Almost three decades past since first neurocognitive models for language processing. None of them have been applied to practical large-scale computational problems. A neurocognitively inspired method is proposed to improve text categorization techniques. Unified Medical Language System metathesaurus is used to create a semantic memory and a neuronal back propagation algorithm is used to obtain weights for the semantic memory so it improves the text categorization. While there is little information in neurocognitive literature about role of inhibition in language processing, findings suggest that inhibitory connection between semantic units are crucial for the categorization task.

Registration required:

No

Free:

Yes