Using Textual and Visual Processing in Scalable Concept Image Annotation Challenge
Alexandru Calfa, Dragoş Silion, Andreea Cristina Bursuc, Cornel Paul Acatrinei, Razvan Iulian Lupu, Alexandru Eduard Cozma, Cristian Padurariu, Adrian Iftene · CLEF (Working Notes) · 2015
This paper describes UAIC 1 's system built for participating in the Scalable Concept Image Annotation challenge 2015. We submitted runs both for Subtask 1 (Image Concept detection and localisation) and for Subtask 2 (Generation of Textual Descriptions of Images). For the first subtask we created an ontology with relations between concepts and their synonyms, hyponyms and hypernyms and also with relations between concepts and related words. For the second subtask, we created a resource that contains triplets ( concept 1, verb, concept 2), where concepts are from the list of concepts provided by the organizers and verb is a relation between concepts. With this resource we build sentences in which concept 1 is subject, verb is predicate and concept 2 is complement.