Why does this release of Similar Posts, Recent Posts, Random Posts and Recent Comments get a clean new number instead of being called a beta? I guess it’s arbitrary! I can’t stop tinkering so in a sense the plugins are perpetually in beta but on the other hand bug reports are drying up (apart from the ones I keep introducing!).
I want to concentrate some more on Similar Posts for a while and try out some ideas that have emerged from reading on the subject. It is a challenge to find good matching algorithms that can be made speedy enough in PHP and not too demanding in memory usage. A few months ago I had a great system running but had to drop it when I moved from my local development server to my live hosted server which immediately froze from lack of memory.
The current release improves the Chinese/Japanese/Korean language matching by searching on bigrams instead of single characters. There are some much better algorithms in the literature but they cost too much in time and memory to be useful here. I hope the bigram scheme is good enough to be useful.