Welcome to the Q&A Forum

PA/DA are based only on the link to the page/domain and don't attempt to distinguish spam from non-spam (except through their dependence on mozTrust, which does do a decent job of identifying spam).  As you've noticed, they should be used in combination with some type of spam filter when evaluating sites.  There are practical reasons why we cannot or do not want to use PR.  One reason is the fact that PR is often out of date as Dr Pete mentioned.  Another is that we need to assign a PA/DA to every page and domain in the Mozscape index (>80 billion pages and >150 million domains) and it isn't possible to get PR at this scale.

We have spent some time investigating spam algorithms and I prototyped one last year that I presented at MozCon.  We do plan to revisit it in the future, but don't have any immediate plans to do so.

Hi Chris,

I don't know of any free tools that do this unless you want to write some code yourself.  If you go that route we have some open source libraries that you might find useful, especially qdr that implements the TF-IDF scoring and dragnet for parsing/cleaning the HTML.  Good luck in your search!

Hi Lesley,

I'm the head Data Scientist here at Moz and can answer some of your questions about the DA changes you've seen.  You are correct that tweeting a URL will eventually include it in our Mozscape index, nice detective work   The blog post announcing Just Discovered Links that Keri quoted (http://moz.com/blog/announcing-the-just-discovered-links-report) is a little outdated and we are now including these URLs in Mozscape.  This is a recent change that overall increases the freshness of the Mozscape data.

You are also correct that including more links to a site in the Mozscape index will generally improve DA and these links were missing before you tweeted them.  We know that the Mozscape index does miss some links, and are working hard to improve the coverage of all sites in our index.  However, we still believe that PA/DA are valuable as overall summaries of the importance for each page or domain.  For example, PA/DA are well correlated with Google search position (see:  http://moz.com/search-ranking-factors).