What to choose for storing chat dialogs: relational, object-oriented or document-oriented databases?

Question

Database

I want to make a centralized storage of my im correspondence.
I have a lot of social media accounts, and looking for this or that dialogue is long and difficult, so I decided to write my bike.
The question arose what architecture to use, the main task is the search.

Answer 1

It is better to choose a document-oriented database, such as RethinkDB or MongoDB. I like the first one better, if only because their development is much faster, even if marketing has not worked out so far.
Some of the most important features in RethinkDB:
- one-click replication
- the ability to observe changes in tables ( rethinkdb.com/api/ruby/#changes )
- lightweight API
- ease of administration
Relational for chat - well, we don't seem to live in the 90s. .. Object-oriented - what is it for? Document oriented databases were created primarily for storing logs. And the message in the chat is nothing but a log.

Answer 2

It is for searching that there is an index of Lucene and Solr, etc.
For good speed, it is better to use them in addition to the NoSQL database, because there is a lot of chat data, and there is little relational data inside the chat.