Should I go for document-based or graph database? (MongoDB vs Neo4j) [closed]

Question

I am currently choosing a NOSQL database for the application which stores a lot of emails to given topics (=mailing lists). There is a huge number of emails for each of them. As an amateur in this area, I dont know if I should go for Document-based MongoDb or Graph oriented Neo4j. It is more like write once, read many type.

The facts for the Neo4j: 1. Emails are basically a graph. Each email is a node and one replies to the other (= an edge). 2. The graph can be really deep, one email can have many responds. 3. It may be needed to reorder some emails (to clean ordering if really needed) in graph.

The facts for the MongoDb: 1. Sometimes may user want to search for some parameter-based queries. For example all emails from the user X. 2. Emails can be quite big. 3. Documents looks much more simpler.

The fact I don't know where belong: Sometimes it is good to search for the parents and check, where did email copied the text from (the specific text which email is replying to).

Can someone please help me?

EDIT: Anyway, I am also open to other nosql databases.

Answer 1

Maybe this blog post will help steer your decision: http://soloso.blogspot.com/2011/07/getting-enron-mail-database-into.html

Take note of the very bottom addendum area -- where @rit expanded on the original code and hosts a dump of the corpus on S3.

That ought to give you quite a bit to experiment with in the MongoDB environment.

Bryan