Re: [TF-DI] Discovery Categories and Tech Landscape

> On 9 Jul 2015, at 14:27, Broering, Arne <arne.broering@siemens.com> wrote:
> 
> Soumya, Dave,
> 
> Again, thanks for your input. I edited the wikipage and incorporate your comments and hints into the set of discovery categories. I also added a new category for peer to peer discovery.
> Please don't hesitate to adjust the wikipage if you think of some missing technology!
> 
> Dave, you mentioned discovery approaches incorporating relationships between people and things. Can you point us at links explaining this in more detail?
> 
>> Another approach is based upon social relationships between people   
>> and things, including abstract things like personal, organisational,  
>> spatial and temporal zones.  This has the advantage of providing a   
>> clear context for discovery and builds upon rich metadata.

I don’t have the links to hand.  However, it should be possible to search for some.   I have recollections of papers on this at a number of conferences. I have also explored the ideas in a couple of EU projects (webinos and Mediascape). The details depend on the kind of social network, e.g. whether it is centralised or distributed, open or closed.

As an example, consider a distributed social network where every person has a URL. An HTTP GET on the URL retrieves linked data describing that person, including relationships to other people, where those people are identified by URLs. FOAF is one such vocabulary, see [1] and can be combined with other vocabularies to describe devices, capabilities and locations etc. The approach can be combined with access control techniques to control who can access which descriptions. By describing your relationships to others using a vocabulary of terms, you can then use this vocabulary as a basis for access control rules. These can also act upon presence information and other contextual data, e.g. your location, the time of day, whether it is a weekend or work day, whether you are on vacation and so forth.

In centralised social networks like Facebook, each user has an account and credentials to log on to that account. Authentication is harder for distributed social networks. Setting up a conventional account with each of a large number of servers would be a pain. An alternative is for each person to provide a public key that can be used to authenticate him or her. This could be published as part of that person’s description.

If you don’t know someone’s URL, then perhaps you know their email address.  RF 7033 defines the WebFinger protocol[2][3] which can be used to discover information about people or other entities on the Internet using standard HTTP methods. WebFist is a fallback when providers don't support WebFinger natively.  More generally, you can use distributed search over peer to peer networks.

For the Web of Things, we could perhaps aim to evolve vocabularies to support rich descriptions of people, devices and contexts as a basis for discovery and access control. WebFinger hasn’t taken off, so we need to understand the business drivers for such services. Web search engines are powered by advertising. So are centralised social networks like Facebook. Open markets of services for the Web of Things may require different approaches, and we should devote time to considering this.


[1] http://www.foaf-project.org/ <http://www.foaf-project.org/>
[2] https://tools.ietf.org/html/rfc7033 <https://tools.ietf.org/html/rfc7033>[3] https://code.google.com/p/webfinger/ <https://code.google.com/p/webfinger/>
[4] http://www.onebigfluke.com/2013/06/bootstrapping-webfinger-with-webfist.html <http://www.onebigfluke.com/2013/06/bootstrapping-webfinger-with-webfist.html>


—
   Dave Raggett <dsr@w3.org <mailto:dsr@w3.org>>