Input Nodes |
---|
The networks will have one input node for each word in the vocabulary. The vocabulary will consist of ten nouns, twelve verb, three adjectives, one article, three prepositions, three pronouns, and one adverb (35 words in total). A complete list of these words is included in Appendix A.
The input to the network will also include nodes to represent the semantics of nouns and verbs. Semantic characteristics for nouns will be: - Human, Animal, or Mechanical (each represented by one of three mutually exclusive nodes). - Animate or inanimate (represented by one node, active if the noun is animate). - Singular or plural (represented by one node, active if singular). - Pronoun (active if true, inactive otherwise). - Proper (active if true, inactive otherwise). - One Noun ID nodes, used to distinguish nouns that might have identical semantical representation otherwise. For example, John and Mary would both be Proper names, singular, active, animate, human. In this case, each noun in the group will be assigned a distinct number. Semantic characteristics for verbs will be: - Tense, represented by activating one of the following mutually exclusive nodes: present, past. - Auxiliary verb. - Movement verb. - Sound produce verb. - Sound receive verb. - visual receive verb. - Two Verb ID nodes, used to distinguish verbs that might have identical semantical representation otherwise. For example, walk and swim would both be movements verbs. In this case, each verb in the group will be assigned a distinct number. There will be 60input nodes. All input nodes will use the identity transfer function.
|