[foaf-dev] beyond foaf:mbox_sha1sum

Story Henry henry.story at bblfish.net
Tue Dec 22 14:15:34 CET 2009


On 22 Dec 2009, at 12:00, Mischa Tuffield wrote:

> As for the blacklist of IFPs, we too have a blacklist of IFPs, which we flag up in the FOAF validator. We also found that many people use the foaf:homepage property to point to their browser's homepage. And as a result we had to blacklist, http://www.google.com, yahoo.com, and bbc.co.uk to name a few. 

:-D That is funny. I had not thought of that way of reading the foaf:homepage relation.

The problem here as with mbox sha1 sum really is not the ontology, it is wrong data on the web. (Though I think with sha1 sum there are reasonable arguments that it is not an IFP).

 To stop this one has to find friendly ways to make people think correctly about what they are doing. This is a subtle user interface job. It has to be somewhat similar to what we do when someone makes a mistake in speaking. So if someone says they are the last remaining ancestor of the King of Scotland [1] we could laugh out loud, or suggest that they look very good for their age.

When someone specifies a foaf:homepage, one can follow through the consequences of this: 
  - ask them if they want to make it their OpenID, and then give them instructions on how to do this, so that it can be verified. 
  - if one knows someone else who has publicly claimed this as their home page, one could ask them if they want to merge their foaf files or keep them separate
  - if from that home page one can find links to twitter accounts, openids, etc,... one can suggest that they be added to the profile.

By showing the consequences of a statement - ie. by using OWL, and the additional semantics defined in a term's rdf:comment - one can help people understand what they are doing when they make a statement.  And this without them having to read any OWL.  In short we are moving from declarative meaning to meaning as use.

Let us turn this around, and look at what it would be for a consumer of foaf - something like the Semantic Address Book, or a foaf+ssl protected resource server - to help improve the quality of linked foaf data. The consumer has some weight on the publisher as a potential publisher himself, or as a controller of access to resources. 

So for the Address Book case. While the user is browsing the linked foaf data web, and comes across someone's foaf file which contains an IFP that would allow it to merge that user with info from another foaf, then the Address Book, could ask the user if it would like to merge the two users, of if more convenient, merge the info but show how this merged information is coming from different graphs. [3] So here the User Agent is keeping track of which graphs it merges and which not. The user himself is the person voting for such merges. If people have wrong IFPs, the users will tend not to merge that user's file, and may want to reduce their direct links to that file too. This in turn creates a pressure on the people with broken foaf to change it - friends calling them, and letting them know about their problem for example.

  Another type of consumer that can create pressure on publishers of foaf, are foaf+ssl enabled web services. If a service gives friends of some group of people access to a resource,
then it can use the inferencing power of what those people's foaf say to create pressure to truthful foaf. So for example imagine a foaf+ssl enabled voting service, where people could vote on foaf specification changes. The service should only allow each person to vote once. If two URIs have the same home page, then if one votes the other will be counted as having voted too. If there is a mistake, this will soon surface, as people find they cannot vote, who would have liked to. The organiser of the vote will then have to decide what is the case, if both users can't get themselves to fix their foaf. 

  So it is only by starting to use foaf in ways that make use of the OWL and other inferencing power of the ontology that we are going to create a pressure - probably describable using game theory - towards growing webs of coherent data. But the applications need to be able to give feedback to the users on the implications of their statements. 

   That is why just publishing linked data is enough. Let the games start !

Henry


[1] http://www.wsu.edu/~brians/errors/ancestor.html
[2] https://sommer.dev.java.net/AddressBook.html
[3] http://sig.ma/ shows a way to do this. The problem with this service is that because it is trying to crawl the whole web of data, and as it shows a view of all the data, it tends to merge way too much information. In a personal address book, the user has anchors he is starting from and has reason to trust: namely URL given to him by people, or blogs he likes to read. From there he is not going to move very quickly to crawl the whole web, so it will be much easier to keep him up to date on how things are moving.


More information about the foaf-dev mailing list