0
$\begingroup$

I need some tool to classify articles based on short category text which consists of two or three words separated by '-'. The RSS/XML tag content is for example:

Foreign - News

Football - Foreign

I created my own categories and now I need to classify categories from parsed RSS of this news source, so it fits news categories defined by me.

I would for example need all articles containing category "football" to be identified as a category Sport but sometimes those categories XML tags contains exact match like Foreign - News should belong in the DB to category defined by me as Foreign.

I can of course also use longer description text if that would be needed but I think for this simple problem that would not be even necessary.

Since I used only trained decision trees frameworks so far for another project, I would like to hear advice about approach, AI technique or particular framework I can use to solve this problem. I don't want to get into a dead-end street by my own poor in this field not experienced decision.

Improve this question
$\endgroup$
1
  • $\begingroup$ Hi. Asking for tools (software) is off-topic here. So, if you don't want me to close this post, I would suggest that you edit this post and ask for an "approach" (using AI techniques) to solve your problem. $\endgroup$
    – nbro
    Feb 16, 2021 at 16:56

1 Answer 1

Reset to default
0
$\begingroup$

For this application, you can frame it as text classification. Look at SpaCy. You just need to create embeddings for your text and put a Softmax in the end. You can get those embeddings from BERT or anything else out there. You can in fact just use GLOVE vectors and others like it, concatenate them and then train a classifier.

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .