On Building an Efficient, Indexed Search Engine With a Word Proximity Algorithm

If you really want to write your own search library, Manning and Raghavan's introduction to IR is highly recommended (and freely available)

http://www-csli.stanford.edu/~hinrich/information-retrieval-...

If you just want to use a search library that gives you word proximity (and more), it would be a lot less work to look at existing open source implementations. Xapian and Lucene are both quite useful and can be invoked from PHP, among other languages.

It's hard to say that this algorithm is very fast and uses very little memory unless you have actually deployed it on some sort of reasonable corpus.