Premium
Using Images As a Foundation for Natural Language Processing
Author(s) -
Kohanim Michel,
Johnson Julia
Publication year - 2000
Publication title -
computational intelligence
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.353
H-Index - 52
eISSN - 1467-8640
pISSN - 0824-7935
DOI - 10.1111/0824-7935.00130
Subject(s) - computer science , premise , natural (archaeology) , sight , artificial intelligence , natural language processing , natural language , consistency (knowledge bases) , representation (politics) , question answering , modality (human–computer interaction) , space (punctuation) , linguistics , philosophy , physics , archaeology , astronomy , politics , political science , law , history , operating system
A new framework for knowledge representation is proposed, based on modality‐dependent (sight, sound, touch) images, within which agents will be better able to communicate using natural languages. The premise is that natural language processing depends not only on the grammatical rules but also on entities (whatever the agents are communicating about), their attributes, and supported events (in a time/space continuum). We call the collection of an entity's associated attributes a state. The description of a concept is refined when new information about the concept is encountered through discovery of consistency between the attributes belonging to different image instances.