SAFE: Policy Aware SPARQL Query Federation Over RDF Data Cubes
Download as PPTX, PDF1 like495 views
- Dr. Ratnesh Sahay presented on the Policy Aware SPARQL Query Federation Over RDF Data Cubes project called SAFE at the SWAT4LS-2014 conference in Berlin, Germany. - SAFE allows for querying across multiple RDF data cube datasets while respecting access policies through policy-aware query rewriting and execution. - An evaluation showed SAFE performed more efficiently than existing federated query systems, with significantly faster source selection times and ability to execute queries that timed out on other systems.
SAFE: Policy Aware SPARQL Query Federation Over RDF Data Cubes
- 1. 10-12-2014 SAFE Policy Aware SPARQL Query Federation Over RDF Data Cubes Dr. Ratnesh Sahay Semantics in eHealth & Life Sciences (SeLS) Insight Centre for Data Analytics NUI Galway, Ireland SWAT4LS-2014, Berlin Germany
- 2. Enabling networked knowledge 2 Linked2Safety - Showcases 1. Showcase #1 – Phase III Clinical Trial: Subject Selection Criteria: the unbiased randomised selection of subjects in phase III clinical trials e.g. return subjects with diabetesValue > 4 and weight >80 and hasCancer 2. Showcase #2 – Phase IV Post Marketing Surveillance trial: the pharmacovigilance of a drug after it receives permission to be sold e.g. Test Drugx association with headaches 3. Showcase #3 – Chemoinformatics: identification of relations between molecular fragments and specific adverse side effect categories. e.g. Test chemicalFragmentX(of DrugX) with rash
- 3. Enabling networked knowledge 3 The Problem return number of patients that have been administered the drug Insulin and exhibit BMI > 25 and Hypertension and Diabetes as adverse events Switzerland Cyprus Greece
- 4. Safety First – Ethical & Legal Aspects Enabling networked knowledge 4 Patients’ anonymity Data Ownership & Privacy Anonymised Clinical Data cubes Insensitive clinical parameters without personal information Access-Control Based Query Federation
- 5. CING (Data Cubes) Enabling networked knowledge 5 SAFE - Secure SPARQL Query Federation CHUV (Data Cubes) ZEINCR O (Data Cubes) RDF Data Cubes Index RDF Data Cubes RDF Data Cubes Access Policy Model SPARQL Query Source Selection Access Policy Filter Query re- Writer Results
- 6. Enabling networked knowledge 6 SAFE SPARQL Query + User Info Source Selection Access Policy Filtering Query Re-writing Oya Clinical Researcher Expertise – Diabetes SELECT ?diabetes ?bmi ?hypertension ?cases WHERE { ?observation a qb:Observation . ?observation l2s-dim:Diabetes ?diabetes. ?observation l2s-dim:BMI ?bmi. ?observation l2s-dim:Hypertension ?hypertension. ?observation sdmx-measure:Cases ?cases. }
- 7. S3={ D ia b e t e s , B M I , H y p e r t e n s i o n , H I V , C }ases Enabling networked knowledge 8 SAFE – Source Selection SPARQL Query + User Info Source Selection Access Policy Filtering Query Re-writing Triples Patterns ?observation l2s-dim:Diabetes ?diabetes. ?observation l2s-dim:BMI ?bmi. ?observation l2s-dim:Hypertension ?hypertension. ?observation sdmx-measure:Cases ?cases. Capable Sources {S1, S2, S3, } {S1, S2, S3} {S1, S2, S3} {S1, S2, S3 , S4} S1={ } S4={ Smoking, Gender, Cases } INDEX Diabetes, BMI, Hypertension, Cases S2={ } Diabetes, BMI, Hypertension, Cases Diabetes, S4 Join Awareness
- 8. Enabling networked knowledge 9 SAFE – Access Policy Access Policy Framework SPARQL Query + User Info Triple Pattern-based Source Selection Access Policy Filtering Query Re-writing Oya Clinical Researcher Expertise – Diabetes Requested Data S1 S2 S3 Input Input Grants Access Denies Access S1 S2 S3
