Rdf conjunctive query selectivity estimation
- 1. RDF conjunc*ve query cardinality es*ma*on Owner : Stama,s/Damian Presenter : Soudip
- 2. Overview • This project provides methods to es,mate the cardinality of (for the result of) a conjunc,ve query • It requires a summary with sta,s,cs informa,on that can be provided as a serialized summary or by providing the database connec,on containing sta,s,cs and dic,onary tables informa,on • This project also include methods to generate a dic,onary-‐ encoded version of the triples table and the triples table sta,s,cs in a database, for both the plain and the dic,onary-‐ encoded triples tables • It is a refactored extrac,on (for code reusability) of the RDFViewSelec,on’s (project) cardinality es,mator
- 3. Few Details • Online repository: – hLps://scm.gforge.inria.fr/svn/distriples/RDFOp,m • Code size (java) – 4041 (LoC), 15 packages • List of people contributed-‐ – Present: Stama,s, Damian, Ioana – Past: Julien Leblay • Current Owner (OAK member) of the Code-‐ Stama,s/Damian • Who is using the code now – Fragmented Query Execu,on (Damian) – CliqueSquare (Stama,s) – Op,mizer (hLps://scm.gforge.inria.fr/svn/distriples/trunk/Op,mizer) (Stama,s/Zoi)
- 4. Func*onal Architecture RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary
- 5. RDF Loader RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary – Parses input RDF files content – Extracts <subject, property, object> triples – Loads the triples into the DB
- 6. Data Store RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary – Stores triples in the DB and creates 3 different tables – Data Table • Stores the basic triples – Summary Table • Stores different summaries of triples • 6 different summary tables – Dic,onary Table • Stores integer values corresponding to each entries in the data table
- 7. Cardinality Es*mator Module RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary – Takes input • A conjunc,ve query • Data from Summary and Dic,onary tables – Outputs cardinality informa,on for the input query
- 8. CQ Parser RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary – Taken as it is from the project Conjunc,ve Query
- 9. DB Interfaces RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary – Contains two interfaces to load data from the DB to the memory • Summary Interface – to get data from the Summary table • Dic,onary Interface -‐ to get data from the Dic,onary table
- 10. Cardinality Es*mator RDF RDF Loader Parser Importer Conjunctive Query CQ Parser DB Interfaces Cardinality Info. Main Modules Cardinality Estimator Triples Summary, Dictonary Data Store (PostgreSQL DB) Data Table Summary Table (6) Dictonary Table S P O S Key Value Count (*) Count (P) Count (O) Min (P) Max (P) Min (O) Max (O) P Count (*) Count (S) Count (O) Min (S) Max (S) Min (O) Max (O) O Count (*) Count (S) Count (P) Min (S) Max (S) Min (P) Max (P) S P Count (*) Count (O) Min (O) Max (O) P O Count (*) Count (S) Min (S) Max (S) S O Count (*) Count (P) Min (P) Max (P) CQ, Summary, Dictonary – Takes as input CQ and Summary data and produces the cardinality info for the input CQ as output – It uses 3 algorithms (two use sta,c summary and one uses MaxMin Summary) for calcula,ng the cardinality (hLps://scm.gforge.inria.fr/svn/distriples/RDFOp,m/)
- 11. Thank you!!