Neural network for javascript text classification?

Question

JavaScript

There is a task of referring a data set to a classifier. The problem is that some of the fields in this dataset have "arbitrary" text. It seems that Google gives out all sorts of things about "convolutional networks" and so on, but there are no more detailed algorithms on how to configure the network to determine, for example, with which checkboxes the text was generated on https://yandex.ru/referats/ . Does anyone have examples or any information how to do this with these libraries?
https://github.com/karpathy/convnetjs
https://github.com/cazala/synaptic
https://github.com/stevenmiller888/mind
https://github.com/harthur-org/brain.js
or with using any other libraries?