This paper examines the use of two kinds of context to improve the results of content-based music taggers. Context in this case means the relationships between tags and the relationships between the clips of songs that are tagged. We show that users agree more on the tags that should be applied to clips that are temporally "closer" to one another; that conditional restricted Boltzmann machine models of tags that take this context into account can more accurately predict related tags; and that when context is used to "smooth" training data, support vector machines can better rank these clips according to the original, unsmoothed tags.
Contextual Tag Inference
AIELLO, LUCA MARIA;SCHIFANELLA, ROSSANO;
2011-01-01
Abstract
This paper examines the use of two kinds of context to improve the results of content-based music taggers. Context in this case means the relationships between tags and the relationships between the clips of songs that are tagged. We show that users agree more on the tags that should be applied to clips that are temporally "closer" to one another; that conditional restricted Boltzmann machine models of tags that take this context into account can more accurately predict related tags; and that when context is used to "smooth" training data, support vector machines can better rank these clips according to the original, unsmoothed tags.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.