readability-wt

This node script is designed to calculate the readability of a webpage (via the Flesch-Kinkaid metric). It attempts to intelligently extract body text from the page - mainly via the assumption that meaningful (non-navigational) text generally comes in blocks longer than a sentence or so. The idea is to run this on webtask.io, so we need to use only the modules they support.

Modules Used

• request
  • Does exactly what it says on the tin.
• node-html-to-text
  • Lazy/concise way of stringifying and sanitising an html document, supported by webtask.
  • There's probably a better way but I lost too much time browsing the list of supported modules and just wanted to get to the interesting part

Issues

• Currently gives results that are lower than other online Flesch-Kincaid calculators. Algorithms need fine-tuning.
• Still not eliminating all extraneous text. Need to look into how node-unfluff handles this.

What next?

• Other readability metrics.
• Maybe return results in a more useful way.