A Path Querying Language for Federation of RDF and Relational - PowerPoint PPT Presentation

A Path Querying Language for Federation of RDF and Relational Database Jiahui Zhang, Xiaowang Zhang, and Zhiyong Feng Tianjin University zhangjiahui@tju.edu.cn xiaowangzhang@tju.edu.cn zyfeng@tju.edu.cn

Outline n Introduction n Federated Path Querying Language(FPQ) n Expressiveness of FPQ n Experiment and Evaluation

Introduction n SPARQL The standard language for querying RDF data since 2008. n Navigational Capability Versa: using XPath over the XML of RDF graphs SPARQLeR: adding path variables CPSPARQL: allowing constraints over regular expressions nSPARQL: applying nested regular expressions SPARQL 1.1: appending property paths

Introduction Q : Go from Rouen to Reims. Rouen Rouen→Dreux→Paris→Reims Dreux Paris Reims Driver Rouen→Paris→Reims Figure: Geographical map.

Introduction Q' : Go from Dreux to Paris. Rouen more orders Rouen→Dreux→Paris→Reims Dreux Paris Reims Rouen→Paris→Reims shorter Figure: Geographical map. ID Time Passenger Driver Start Point End Point 1 6:40 a.m. B E Rouen Reims 2 6:50 a.m. C ? Dreux Paris Table: Taxi-hailing orders.

Introduction Rouen n RDF Rouen→Paris→Reims n RDF+Relational Database Dreux Paris Reims Rouen→Dreux→Paris→Reims Figure: Geographical map. ID Time Passenger Driver Start Point End Point 1 6:40 a.m. B E Rouen Reims 2 6:50 a.m. C E Dreux Paris Table: Taxi-hailing orders.

Outline n Introduction n A Path Querying Language for Federation of RDF and Relational Database n Expressiveness of FPQ n Experiment and Evaluation

FPQ n Federated Path Querying Language(FPQ) n Conjuction     q ( u , v ) : ( u , e , v ); i i i  i 1 where • q is the name of FPQ;  • is a conjunctive combination of relations; • each for is a NRE triple pattern. i  { 1 ,..., n } ( u , e , v ) i i i

FPQ n Query: At a certain time, whether a passenger take a ride?      q (? x , ? y ) [ R ( v ) R ( e )] [(? x , exp , ? y ) (? x , exp , ? y ) (? x , exp , ? y )] 1 2 3 Conjuction where • R(v) := Position(Time, ?x, ?y, ?driverId); • R(e) := Orders(Time, lonUp, latUp, ?x, ?y, ?driverId); • exp1 : next -1 :: lon; • exp2 : next :: lat; • exp3 : next :: nd / next :: ref / next -1 :: id.

Outline n Introduction n Federated Path Querying Language(FPQ) n Expressiveness of FPQ n Experiment and Evaluation

Expressiveness of FPQ | : Disjunctive operator; N : Nesting operator; ∧ : Conjunctive operator; R : Federated operator; UFCNRPQ ∨ : Union Operator. FCNRPQ CNRPQ NRPQ The query evaluation of FCNRPQ: Data complexity → Polynomial time Combined complexity → NP-complete time

Outline n Introduction n Federation Path Querying Language n Expressiveness of FPQ n Experiment and Evaluation l Query and Result l Extention

Query n Query 1: At a certain time, where did the passengers get on vehicles or off? n Query 2: At a certain time, did the passengers visit tourist attractions on the map? n Query 3: At a certain time, which roads did the passengers go down from the vehicles?

Query Rouen n Query 1: Dreux Paris Reims At a certain time, where did the Figure: Geographical map. passengers get on vehicles or off? n Query 2: RDF At a certain time, did the passengers visit tourist attractions on the map? n Query 3: Conjuction At a certain time, which roads did the passengers go down from the vehicles? ID Time Passenger Driver Start Point End Point Relational Database 1 6:40 a.m. B E Rouen Reims 2 6:50 a.m. C E Dreux Paris Table: Taxi-hailing orders.

Result

Extention n Distributed environment to solve the memory overflow n Streaming data to process the real time query

Q & A n Jiahui Zhang zhangjiahui@tju.edu.cn n Xiaowang Zhang xiaowangzhang@tju.edu.cn n Zhiyong Feng zyfeng@tju.edu.cn Thank you!