Executing SPARQL Queries over Mapped Document Stores with SparqlMap-M

Executing SPARQL queries over Mapped DocumentStores with SparqlMap-M

J. Unbehauen M. Martin

IIS // AKSW // BIS // IfILeipzig University

SEMANTiCS 2016

J. Unbehauen, M. Martin (Leipzig Univ.) SPARQL over Document Stores: SparqlMap-M SEMANTiCS 2016 1 / 25

Outline

1 Motivation and Scope

2 Approach

3 Evaluation

4 Conclusions and Future Work

Scoping

[1] S. Auer, J. Lehmann, A. Ngonga Ngomo. Introduction to Linked Data and ItsLifecycle on the Web, Reasoning Web. Semantic Technologies for the Web of

Data, LNCS 6848, 2011

Motivation

NoSQL DBMS and document stores are thriving

Document stores used in Rapid Application Development Frameworks

Visit our Adding Semantics to Model-Driven Software DevelopmentPoster

Use cases in both research and industry

Current solutions support R2RML and relational databases

Motivation

Outline

2 Approach

3 Evaluation

SparqlMap Architecture

BindingTranslat.

SparqlMap

QueryAnalysis

QueryParsing

MappingBinding

QuerySELECT DISTINCT ?name {

?person foaf:name ?name. #(tp1)

?person :inDepartment ?dep. #(tp2)

?dep rdfs:label ’Research’ #(tp3) }

Result?name

------------

’Mary R.’

’James T.’

Translat.Exec.

[2] J. Unbehauen, C. Stadler, and S. Auer. Accessing relational data on the webwith sparqlmap. In JIST. 2012.[3] J. Unbehauen, C. Stadler, and S. Auer. Optimizing sparql-to-sql rewriting. InIIWAS, 2013.

SparqlMap-M Architecture

BindingTranslat.

SparqlMap-MSparqlMap

QueryAnalysis

QueryParsing

MappingBinding

SelectiveMaterialization

?dep rdfs:label ’Research’ #(tp3)

}Mapping

DeduplicationUnion Decom-

position

MaterializedExecution?name

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

1 Data Models and Mapping

2 Query Structure

3 Querying Capabilities

4 Data Model Specific Optimization

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

position

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

2 Query Structure

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

position

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

2 Query Structure

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

position

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

2 Query Structure

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

position

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

2 Query Structure

Data Models and Mapping

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

Deduplication

Union De-composition

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

Key-Value pairs

Nested documents

Schema less

A relational view on documents by:

Goal: reuse existing (R2RML) concepts

Unnesting documents by joining them with parent → Flat structure

Naming attributes to reflect hierarchy → Key-Value treated as tuples

Schema imposed by mapping

#Department{ i d : 2 , name : ” Resea rch ” ,emp : [{ i d : 1 , name : ”Mary R. ”} ,

{ i d : 2 , name : ”James T. ” } ] } ,

Query Structure

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

Deduplication

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

Query Structure

SparqlMap

Recursive translation yields nested unions

Index hits require careful query design

Complex expressions for joins

SparqlMap-M / MongoDB

No direct equivalents for joins

No complex equivalence expression

Query Structure

SparqlMap

Recursive translation yields nested unions

Index hits require careful query design

Complex expressions for joins

SparqlMap-M / MongoDB

No direct equivalents for joins

No complex equivalence expression

Query Structure: Union Decomposition

Nested Unions:

./?dep=?dep

σname=Research

./?person=?person⋃trm1 trm4

⋃trm2 trm5

Pushed Union: ⋃./?dep=?dep

trm3 ./?person=?person

trm1 trm2

./?dep=?dep

trm4 trm5

Query Structure: Union Decomposition

Nested Unions:

./?dep=?dep

σname=Research

./?person=?person⋃trm1 trm4

⋃trm2 trm5

Pushed Union: ⋃./?dep=?dep

trm1 trm2

./?dep=?dep

trm4 trm5

Selective Materialization

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

Deduplication

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

Selective Materialization

Delegate to abstraction layer (Apache MetaModel)

Execute unpushable SPARQL operators in memory

Πname

./id=depid

σname=”Research”

department employee

MaterializedExecution

De-Duplication

BindingTranslat.

QueryAnalysis

QueryParsing

MappingBinding

}Mapping

Deduplication

------------

’Mary R.’

’James T.’

Translat.Exec.

Result

De-Duplication

Documents are nested for fastretrieval and filtering

Naive mapping introduces overhead

Declaratively labelR2RML-TriplesMaps as duplicated

Only use denormalized data in joins

De-Duplication

Documents are nested for fastretrieval and filtering

Naive mapping introduces overhead

Declaratively labelR2RML-TriplesMaps as duplicated

Only use denormalized data in joins

Outline

2 Approach

3 Evaluation

Benchmark Setup

BSBM for availability of both SQL and RDF representation

SQL representation translated into MongoDB documents

Additionally performed denormalization

Benchmark Results

BSBM 10 million triples

PostgreSQL Fastest

MongoDB-Naive/-Dup Dup required for performance

SparqlMap-M-Naive/ -Dup/ -DupAwareOverhead by rewriting/materialization

Benchmark Results

BSBM Q4

Medium selectivity

Naive modes touch a lot of data

Performance gain by duplicatedata (MongoDB, SparqlMap-M)

Benchmark Results

BSBM Q5

Low selectivity join

SparqlMap-M: expensive selfjoin in memory, dominates cost

MongoDB: Self-join inaggregate pipeline, slower thanPostgreSQL

BSBM Q9

High selectivity join

SparqlMap-M-Dup(Aware):duplicates increase overhead.Unpushable join dominates cost

Outline

2 Approach

3 Evaluation

Future Work

Enable Updates

Integrate Caching

Evaluate Join capable query language

MongoDB left outer join ($lookup)Multimodel databases: ArangoDB, OrientDBDB virtualizations: JBoss Teiid, Apache HAWQ

Conclusion

Architecture for a SPARQL execution layer over document stores

Harness duplicates for increasing performance

Evaluated with BSBM on MongoDB

Executing SPARQL Queries over Mapped Document Stores with SparqlMap-M

Data & Analytics

Practical SPARQL Benchmarking

C-SPARQL: A Continuous Extension of...

RDF and SPARQL - Part IV: Syntax of SPARQL

How good is your SPARQL endpoint? A QoS-Aware SPARQL...

Requêtes sparql

An Introduction to the SPARQL RDF Query...

Extending SPARQL Algebra to Support Efficient Evaluation of....

Querying Wikidata: Comparing SPARQL, Relational...

x Sparql PDF

Repositories and SPARQL

S Stavros Christodoulakis ACM - NTUAbikakis/papers/SPARQL-RW...

SPARQL Endpoint

SPARQL & JENA

SPARQL - Free University of...

SPARQL 1.1

OWL & SPARQL -...