Evaluating the value of Named Entity Recognition for non-structured metadata with the help of OpenRefine
Seth van Hooland, Ruben Verborgh and Simon Henghen
Certain metadata (notoriously description) contain unstructured text, yet they usually convey a high amount of interesting information. To capture this in a machine-processable format, Named Entity Recognition (NER) can be used. Thanks to the brand new OpenRefine NER extension the Free Your Metadata (freeyourmetadata.org) team developed, you can enrich your description fields right from your workspace. During the boot camp, participants will be asked to perform the NER upon a sample of their own metadata. During the second half of the boot camp, we will discuss the quality of the outcomes of the different services (DBPedia Spotlight, Zemanta and Alchemy) and how the obtained URIs can be used for disambiguation and vocabulary reconciliation.
Understanding the general context of Linked Data and awareness regarding metadata standards
Running installation of OpenRefine and the NER extension