Tutorial: [only for the desktop version] follow the magnifying lens (on the left besides to this text in a maximized browser.)
Discussion: Discovering Entities #3, video: Entity Extraction of free (English) text.
This demonstrator analyses the text in the text area, linking nouns with a logically linked data space and tagging linked nouns with respect to the discovered entities, showing for each tagged noun the corresponding entity and its RDF records (triples) found in DBPedia. This is the first and most sensible step inside an ETL process /etl/. The demonstrator discovers semantic correlations between concept pairs on the basis of a SKOS taxonomy taken from DBPedia. This demonstrator was realized using the Linked Data source DBPedia hosted in one of the current best RDF Triplestores – Blazegraph™.
Let yourself go (notes)
– Each HTML5 supporting device will be able to run this demonstration (best if Chrome, Safari on Desktop with at least 8GB main memory – for graphics).
– The data for this demonstration are taken from the DBPedia 2014 dataset – Please note that the use of this dataset is for pure demo convenience only, since it is a quite complete dataset. Of course we would be happy to mount „your“ data set as a basis for this technology.
– The provided demonstrator runs on a dedicated server without a NIS registration. In some cases when company policies are restrictive the url of this dedicated server will be blocked. Please access this page then from another place without these restrictions.
– The provided service is connected via IP to the demonstration page. In case of connection problems we will not be able to provide this service. We ask for your understanding.
– Entities are linked as they are found in the triplestore. Some results could be funny because they were found for the given text token.
– We thank DBPedia for providing these datasets but do not feel responsible for any data corruption or peculiarity found in them. We ask for your understanding.
