id author title date pages extension mime words sentences flesch summary cache txt work_wysobjngavdtdivi2qo2jkd6fi Peter Young From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions 2014 12 .pdf application/pdf 8217 754 67 From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions construct a denotation graph, i.e. a subsumption hierarchy over constituents and their denotations, based on a large corpus of 30K images and 150K descriptive captions. everyday activities (each paired with multiple captions; Section 3) to construct a large scale visual denotation graph which associates image descriptions that constructs the denotation graph uses purely syntactic and lexical rules to produce simpler captions variety of descriptions associated with the same image is what allows us to induce denotational similarities between expressions that are not trivially related similarities (cos, Lin, Bal, Clk, Σ, Π) on our image captions ("cap"), the BNC and Gigaword. To allow a direct comparison between distributional and denotational similarities, we first define P(w) (and P(w,w′)) over individual captions the vectors obtained from the image-caption training data) and denotational similarity features to this ./cache/work_wysobjngavdtdivi2qo2jkd6fi.pdf ./txt/work_wysobjngavdtdivi2qo2jkd6fi.txt