Acropolis
Acropolis is a toolkit for building and maintaining knowledge graphs. The Acropolis stack powers the BBC's Research & Education Space and Datalab projects.
RES's purpose is to construct a knowledgebase of Linked Open Data, a reasonable amount of which (thanks in part to our intervention) ought to concern or relate to creative works with associated accessible (subject to licensing/access controls) media assets which may be presented in a variety of ways, and in doing so always represent the origins of the data and ensure where possible that applications can retrieve it, both because RES won't always retain the finer (domain-specific) detail, and because RES doesn't claim to be authoritative.
Build system support macros
URI parsing and manipulation library
Client access library for services speaking AWS protocols
Client access library for SPARQL-savvy graph stores
Client access library for SQL databases
Self-balancing cluster library
- Add support for https://github.com/i-scream/libstatgrab?
Should be renamed to libappconfig
and have logging elements removed (replaced with liblogging
)
Web crawler
RDF-oriented workflow processing server
Linked Data server
- Support
GET
requests including a structured request payload? (JSON?)- Can this be implemented without breaking caching? Is there a
Vary: body
? - More trouble than it's worth?
- Should be a
POST
? But then breaks semantics?
- Can this be implemented without breaking caching? Is there a
- Other HTTP methods should be at least handled; it should then be up to the engine to decide what to do.
- e.g., Patchwork could accept S/MIME-signed
PUT
requests to enqueue graphs for processing?
- e.g., Patchwork could accept S/MIME-signed
- Configurable routing (e.g., send different URI patterns to different engines)
- Incoming URI and query-parameter transformation
- It should be straightforward to implement this with neglible loss of efficiency
Knowledge-base processing engine for Twine
Knowledge-base server engine for Quilt
- Add support for multiple
?uri=URI
query-parameters, return a single RDF doc containing all of the requested entities.- Add a configurable soft-limit for the number of URIs which will be processed in a single query
These wiki pages are the source for my notebook; I use Travis and some vaguely interesting build scripts to publish their contents as part of my website on a regular basis.
In other words, I use Github Wiki as a content management system for parts of my website. It works quite well.