Two people confronted with the same document will typically apply different tags, even if they are trained editors using a controlled set of terms. That's a clear sign that it can't be simple to automate tagging. 

Dozens of companies  claim to have solved the problem with revolutionary algorithmic breakthroughs. Using their products can be a bit of a letdown, as you rapidly see mistakes and oddities that don't quite fit the miraculous storyline spelled out in the marketing literature. That's why very few companies make it easy to try their software on your own content. 

 No single algorithm will ever solve the problem of tagging. There are many types of tags, and a nearly infinite variety of content. Matching them up requires high-quality execution on hundreds, if not thousands, of critically important details.

At TopicTagger, we are pushing the algorithmic frontiers, as is the case at many of our competitors. But we also recognize that some of the best ideas in search and tagging have been proposed as long ago as the 1970s.  Getting good results means being open to the best means available, and orchestrating a vast array of tools.

At TopicTagger, we respect the difficulty of the problem. Our software will make mistakes.  But we'll make fewer misakes, because we're up-front about the difficulty of the problem, and the inherent imperfections of any solution.