DefiningImageAccess/Tool/AllegroGraph
From ImageWeb
| DefiningImageAccess/Tool/AllegroGraph | |
|---|---|
| Link:=http://www.franz.com/products/allegrograph/}} | |
| Status:=Active}} | |
| JISCTool:=False}} | |
| Focus:=DataEngine}} | |
| [[Publishes::{{{Publishes}}}]]}} | |
| [[Uses::{{{Uses}}}]]}} | |
| [[RelatedTo::{{{RelatedTo}}}]]}} | |
| [[Partner::{{{Partner}}}]]}} | |
| [[Contact::{{{Contact}}}]]}} | |
| Defining Image Access | DefiningImageAccess/RelatedWork |
AllegroGraph
A commercial high-capacity triple store for RDF data.
The web page really says most that is to be said. Their claims include:
- can handle datasets consisting of billions of RDF triples
- supports SPARQL and other query mechanisms
- supports a variety of inference techniques (prolog-based and RDFS reasoning). Integrated with (commercial?) version of Racer.
So if very large triple stores are what you need, this is one to consider ... at a price.
Observations (based on the Free Java Edition for Windows system)
Outline
- loading: supports both NTriple and RDF/XML, and programmatically loading
- some optimisations, such as dictionary, indices
- how it works best
- load a large bulk of triples
- index them (indexAll())
- adding new triples by loading new files or programmatically
- indexTriples(), building new indices in addition to the old indices
- indexAll() will rebuild all the indexes and merge them into a new one. This operation should be avoided for a large triple database.
Quotes from AllegroGraph's Java Edition documentation (http://www.franz.com/products/allegrograph/doc/java/agraph-introduction.html)
- The standard query language in AllegroGraph is SPARQL
- AllegroGraph supports accessing a triple-store through JAVA, HTTP protocol, Lisp, and a visual inspector -- TopBraid Composer
- It provides customisable indexing strategy. The most comprehensive choice provides 6 different types of indices. This could greatly improve the query performance, although the indexing brings storage overhead too.
- It uses Lucene engine for string queries.
- It has built-in RDFS++ reasoner, which supports all the RDFS predicates and some of OWL's:
It supports the following predicates:
|
