Grammatical word-disambiguation again

The challenge is especially that of generalizing the grammatical word-disambiguation to several languages. Creating a module of grammatical word-disambiguation for each language appears to be a long and arduous task. This seems to be the main difficulty. But if a module specific to a given language can be generalized to several other languages, this could be an important advance in the field of rule-based machine translation (which simulates human reasoning seems to me a more appropriate term).

We can describe the problem more precisely. We have about 100 grammatical categories for a given language. We also have about 300 ambiguous grammatical types – to fix ideas – which are: e.g., adverb or preposition, singular masculine noun or singular masculine adjective, etc. The problem is to describe an algorithm to remove the ambiguity and determine the corresponding grammatical type according to the context.

Now rewriting the complete module of disambiguation by grammatical type, so that it can be used and adapted to other languages (Italian in the first place). It remains to be seen if this can be done.

This entry was posted in blog and tagged , , , , . Bookmark the permalink.