The (hopefully near) future of human language technologies
235
The (hopefully near) future of human language technologies
Today’s language technology applications usually rely on either humandesigned rules (used sequentially by computers) or large amount of sequential data, that is, spoken or textual corpora. Today, computer modeling of human language abilities does not use parallel methods. In current natural language processing paradigms the notion of parallelism is almost totally missing. Multi-core processors are nowadays available even in commercial computers. On the other hand, results of brain research are quite far from existing language technology applications. Applying parallelism would lead us to a more realistic architecture for language understanding, with an increased processing speed.
More