linked data+top-k query processing

7/27/2019 Linked Data+Top-K Query Processing

http://slidepdf.com/reader/full/linked-datatop-k-query-processing 1/24

Top-k Linked D

Query Process



• Linked Data is about using the Web to connect related data that w

previously linked.

Linked Data - Definition



• Use URIs to define things

• Use HTTP URIs so that these things can be referred to andup

• Provide useful information in RDF – when someone looks

• Include RDF links to other URIs – to enable discovery of reinformation

Linked Data - Principle

http://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol

http://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol



• URIs

– Global unique identifiers for entites

– Pointers to data

• HTTP to access data on the Web

• RDF as a share data model

• FORMATS ( RDF/XML, RDFa,…) / HYPERLINKS

Linked Data - Component



• A resource is basically everything

– E.g. persons, places, Web documents, abstract concepts

• Descriptions of resources

– Attributes

– Relations

• The framework contains:

– A data model

– Languages and syntaxes

RDF



• Data comes as a set of triples (subject, predicate, objec

• Subject: resources

• Predicate: properties

• Object: literals or resources

• Examples: – ( Mount Baker , last eruption , 1880 )

– ( Mount Baker , location , Washington )

RDF Data Model



• RDF is also a graph model

– Triples as directed edges

– Subjects and objects as vertices

– Edges labeled by predicate

• Example:

– ( Mount Baker , last eruption , 1880 ) – ( Mount Baker , location , Washington )

RDF Data Model



• URIs extend the concept of URLs

– Globally unique identifier for resources

– URL of a Web document usually used as its URI

– Attention: URIs identify not only Web documents

• Example:

– Me: http://olafhartig.de/~hartig/foaf.rdf#olaf – RDF document about me: http://olafhartig.de/~hartig/foaf.rdf

– HTML document about me: http://olafhartig.de/~hartig/index.h

URI

http://olafhartig.de/~hartig/foaf.rdf










URI



• Literals may occur in the object position of triples

• Represented by strings

• Literal strings interpreted by datatypes

– Datatype identified by a URI

– Common to use the XML Schema datatypes

– No datatype: interpreted as xsd:string

– Untyped literals may have language tags (e.g. @de)

Literal



• Blank nodes represent unnamed, anonymous resources

– Not identified by a URI

• Blank node identifiers

– Identification of blank nodes in triple serializations

• Form: _:xyz

• Scope: a single RDF graph

Blank Nodes



• Extends the definition of RDF nodes and RDF triples

– RDF node: I, B, and L, which are pair-wise disjoint infinite sets of IResource Identifiers (IRIs), blank nodes and literals

– RDF triple: (s, p, o) ∈ IB × I × IBL, where IL = I ∪ L, IB = I ∪ B and IBL

Linked Stream Data model



• White box architecture

– Implements all required components

– physical operators (e.g. windows, join, triple pattern matching)

– data structures (e.g. B+-Trees, hashtables)

– query generator/optimizer/executor

•

Black box architecture – Uses existing RDF and data stream processing systems as sub-co

– Query rewriter, data translator and orchestrator among subcomneeded

System architecture



<?xml version="1.0" encoding="UTF-8"?>

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"

xmlns:foaf="http://xmlns.com/foaf/0.1/"

xmlns:dc="http://purl.org/dc/elements/1.1/"

xmlns:ex="http://example.com/">

<foaf:Band rdf:about="http://example.com/Beatles/">

<foaf:name>Beatles</foaf:name>

<ex:album rdf:about="http://example.com/Beatles/Sgt_Pepper/">

<ex:name>Sgt_Pepper</ex:name>

<ex:song rdf:about="http://example.com/Beatles/Sgt_Pepper/Lucy">

<ex:name>Lucy</ex:name>

</ex:song>

</ex:album>

<ex:album rdf:about="http://example.com/Beatles/Help!/">

<ex:name>Help!</ex:name>

<ex:song rdf:about="http://example.com/Beatles/Help!/Help!">

Data model



-Get all songs from Beatles’s album

PREFIX ex: http://example.com/

PREFIX foaf: http://xmlns.com/foaf/0.1/

PREFIX : http://example.com/resource/

SELECT * WHERE

ex:beatle ex:album ?album

?album ex:song ?song

}

Query

http://example.com/

http://xmlns.com/foaf/0.1/

http://example.com/resource/

http://example.com/resource/

http://xmlns.com/foaf/0.1/

http://example.com/



• Purpose

– Efficiency and scalability are essential problems in the Linked D

– Instead of computing all results, top-k query processing approacproduce only the “best" k results

Top-k Linked Data Query Proces



• Source index

– Map or match triple pattern to sources containing bindings

Top-k Linked Data Query ProcessingRequirement (just use for binary op

Linked DataQuery ProcessingEngine

ex:sgt_pepper foaf:n

"Sgt. Pepper";

ex:song "Lucy".

Src.2

ex:help foaf:name

"Help!";

ex:song "Help!".

Src.3ex:beatles foaf:name

"The Beatles";

ex:album ex:sgt_pepper;

ex:album ex:help.

Src.1

TP1: ex:beatles ex:album ?album .

TP2: ?album ex:song ?song .

sourceindex



• Ranking function

– Determining the relevance of triple pattern bindings

– For instance, scores for triples can be obtained through PageRank ranking

– However, no triples are indexed (i.e., each source must be scanne




• Sorted access


TP2: ?album ex:song ?song

Src.2

TP1:

ex:beatles ex:album ?album

Bindidesc score

Sche

Str

Src.3

2

Src.1

1



• Push Based Rank Join


Sorted Access forex:beatles ex:album ?album .

Sorted Access for?album ex:song ?

Score Query Bindings – Output Queue

ex:beatles foaf:name

"The Beatles";

ex:album ex:sgt_pepper;

ex:album ex:help.

Src.1 ex:help foaf:name

"Help!";

ex:song "Help!".

Src.3

Score Seen Triples (TP2)

3 ex:help ex:song "Help!"

2 ex:sgt_pepper ex:song

“Lucy“ (skip because of

score 2 <3, just only pus

“help!” )


1 ex:beatles ex:album

ex:sgt_pepper

1 ex:beatles ex:albumex:help (because

1=1 so choose both

of them)



• Push Based Rank Join



4ex:beatles ex:album ex:help.ex:help ex:song "Help!" .

Threshold: 4(max (1+3, 1+3))

Sorted Access forex:beatles ex:album ?album .

Sorted Access fo?album ex:song


3 ex:help ex:song "Help



ex:sgt_pepper

1 ex:beatles ex:albumex:help

Found query bindingwith score ≥ threshold

STOP



• Improving threshold estimation

– Origin threshold estimation:

– How to improve:

– Star-shaped entity query bounds

– Look-ahead bounds


Threshold: max { max_1 + min_2, max_2 + min_1}



• Star-shaped entity query bounds

– Problem:

– In Linked Data query processing, every result for entity queries to suis contained in one single source.

– Reason:

– A result here is an entity,

– Information related to that entity comes exclusively from the one sorepresenting that particular entity.

– Idea:

– Upper bound scores for triple pattern bindings via the maximal posscore




• Look-ahead Bounds:

– Provide a more accurate upper bound for the unseen bindings

the next possible score


ma

mi

Threshold: max { 1 + 2 , 1 + 3 } = 4


3 ex:help ex:song "Help!"

Sorted Access for?album ex:song ?song

Src. 2

S

mi

max_1 = 1

min_1 = 1



ex:sgt_pepper


ex:help

Sorted Access for


4 ex:beatles ex:album ex:help .ex:help ex:song "Help!" .

linked data+top-k query processing

Documents