Keyword Categorisation Tool

The best way to visualise this potential project is through a demo and this is neatly already demonstrated by someone who offers a very similar functionality. ([login to view URL]) In essence I'm looking for the ability to upload a list of up to 10,000 keywords would could then automatically categorised into topical groups. This example above sources its data by crawling a website but I want to be able to upload a CSV with the keywords prelisted. (no page crawl required) So for instance, a list with a word like dog would be allocated to a category called 'pet' and also 'animal'. If the word cat was also in the list then this would be allocated into the same two categories and we could assume the 'topics' featured on the document (word list) were pets and animals. The categories would have 'weights' i.e there were more 5 times more words associated with 'comedy' than pets then this would be identifiable through some kind of index or relative weight scale. Ideally i would not have to pre-define the categories (there may be 1000- these would be done by you or by copying preexisting category taxonomy - I believe the above site used / has copied wiki-data categorisations which would suffice. This is very top-line brief obviously and full scope would need to be mapped. In my view there are two ways to achieve this - an unsupervised ML appraoch which could feature constantly semantic look-up tables. Id be interested to hear any questions or proposal as to how this might be achieved. Kind regards Mike