14
1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

Embed Size (px)

Citation preview

Page 1: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

1

SWAD Europe Storage and Retrieval Workshop

Dave Beckett

Page 2: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

2

Overview

• Goals• Discussion outcome• Key issues

Page 3: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

3

Goals

• Gather from implementors:– Techniques– Problems met

• Discuss– Storage models, schemas, experience– Aggregation, provenance– Test data for stores, queries– Implementing datatypes, RDFS, OWL

Page 4: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

4

Storage Issues

• RDBMs or not• When to optimise from relational?• Scalability - today ~40M triples• Planning for 10x increase• Inference and text-searching

Page 5: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

5

Retrieval Issues - APIs

• Language APIs– Some languages have many RDF APIs

(Good thing / bad thing)

• Cross-language APIs– like DOM, IDL for RDF– That might be a query language

• Networked APIs

Page 6: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

6

Retrieval Issues - Contexts

• Not all software has support• Two main uses of context seen:

– Source URI (where triples came from)– Sub-graph identifier (modelID)

• Others:– Statement identifier (not reification)– Internal identifier

• Little query language support

Page 7: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

7

Lessons from implementing OWL, RDFS• RDFS easy except for 2 items• User-driven datatype support• XSD via third-party libraries• Interest in understanding OWL

layering• “OWL Tiny” the most implemented

Page 8: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

8

OWL Tiny

“OWL Tiny is the subset of OWL Lite that contains property characteristics and identity relations from OWL Lite. It is the part of OWL that cannot produce conflicts.”Steve Harris,University of Southampton.

Page 9: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

9

Retrieval and Querying

• Retrieval by query• Several QLs in similar SQLish style• RDQL, RQL, SeRQL, more…• SQL familiarity is a good thing• Felt now is time to standardise

Page 10: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

10

Query language standardising

• Conjunctive query with constraints• Maybe: +contexts, optionals• Not now: updates• Multiple other QL requirements• RDF graphs in and out useful• Path based querying• Query across graphs

Page 11: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

11

Web APIs and Web Services

• Networked services• Several tools provide WS APIs• SOAP use• REST architectural style preferred• RDF Net API

Page 12: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

12

Software APIs

• FAQ #1:What is the best software?[in my favourite language(s)]

• Healthy competition in APIs

Page 13: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

13

Summary

• Stores exist in substantial sizes• Please standardise a QL soon• Contexts are needed but messy

Page 14: 1 SWAD Europe Storage and Retrieval Workshop Dave Beckett

14

Thanks

See: http://www.w3.org/2001/sw/Europe/ Questions?