full http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=7030117&tag=1 I hope they somehow help. br, -Xiang On 4.8.2015 15:07, Haroon Rashid wrote: > Thanks Fariz for the detailed explanation. Although I understood the > importance of RDF, but I am still curious to know why people are not > interested in it. I think people either find it difficult to adapt or > find it much more verbose. > > On Tue, Aug 4, 2015 at 2:02 AM, Fariz Darari <fadirra@gmail.com > <mailto:fadirra@gmail.com>> wrote: > > Hi Haroon, > > I would like to give a try answering your questions: > > 1. RDF is good at interoperability. Consider the use case of stream > data integration. There can be two different situations: > a. When your data and the others have already been described using a > shared, common ontology (e.g., Semantic Sensor Networks ontology > [1]), you basically need minimal/no integration effort. > b. When your data is in RDF but is described in different > ontologies: you can use some off-the-shelf stream integration > techniques like from [2], which uses the R2RML standard as a mapping > language. > > 2. RDF is good at reasoning. On top of RDF, there is OWL, which is > an ontology language for additional inference over your stream data. > TrOWL is an example of an OWL reasoner for stream services [3]. > > 3. RDF is good for the Web. RDF by design is a data model for the > Web. Everything is represented by URIs, can ideally be dereferenced > via HTTP, and can be linked with other URIs [4]. Thus, the power of > the Web is inherited to RDF. As an example, you need not stream > complete description of some resources. You can just use URIs in > your stream, and the complete descriptions can be dereferenced on > demand depending whether your data stream consumers are interested in. > > I hope this helps. > > Best, > Fariz > > [1] > http://www.slideshare.net/rgcmme/overview-of-the-w3c-semantic-sensor-network-ssn-ontology > > [2] https://github.com/jpcik/morph-stream > [3] http://trowl.org/about/ > [4] http://www.w3.org/DesignIssues/LinkedData.html > > > Regards, > Fariz > > On Sat, Aug 1, 2015 at 9:53 AM, Haroon Rashid <haroonr@iiitd.ac.in > <mailto:haroonr@iiitd.ac.in>> wrote: > > Dear All, > > Sorry for disturbing you during the weekend. > > Thanks > Jean > for the explanation. > Here I am considering only IoT data. Generally, we send data > from sensors in Json/ Xml format, where a specific value > /reading > is represented by different key-value pairs > as > > {sensor/device: > > device_name > type: > > temperature > value: > > 32 > unit: > > degree > time: > > 12:12:12 > } > I > k > now that RDF/RDFa data is machine interpretable because of URIs, > which make it special. Things I am not able to understand > include: > > 1. How RDF makes data more discoverable? I mean even JSON/XML > data is discoverable because data is associated with a > number of attributes as shown in > the > example. In both representations, > > i.e., json/xml or RDF representation > , > > we must be knowing the attribute names or URIs before hand. > 2. Also, you are saying that if we don’t know the data > source/structure a-priori, then RDF data allows us to do > some fancy tasks. Can you please elaborate it with an > example? I think if we don’t have any clue about data > structure then It does not matters whether it is in json or > xml or RDF. > > The main point is here to find the importance of RDF/linked > data. How it makes a different impact on the research community. > Why should I represent my IoT data in RDF > steams > ? > > > > On Thu, Jul 30, 2015 at 1:54 PM, Jean Paul Calbimonte > <jpcalbimonte@delicias.dia.fi.upm.es > <mailto:jpcalbimonte@delicias.dia.fi.upm.es>> wrote: > > > > Hi Haroon, > > > > I guess there would be different answers to this. > > One example can be data discovery. If you already know your > data sources then it's usually fine to use existing > technologies: you know your schema and you can use your old > pub-sub stuff. > > But in IoT and other domains you sometimes do not know that > beforehand, and the interpretable data comes very handy to do > fancy data discovery tasks. On the contrary, if we use e.g. CSV > values 3.0, 4.5, 3, 6.7 without any semantic metadata it's > impossible to know what your data source is about. > > > > I think there are many other examples, this could be one. > > > > Jean-Paul > > > > ________________________________ > > Date: Mon, 27 Jul 2015 13:34:48 +0530 > > From: haroonr@iiitd.ac.in <mailto:haroonr@iiitd.ac.in> > > To: public-rsp@w3.org <mailto:public-rsp@w3.org> > > Subject: Why we work on RSP > > > > Hello Everyone, > > > > > > My apologies if I have sent this mail to the wrong discussion > group. > > > > > > My name is Haroon and I have joined the RSP W3C group > recently. I am working on linked, streaming data from last one > month approximately. I am considering IoT data (say temperature, > humidity readings) as a source of continuous data streams. I > find this area exciting; recently I had a discussion with some > of my colleagues about linked stream data processing. During > discussion we mainly discussed around the following points: > > > ______________________________________________________________________________________________ > > > > > 1. Why we need to work on linked-data (RDF) streams ? > > My response: Linked data is machine interpretable. Therefore > > , > > we need to represent our data into linked data form so that > machines can > > understand it and possibly can reason over it. This also > makes data sharable/reusable. Other data representations are not > machine interpretable. > > > > 2. We have several data representations available (say XML, > JSON, ….) and we have some efficient publish-subscribe systems, > which consume IoT data streams and then push > > the > > relevant data to end users/applications. Existing data > representations and publish-subscribe systems suffice the > current needs > > , > > then why should we go for linked streams data representation. > Apart from machine-interpretable feature it does not add > anything. Also > > , > > it makes data much more verbose and hence it might take more > time to process the data at processing engine. > > Response: …. > > > ______________________________________________________________________________________________ > > > > At the end of > > the > > discussion I found my colleagues were not satisfied because > none of them was > > an > > expert in semantic technologies. Although I am satisfied > about this area > > , > > but I need genuine feedback/comments from your side about the > above mentioned points. What makes linked > > , > > streaming data representation so special that we need to work > over it further? > > > > > > -- > > Haroon Rashid > > > > > -- > Haroon Rashid > > > > > > -- > Haroon Rashid > >
Received on Tuesday, 4 August 2015 12:46:57 UTC