This assignment is about SPARQL queries. To prepare the queries, study this dataset,

http://liyangyu.com/careprovider.txt

This is the dataset your queries will run against (it has been edited slightly to make it easier to understand), and here is a brief description of this dataset.

First off, resource

represents a care provider, as shown by the following statement,

"CP".

Where CP is short for Care Provider. You can see there are quite some statements showing other properties of this resource. For example, givenName and familyName.

Now, this care provider is working with several targets, identified by statement like this:

schema:seeks .

And a target is identified by the following statement,

"TG".

Where TG is short for target. From the dataset, you can also see this resource is now working with 5 targets, and for each target, there are quite some statements describing his properties.

A target can convert to a patient at some point. For example, in this dataset, only 1 targets has become a patient, as shown by this statement,

schema:sameAs .

And a patient has a different type code as shown below:

"PT".

Where PT is short for patient. Obviously, there are also quite some statements describing a given patient.

Now, we can write queries to get information we need. For example, the following query,

prefix schema:

SELECT * WHERE { ?patient schema:additionalType "PT". ?patient schema:email "sam.rose[at the google mail].com". }

Tries to find a patient, who has an email address given by sam.rose[at the google mail].com.

In this assignment, you will be writing queries against this dataset. You can also use any features from SPARQL 1.1 to finish the assignment. You do NOT have to actually run the queries, but you certainly need to write out the queries.

Each one of the following is 15 points.

For care provider with id 001, who are his targets? And what do we know about these targets?

For care provider with id 001, how many targets he is currently working with?

For care provider with id 001, how many his targets have been converted to patients?

For care provider with id 001, how many his patients are now living in an area with zip code 30092?

Finally, for your convenience, you can use the following when construct your queries.

PREFIX xsd:

PREFIX rdf:

PREFIX rdfs:

PREFIX owl:

PREFIX fn:

PREFIX dc:

PREFIX schema:

The following two questions are not related to the above query situation, but are again on SPARQL queries. Each one of the following question is 10 points each.

A triple store is a data store that contains RDF triples. Consider a triple store that contains RDF triples in the form of FOAF profiles. In other words, these RDF triples are written by using terms from one single ontology, the FOAF ontology (http://www.foaf-project.org/). However, since these RDF triples are collected from various web sites, the same person is likely identified by using different URIs.

For example, one site creates several RDF triples about Tim Berners-Lee, saying he is the inventor of the web we are using today. Another web site also talks about Tim Berners-Lee, by adding some more information about his contribution to the idea of the Semantic Web. These two web sites are actually using different URIs to represent Tim Berners-Lee as a resource.

Our goal is to integrate all the information about Tim Berners-Lee (and of course for all other similar cases). To do so, we need to link seemingly different resources. In fact, one way to generate links for FOAF triples is when two resources share the same value for the foaf:mbox property, these two resources must be identical, even the URIs used to identify them are different.

Write a SPARQL construct query to generate owl:sameAs statements based on this property. Make sure your solution only generates triples where the subject and object are different (i.e., non-trivial owl:sameAs statements).

What are the pros and cons of generating links as in #1 above? In your answer, consider both precision (what percentage of the generated links are correct) and recall (what percentage of the needed links are generated by the method).