Tutorial "Linked Data Query Processing" Part 1 "Introduction" (WWW 2013 Ed.)

Linked Data Query Processing
Tutorial at the 22nd International World Wide Web Conference (WWW 2013)
May 14, 2013
http://db.uwaterloo.ca/LDQTut2013/
Olaf Hartig
University of Waterloo

WWW 2013 Tutorial on Linked Data Query Processing [ Introduction ] 2
Tutorial Outline
(1) Introduction
(2) Theoretical Foundations
(3) Source Selection Strategies
(4) Execution Process
(5) Query Planning and Optimization

Tutorial at the 22nd International World Wide Web Conference (WWW 2013)
May 14, 2013
http://db.uwaterloo.ca/LDQTut2013/
1. Introduction
Olaf Hartig
University of Waterloo

Outline
 The Linked Data Principles
 Paradigms for Querying Linked Data
 Characteristics of the “Database System”

The Traditional, Hypertext Web
MovieDB
Data exposed
to the Web
via HTML
CIA World
Factbook

Towards a Web of Linked Data
MovieDB
:
( Albania , unemployment rate , 13.2% )
:
Data model: RDF
( War Child , release date , 12 July 1999 )
( War Child , filming location , Albania )
( Michael Davie , directed , War Child )
:
CIA World
Factbook

MovieDB
( http://...imdb.../WarChild , release date , 12 July 1999 )
( http://...imdb.../WarChild , filming location , http://cia.../Albania )
( http://...imdb.../MichaelDavie , directed , http://...imdb.../WarChild )
:
( http://cia.../Albania ,
unemployment rate , 13.2% )
:
Data model: RDF
Global identifier: URI
CIA World
Factbook

MovieDB
:
Data model: RDF
Access mechanism: HTTP
:
CIA World
Factbook

MovieDB
CIA World
Factbook
:
:
Data model: RDF
Access mechanism: HTTP
Connection: data links

Supplementary Access Methods
● RDF dump: the whole dataset provided as a big file
● SPARQL endpoint: Web service that allows for executing
SPARQL queries over the dataset
● Caveat: these access method cannot be assumed
to be available for all datasets
● Creating dumps is not feasible if data changes very frequently
● Dumps or endpoints may not be feasible if Linked Data
interface is simply a wrapper for some other data source
● Providing and maintaining a reliable SPARQL endpoint
is a significant additional effort

Outline
√

Traditional Paradigm 1: Warehousing
● Copy data into a centralized repository
● Query this repository
+ Almost instant results
– Misses unknown or new sources
– Collection possibly out of date

?
??
?
Traditional Paradigm 2: Federation
● Distribute query execution over a
federation of SPARQL endpoints
+ Current data
– Misses sources without
SPARQL endpoint

Principle 1: Rely on the Linked Data principles only
Principle 2: On-line execution

Principle 1: Rely on the Linked Data principles only
Principle 2: On-line execution
Consequence: Obtain data for executing a given query by
looking up URIs during the query execution process itself

“Ingredients” for LD Query Execution
Query-local data

● Data retrieval approach
● Data source selection
● Data source ranking
(optional, for optimization)
Query-local data
GET http://.../movie2449

Query-local data
http://mdb.../Paul http://geo.../Berlin
http://mdb.../Ric http://geo.../Rome
?loc?actor
● Result construction approach
● i.e., query-local data processing

Query-local data
http://mdb.../Paul http://geo.../Berlin
http://mdb.../Ric http://geo.../Rome
?loc?actor
● Result construction approach
● i.e., query-local data processing
● Combining data retrieval
and result construction

+ Current data
+ May make use of any Linked Data available on the Web
– Least efficient due to data shipping
Use cases: live querying where freshness and discovery of
results is more important than an almost instant answer
Properties of LD Query Processing

Combination with other Paradigms
● Linked Data query processing with a query-local dataset
● Query-local dataset contains additional data [LT11]
● Query-local dataset for caching [Har11b, HH11]
● Linked Data query processing with a SPARQL endpoint
● SPARQL endpoint exposes a cache of Linked Data [UKH+12]

Our Topic Today …
… pure Linked Data query processing
Linked Data query: a query that ranges over
data made available using
the Linked Data principles
Web of Linked Data: network of data that evolves
by publishing data according
to the Linked Data principles

Outline
√
√

An Analogy ...

Traditional, Central Database Systems

Distributed Database Systems

The Web of Linked Data

● Number of
potential data
sources infinite

● Number of
potential data
sources infinite
● No (a priori)
information

● Number of
potential data
sources infinite
● No (a priori)
information
● Number of
actual data
sources infinite

Issues due to the Openness
● Data quality issues
● Accuracy
● Freshness / timeliness
● Believability / trustworthiness
● Data source quality issues
● Availability
● Reliability
● Data integration issues
● Coreferences: Publishers may use different URIs
for denoting the same entity
● Schema heterogeneity: Publishers may use different
vocabularies for their data
For the purpose of discussing
execution of queries in this tutorial,
we largely ignore these issues.

Outline
√
√
√
Next part: 2. Theoretical Foundations ...

These slides have been created by
Olaf Hartig
for the
WWW 2013 tutorial on
Link Data Query Processing
Tutorial Website: http://db.uwaterloo.ca/LDQTut2013/
This work is licensed under a
Creative Commons Attribution-Share Alike 3.0 License
(http://creativecommons.org/licenses/by-sa/3.0/)

Tutorial "Linked Data Query Processing" Part 1 "Introduction" (WWW 2013 Ed.)

More Related Content

What's hot

Similar to Tutorial "Linked Data Query Processing" Part 1 "Introduction" (WWW 2013 Ed.)

More from Olaf Hartig

Recently uploaded

Tutorial "Linked Data Query Processing" Part 1 "Introduction" (WWW 2013 Ed.)