discover-rewards-notifier icon indicating copy to clipboard operation
discover-rewards-notifier copied to clipboard

Better tab title matching

Open nareddyt opened this issue 7 years ago • 3 comments

In #66, we introduced a new way to determine which offers should appear on a webpage. Specifically, we now fallback to checking the title of the tab and trying to match that to a retailer name in the data files.

For example, when I google for j. crew, the following appears because the tab name has j. crew in it:

image

However, this doesn't work if I google for j.crew or j crew because our matching algorithm just does a ignorecase string match.

We should add support for the following string match algorithms:

  • Ignore whitespace
  • Ignore punctuation
  • Ignore whole worlds. For example, I need to search for American Eagle Outfitters instead of just American Eagle. This is a harder problem, so this doesn't need to be solved yet.
  • Ignore special characters, like the trademark symbol.

Also, we should also try to match the tab title against the hostname. That way, if I search for ae in google, it gives me a deal for American Eagle because the site url is www.ae.com.

nareddyt avatar May 04 '18 02:05 nareddyt

See #70 for another example of bad title matching. In #70's case, we were not strict enough and were matching anything, even if it wasn't a full word.

nareddyt avatar May 05 '18 00:05 nareddyt

Hey @nareddyt , would love to work on this

itsthecheat avatar Oct 09 '18 21:10 itsthecheat

Go ahead! Let me know if you have any questions

nareddyt avatar Oct 10 '18 18:10 nareddyt