Oppgaven er ikke lenger tilgjengelig

Efficiently import and index RDF data in fully indexed stores

RDF is nice in that you can index everything with just 6 indices, but that's still many indices to update, possibly causing low write performance. Can you find ways to fix this?

In a fully indexed triple store, you have 6 indices for triples or 24 for quad stores. When adding data, there are many indices to update, potentially causing low write performance. To understand this problem better, consider that you first write a lot of data to the datastore, and immediately thereafter, you query for parts of that data, and expect to see them, but since they may not yet be written to the database, much less indexed, they don't show up for various reasons. Just imagine how upset you would be if your post to Facebook didn't show up immediately! What can you do to balance the time it takes to write a lot of data to the store and the time it takes to query it? You may investigate ideas like using a sparse index that is gradually populated as computing resources become available, or other interesting directions.

Emneord: Semantic Web, Databases
Publisert 10. sep. 2013 23:04 - Sist endret 30. jan. 2014 13:43


Omfang (studiepoeng)