0
$\begingroup$

Inputs:

  • Time series of spectra representing human speech
  • Semantic network (as a directed graph) associations

Outputs:

  • Modified version of the semantic network input

Edge types in the graph (semantic, not neural, network connections):

  • Adjacency
  • Analogy of
  • Generalization
  • Example of
  • Composed of

Ideally, each edge would have the following two numeric attributes:

  • Median recency
  • Probability or strength

Characteristics of the permissible network topology:

  • Multiple edges types connecting the same two vertices
  • Cycles

This goes beyond parsing to the modification of semantic models in ways that simulate listening to language.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.