Full text indexing with MongoDB
We’ve seen this done before with CouchDB, but I still think the full text indexing should be delegated to specialized solutions like Lucene and Solr.
Secondly, if I got it right, the proposed solution is re-inverting the index and store it based on the indexed documents which I don’t think is a good idea either. It will lead to basically doubling the number of documents stored, plus searching will depend linearly on the number of indexed documents instead of depending on the number of terms which normally would very less over time (nb this being the whole idea behind inverted indexes). Unfortunately even if just a proof of concept, this approach shows one of the pitfalls of data modeling with NoSQL systems.
via: http://hmarr.com/2010/mar/18/full-text-search-with-mongodb/
