wiki:DBTropes

Version 6 (modified by kiesel, 8 years ago) (diff)

added subclassOf statements

DBTropes logo
...it's like TV Tropes, but LINKED DATA!

About

TVTropes.org is a wiki-based catalog of the tricks of the trade for writing fiction. It contains pages for numerous movies, books, and other items, and associates these with tropes (writing devices and conventions).

DBTropes.org is a Linked Data wrapper for TVTropes.org. It parses TVTropes pages and outputs machine-interpretable RDF. For formalizing items and tropes, the Skipinions ontology created in Skipforward is used. Links to DBpedia can be generated, too.

The original idea of DBTropes was to be able to import TV Tropes data into Skipforward, then being able to use Skipforward's recommendation and annotation features, and also being able to use the TV Tropes feature hierarchy for Skipforward items. However, DBTropes data can also be used outside of the scope of Skipforward.

Example Pages

Documentation

Since DBTropes is an online wrapper, fixes in the way DBTropes parses TV Tropes pages are effective immediately. The following settings influence page parsing.

Page Type Settings

Each page can be interpreted as...

  • Item: A movie, game, etc. - links to other pages in itemizations are parsed if pointing to Feature pages only and output as hasFeature statements.
  • Feature: A trope. Links to other pages in itemizations are parsed if pointing to Item pages only and output as hasFeature statements.
  • Feature Cat: A trope category. Links to other pages in itemizations are parsed if pointing to Feature pages only and output as subclassOf statements.
  • Ignore: Page will be ignored - no type will be set.

After clicking on (Re)link with DBpedia (available for Items only), you will be presented a list of links found by Sindice when searching for the item label. "Set" the link that represents the current page in DBpedia. An owl:sameAs statement will be generated.

Hiding Statements

Using the buttons to the left of statements, individual statements can be selected as incorrect and hidden from the RDF output (currently not implemented).

Processing Comments

Within the RDF output typically there are some processing comments created by the wrapper. Most notable are the processingUnknown triples indicating referenced pages whose type is not known yet. If you want to help, visit these pages and set the proper page type.

Frequently Asked Questions

Is there a SPARQL endpoint?

No, not currently. Just download the dump (see below) and use a triple store of your choice to do SPARQL queries.

Shouldn't DBTropes HTML pages look better?

The only purpose of the HTML pages is to provide a simple frontend for fixing the generated RDF. Proper visualization should be done in other applications that use the RDF data (e.g., Skipforward).

Status

  • Item hierarchy support very basic only
  • Hiding statements in RDF output not supported currently
  • Some of the used properties are not available as Linked Data yet
  • On page type changes, pages referring to the current page should be reparsed, too
    • The RDF snapshot download should be fairly consistent though.
  • Recent Changes (for page type changes, etc.) is missing
  • No 303 redirects for Linked Data yet

Download

A daily snapshot of all data can be found here (NTriples format).

Timeline

  • 2010-03-18: Added subclassOf information for feature types. Currently generated offline.
  • 2010-01-21: Better URI normalization and sameAs resolving. LD output now includes referencing statements.
  • 2009-12-09: Initial release

Contact

DBTropes has been created by Malte Kiesel, a researcher at DFKI Knowledge Management department. Thanks to Gunnar Grimnes for help with the design.

If you want to contribute, please contact us.

Attachments (1)

Download all attachments as: .zip