Use Jupyter Notebook with an Ontop SPARQL endpoint

This tutorial shows how to use a Python/Jupyter Notebook to interact with an Ontop SPARQL endpoint.

Link to the Jupyter Notebook: ontop-jupyter.ipynb

Assume that the endpoint is already set up and the URL is http://localhost:8080/sparql (opens new window).

SPARQLWrapper

You can use the SPARQLWrapper library (opens new window) to send SPARQL queries and get results. The following code gets the result as JSON documents and convert it to a Python dict object.

from SPARQLWrapper import SPARQLWrapper, JSON
q = """
  PREFIX : <http://example.org/voc#>
  PREFIX foaf: <http://xmlns.com/foaf/0.1/>
  SELECT ?lname ?title 
  WHERE {
    ?attendee foaf:lastName ?lname ; :attends ?course .
    ?course :title ?title .
  }
"""
sparql = SPARQLWrapper("http://localhost:8080/sparql")
sparql.setQuery(q)
sparql.setReturnFormat(JSON)
results = sparql.query().convert()
print(results)

Running the code above yields the following output (JSON indentation added for clarity):

{
   "head":{
      "vars":[ "lname", "title" ]
   },
   "results":{
      "bindings":[ {
         "lname":{ "type":"literal", "value":"Robards" },
         "title":{ "type":"literal", "value":"Data Mining" }
      }, {
         "lname":{ "type":"literal", "value":"Smith" },
         "title":{ "type":"literal", "value":"Operating Systems" }
      }, {
         "lname":{ "type":"literal", "value":"Doe" },
         "title":{ "type":"literal", "value":"Linear Algebra" }
      }, {
         "lname":{ "type":"literal", "value":"Combs"  },
         "title":{ "type":"literal", "value":"Operating Systems" }
      }, {
         "lname":{ "type":"literal", "value":"Doe" },
         "title":{ "type":"literal", "value":"Analysis" }
      }, {
         "lname":{ "type":"literal", "value":"Hinkley" },
         "title":{ "type":"literal", "value":"Data Mining" }
      }, {
         "lname":{ "type":"literal", "value":"Robards" },
         "title":{ "type":"literal", "value":"Research Methods" }
      }, {
         "lname":{ "type":"literal", "value":"Smith" },
         "title":{ "type":"literal", "value":"Analysis" }
      }, {
         "lname":{ "type":"literal", "value":"Smith" },
         "title":{ "type":"literal", "value":"Linear Algebra" }
      }, {
         "lname":{ "type":"literal", "value":"Combs" },
         "title":{ "type":"literal", "value":"Linear Algebra" }
      }, {
         "lname":{ "type":"literal", "value":"Hinkley" },
         "title":{ "type":"literal", "value":"Theory of Computing" }
      }, {
         "lname":{ "type":"literal", "value":"Alfaro" },
         "title":{ "type":"literal", "value":"Introduction to programming" }
      }, {
         "lname":{ "type":"literal", "value":"Hinkley" },
         "title":{ "type":"literal", "value":"Software factory" }
      }, {
         "lname":{ "type":"literal", "value":"Mendez" },
         "title":{ "type":"literal", "value":"Software factory" }
      }, {
         "lname":{ "type":"literal", "value":"Mendez" },
         "title":{ "type":"literal", "value":"Software process management" }
      }, {
         "lname":{ "type":"literal", "value":"Moses" },
         "title":{ "type":"literal", "value":"Information security" }
      }, {
         "lname":{ "type":"literal", "value":"Moses" },
         "title":{ "type":"literal", "value":"Discrete mathematics and logic" }
      } ]
   }
}

Pandas DataFrame

You might want to convert the SPARQL results to a pandas DataFrame for data analysis. The library sparql-dataframe (opens new window) is handy for this.

import sparql_dataframe
endpoint = "http://localhost:8080/sparql"
q = """
    PREFIX : <http://example.org/voc#>
    PREFIX foaf: <http://xmlns.com/foaf/0.1/>
    SELECT ?lname ?course_title 
    WHERE {
      ?attendee foaf:lastName ?lname ;
          :attends ?course .
      ?course :title ?course_title .
    }
"""
df = sparql_dataframe.get(endpoint, q)

The result df can be manipulated as any Pandas DataFrame. For instance:

df.head()

	lname	course_title
0	Robards	Data Mining
1	Smith	Operating Systems
2	Doe	Linear Algebra
3	Combs	Operating Systems
4	Doe	Analysis

df.describe()

	lname	course_title
count	17	17
unique	8	11
top	Smith	Linear Algebra
freq	3	3

df['course_title'].value_counts().plot(kind='bar')

<matplotlib.axes._subplots.AxesSubplot at 0x11475ebd0>