6

I have created a nice LaTeX resume.

However, having uploaded it to an online job search I am worried the automated extraction does not perform well.

Are there any specific packages or other tips to improve the machine-readability of a PDF file (particularly with a view to producing resumes)?

I am using multicols which messes up extraction. Disabling multicols is easy but then I cannot get the content onto 2 sides. Is there anyway to make multicols output easy to extract for an automated system?

Does anyone have any strategies for maintaining two versions of the resume, one machine readable and one typeset for printing? Is there anyway to have some kind of 'alt-text' attached to the PDF that these systems would pick up? The problem is when submitting an application there is no way to know if they will use an automated feature extraction.

The biggest problem is that the extracted text seemed to have no spaces (allthewordscomeouttogether). Is there any way to force LaTeX to include a space character in the text?

  • Previous questions of this form have been closed as off-topic as the machine-readablity of a file is dependent on the reading software, not the producer. – Joseph Wright Aug 10 '13 at 6:30
  • As for different versions, check out the (I think) CurVe document class. You can have different 'flavors'. If you don't want to use that, keep in mind that LaTeX's great strength is content-format separation. – Sean Allred Aug 10 '13 at 12:01
  • @JosephWright: I understand that, but there is a lot of discussion of LaTeX for resumes and this is a really important consideration for the current job market which I have not seen discussed elsewhere. – robince Aug 10 '13 at 13:07
  • @thrope Quite possibly, but that per se doesn't mean TeX expertise will help answer. – Joseph Wright Aug 10 '13 at 13:18
  • @JosephWright the justification for claiming that readability depends on the reader is spurious. the reader stands no chance with general tex output, since tex font encodings aren't standard. the cmap package (and related ones) provide translation tables from the tex glyph codes to standard ones; unfortunately, i'm not expert in use of the package. (i assume matters are better with unicode-coded pdf, as from xetex or luatex ... i have no evidence, though). – wasteofspace Aug 21 '13 at 10:27
2

As you mentioned your isues relate to multicol (but without giving an ME) I did a quick test:

\documentclass{article}

\usepackage{multicol,lipsum}

\begin{document}
\begin{multicols}{2}
\lipsum[2]
\end{multicols}
\end{document}

If you run this through pdftex you get the following pdf and as you can see there is no issue whatsoever to select the left column material, or parts of it as I did (at least in Acrobat):

enter image description here

And when I paste the selected text back into some editor I get:

Nam dui ligula, fringilla a, euismod
sodales, sollicitudin vel, wisi. Morbi
auctor lorem non justo. Nam lacus
libero, pretium at, lobortis vitae, ul-
tricies et, tellus. Donec aliquet, tortor
sed accumsan bibendum, erat ligula

i.e., all spaces are there (or rather the extractor correctly deduced them). So I guess you need to be more precise in what you are doing and what the results are for anybody to help you.

As to auto-extraction: I guess this really depends on how clever the software is doing it. As the above example shows the structure in the pdf is rich enough to be interpreted correctly, e.g., if you turn on reading mode in Acrobat then you will get the right words spoken in the right order and not in screen order.

Concerning the spaces: TeX doesn't output space characters but positioning commands that move to a new position at a word boundary (given that it does justification this isn't surprising) and there isn't really any way around this. But for exstraction software that should not be a problem (in theory).

  • Thanks. At least bright.com doesn't pick up the columns correctly - but even with columns disabled a lot of times the spaces are missing. I have no idea what software companies might use for automatic keyword screening. I guess it may be safer to stick to html if they accept that format (the output from pdf2htmlex looks really good). – robince Aug 10 '13 at 13:05
  • @thrope you may be right and you don't have control over this. But on the other hand, my guess would be that crappy extraction software would choke on any pdf not just those generated by TeX, in which case you should be fine. Try a pdf generated from Word (with justification) and look if it behaves better with your software. – Frank Mittelbach Aug 10 '13 at 13:57
0

I just found pdf2htmlEX which may solve some of my problems. Still interested to hear other possibilities though.

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.