- 9. Enabling networked knowledge 10 SAFE – Access Policy • Example Access Policy AP1 type Access_Policy AP1 applies_to {S1, S2} AP1 grants_access Read AP1 assigned_to Oya SPARQL Query + User Info Source Selection Access Policy Filtering Query Re-writing Oya type User Oya haslocation Galway Oya hasPurpose Perform p-value analysis Oya hasRole Clinical Researcher Oya hasDomain Diabetes • SPARQL Query ASK WHERE { ?accessPolicy a AccessPolicy. ?accessPolicy appliesToNamedGraph S1. ?accessPolicy :grantsAccess rantsAccess acl:Read_l2s, ?accessPolicy hasUser Oya. }
- 10. Enabling networked knowledge 11 SAFE – Query Rewriting SPARQL Query + User Info Graph Information will be added to the query triples SELECT …. WHERE { GRAPH <S1> { …. } } SELECT …. WHERE { GRAPH <S2> { …. } } Sub queries sent to relevant sources S1 S2 Integration of results obtained from each sources Source Selection Access Policy Filtering Query Re-writing Diabetes BMI Hypertension Cases 0 0 0 40 1 0 1 50 0 1 1 120 1 1 1 90 S1 S2
- 11. Enabling networked knowledge 12 Evaluation - DataSets Dataset # triples # obs # sub # pred # obj # size # index size # index generation time Internal Dataset CHUV 0.8 M 96 K 96 K 36 88 31 MB - - CING 0.1 M 17 K 17 K 21 51 5 MB - - ZEINCRO 0.4 M 49 K 49 K 24 59 15 MB - - Total 1.3 M 162 K 162 K 81 198 51 MB 8 KB 10 sec External Dataset World Bank 77 M 10 M 10 M 58 40 K 19 GB - - IMF 18 M 1.8 M 1.8 M 30 3151 3.51 GB - - Eurostat 0.3 M 38 K 44 K 31 5717 205 MB - - Trans. Int. 43 K 3939 4286 64 5290 9.2 MB - - Total 95 M 12 M 2 M 183 54 K 23 GB 12 KB 571 sec
- 12. Enabling networked knowledge 13 Evaluation - Berlin SPARQL Benchmark Characteristics Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 # of Triple Patterns 9 7 9 16 7 8 11 10 7 7 3 7 # of Sources 3 4 4 3 4 3 3 4 3 3 3 3 # of Results 41 50 348 41 62 1983 5 10 1701 19656 570 41 Filters > 9 Patterns Negation LIMIT Modifier Order By Modifier DISTINCT Modifier REGEX Operator UNION Operator
- 13. • Sum of triple-pattern-wise sources selected for each query • Number of SPARQL ASK requests used for source selection Enabling networked knowledge 14 Evaluation – Source Selection Systems Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Avg SAFE 8 10 13 16 15 13 15 16 7 7 9 7 11 FedX 9 13 16 24 20 14 16 19 15 17 9 16 16 Systems Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Avg SAFE 0 0 0 0 0 0 0 0 0 0 0 0 0 FedX 36 28 40 64 48 40 44 40 21 21 9 21 35
- 14. Enabling networked knowledge 15 Evaluation - Source Selection Time • Source Selection Time
- 15. Enabling networked knowledge 16 Evaluation - Query Execution Time • Query Execution Time Query times-out for FedX
- 16. Enabling networked knowledge 17 SAFE – Highlights Source Selection SPARQL SERVICE Using SPARQL ASK queries Using a catalog/index SAFE - Hybrid (catalog/index + ASK) Lightweight Cache – RDF Cube Data Structure – AccessPolicy Join Aware Excludes ineligible sources before actual query join Provenance – via RDF Named Graphs Self-contained Data Cubes Creator Location Date Access rights
- 17. Conclusion & Future Work Enabling networked knowledge 18 Efficient source selection with a lightweight indexing Policy aware query execution Evaluated against internal and external sets Performance is significantly improved compared to FedX Cooking at the moment ! Evaluation extended to federation engines (ANAPSID, HiBISCuS) Benchmarking for query federation over statistical data cubes SAFE extension for normal RDF data
- 18. http://linked2safety.hcls.deri.org:8080/SAFE-Demo/ Enabling networked knowledge 19 SAFE - Team • Yasar Khan • Muhammad Saleem • Aftab Iqbal • Muntazir Mehdi • Aidan Hogan • Panagiotis Hasapis • Axel-Cyrille Ngonga Ngomo • Stefan Decker • Ratnesh Sahay Thank You