Hugo ter Doest
Hugo ter Doest
Please elaborate a bit more on what you want to achieve. tfidf is a weighting algorithm, it gives insight in how (un)important words are in documents. It is not a...
Imho that is not what tokenization is meant for. Tokenization splits a text into words (and punctuation, if necessary) and "take off" consists two words. Combining them into a phrasal...
It's not yet in natural, but I'm working on that to use it for named entity recognition. You can have a preview at a CYK and Earley parsers here in...
The number of layers is not correctly set after reading the Json file. It is always 3 and therefore returns the result of the third layer. Will correct this in...
Naar prio laag, versiebeheer op documenten wordt default.
I don't see a dependency here on webworker-threads. I think you mean the dependency in NaturalNode, here: https://github.com/NaturalNode/natural/blob/master/lib/natural/classifiers/classifier_train_parallel.js Is that what you mean? I can replace it with webworker.
I will ook into this.
I checked the source files that are mentioned in your stack trace. But I don't see a dynamic require of `util`. Any clue what is going on?
Oke, let's have a look. Webworker threads are not dynamically required, a try-catch construct is used: ``` try { Threads = require('webworker-threads') } catch (e) { // Since webworker-threads are...
@andynunes @justinfagnani I am in the process of adding TypeScript support. Would be great if could take a look at the first steps. I added a `index.d.ts` to the tokenizers...