We're doing something similar, but I am afraid I can't release the code quite yet. Great to have a free example out there though. One problem I found with using n-grams and almost any database, neo4j included, is that speed is very important if you want to use this in auto-complete. Therefore, I would highly recommend doing one or more of the following.
1. Cache the entire dataset if possible 2. If relying solely on neo4j, as a corollary to #1, write a warm-up script 3. Use with a very fast caching layer such as memcached or redis in addition to or instead of neo4j. You can always have a script that loads redis with neo4j's data for instance as an external index. The gist is that if your auto-complete cannot do lookups in fractions of a millisecond, it will just "feel" wrong even if it is below one second. For that reason, we are going with #3 for web form auto-complete. For anything like a spell-checker where speed is important, but not the only thing that matters, a pure neo4j solution gives more sophisticated levels of checking and algorithms to leverage. -- View this message in context: http://neo4j-community-discussions.438527.n3.nabble.com/Neo4j-google-n-grams-data-set-in-neo4j-tp3540107p3543961.html Sent from the Neo4j Community Discussions mailing list archive at Nabble.com. _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
