Currently working on Resume Rarser tool using doc2vec.

The main assumption that I take when parsing resume is that each line of text (docx, pdf etc) contains information of one class.

Although there are some cases when one line of text could contain several classes. For example,

2016-2018 University of Glasgow, Nanotechnology

It contains education period, university and specialization.

To fix this, I am able to split the string into bi/trigrams and classify the string correctly.

If I classify the bigrams of normal strings though it gives much worse results.

The question that I have is How to detect this kind of multi-class strings?

  • $\begingroup$ Do the classes just have deterministic and clear category of data? (Ex. name, year, location, etc.) $\endgroup$ – Alireza Zolanvari Mar 14 at 10:57
  • $\begingroup$ @alirezazolanvari yes, actually text also gets clusterized into info, education, experience etc for better further classification of each cluster $\endgroup$ – Graygood Mar 14 at 12:17
  • $\begingroup$ What I also found, that numbers in such strings as sample take too much 'attention' for classifier. $\endgroup$ – Graygood Mar 14 at 12:18
  • $\begingroup$ According to what I know, all of these clusters have some specific (and not necessarily unique) keywords which can be used in this case (Ex. Education cluster keywords: university, school, etc.) $\endgroup$ – Alireza Zolanvari Mar 14 at 12:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.