Analysis of Image Captions in Scientific Publications

14:41 by Diana Unander

On 28 May from 13 till 14.30, in F322 (Växjö) we will have a talk by a visiting professor to the iInstitute, Christian Wartena from University of Applied Sciences and Arts Hanover.

In the project NOA of the Technischen Informationsbibliothek (TIB) and the University of Applied Sciences and Arts Hanover we collect images with a free license from open access publications in order to make them available via Wikimedia Commons and thus to increase the potentials to reuse scientific images. In order to annotate and describe the content of the images techniques from image recognition are not suited.  Instead we analyze the captions and the passages referring to the image. We will show what specific problems the extraction of keywords from captions provides and how we solved them. Especially we will show, what role the imagery and concreteness of words, two notions form psycholinguistics, can play and how we can determine these values for arbitrary words.

Welcome!

For more information contact: Koraljka Golub – koraljka.golub@lnu.se

Diana Unander

Comments are closed.