AutoBlogged can automatically identify tags from the original article based on frequency, importance, sentence structure, HTML formatting, and other factors. The purpose of this is to identify relevant and natural keywords and phrases that the author did not include as tags. This is important for search engine results and also to assist visitors in finding the content of your site.
The tagging engine will quickly fill your blog pages with relevant keywords that will help your site dominate the search engines for your topic.
Although the internal tagging engine is quite effective in most cases, it is still just a script and could never accomplish what a human could. Often you will encounter keyword phrases that are not relevant, consist of sentence framgents, or simply do not make sense.
Although you might want to manually delete tags that detract from your content, you should also consider that a search engine will see the tags differntly than a human would. Often those sentence fragments will help to make your content unique and will help to diversify your page content while still keeping it on topic. It is not uncommon to find that some tags you would normally delete turn out to be the ones that bring in the most search engine traffic.
For example, you might have a blog that covers topics related to Microsoft Windows. Invariable you will pick up tags related to Linux operating systems, or software that is only relevant because it runs on Windows. These tags might turn out to be extremely helpful in search engine positions due to the unique mix of words.