Skip to main content

Spam Classification And Review Accuracy Improves

At any given time, we can see a small sample of the Blogger blog universe, as reported in Blogger Help Forum: Get Help with an Issue.

One sample, that we may see, is composed of the blogs which have been deleted / locked, by the Blogger spam classifier - which the owners want restored.

If properly requested by a former owner, we may request review of a blog, that appears to be improperly classified.

We sample the Blogger spam population, using forum spam reviews.

To request review, we submit a blog in a database. The database is read by the Google staff, which hand review blogs classified by the automated processes.

Having submitted a handful of review requests, we wait for the review results. The results of the reviews provide a sample, of blogs being classified, and reviewed.

Seeing a trend of spam review results, we observe what is being classified.

The general trend would be between 33% and 66% of righteous / spurious spam classification ratio (in other words, varying between a 1/2 to a 2/1 ratio). Instinctively, that should be normal - since Blogger tries to get as many spammers out of business - but without disturbing too many legitimate blog owners.

Occasionally, we see the ratio more like 1/9 - or 9/1. Then, we see a predominance of one or two classes of blogs, as reviewed.

  1. Blogs not spam.
  2. Blogs marginally spammy.
  3. Blogs blatantly spammy.

Currently, we are seeing more legitimate blogs, being spuriously classified.

Most recently, we saw a large population of Groups #1 and #2. When review was requested, 95% of those submitted were restored.

There will always be some spam blogs, not classified - that should be. And there will always be some blogs spuriously classified - that should not be.

But when the majority of the blogs for which review is requested, are subsequently restored, that tells us that the Blogger spam classifiers are having to reach deeper into Groups #1 and #2, above. And that implies that Group #3 is becoming smaller. And that Group #3 includes less blogs which blatantly imitate Group #1.

There will always be spammers, trying to discourage spam reviews.

In spite of the devious maligning of the Blogger spam mitigation policies

The Blogger system of preventing spam is full of failures - and the support team don't remove blogs with spam/malware/nudity and other offenses.

We can tell, from the samples, that the system is working. And that of the people who suggest the negatives

The Blogger system of preventing spam is full of failures - and the support team don't remove blogs with spam/malware/nudity and other offenses.

many of them are non self aware spammers, who are lamenting loss of their blogs.

People who want spam classification improved have to request review.

If spam filter tuning is to continue successfully, everybody who is not a spammer, but who is treated as if they are, must request review of their blogs. And the majority of the review requests must produce blogs restored - which gives Blogger details to tighten the filters, and classify less blogs that are legitimate, during the next classification cycle.

Blogger can't tune their filters based upon non responding legitimate blog owners. People who post
My blogs were deleted - but I'm not providing the URLs, because the Blogger anti-spam policies don't work!
Either

  • Are spammers, trying to discourage the spam classification and review process.
  • Are non spammers who will, unfortunately, never see their blogs again.

Which group each blog owner falls into, remains to be seen. If your blog was recently classified, and you believe classification was unfair, then you have to submit your blog for review.

Comments

Popular posts from this blog

Custom Domain Migration - Managing The Traffic

Your blog depends upon traffic for its success.

Anything that affects the traffic to your blog, such as any change in the URL, affects the success of your blog. Publishing the blog to a custom domain, like renaming the blog, will affect traffic to your blog. The effects of the change will vary from blog to blog, because of the different traffic to every different blog.Followers. People who find your blog because of recommendations by other people.Search engines. Robotic processes which methodically surf your blog, and provide dynamic indexing to people who search for information.Subscribers. People who read your content from their newsfeed reader, such as the dashboard Reading List.Viewers. People who read your content from their browser.No two blogs are the same - and no two blogs will have the same combinations of traffic sources.

Stats Components Are Significant, In Their Own Context

One popular Stats related accessory, which displays pageview information to the public, is the "Popular Posts" gadget.

Popular Posts identifies from 1 to 10 of the most popular posts in the blog, by comparing Stats pageview counts. Optional parts of the display of each post are a snippet of text, and an ever popular thumbnail photo.

Like many Stats features, blog owners have found imaginative uses for "Popular Posts" - and overlook the limitations of the gadget. Both the dynamic nature of Stats, and the timing of the various pageview count recalculations, create confusion, when Popular Posts is examined.