Currently lists of ngrams and their history of changes are saved as files (NodeStory type), it is loaded in memory (RAM) and it should be saved in database to scale up its capacity. For now big lists and big corpora are rather slow and its behavior is quite erratic.

Roadmap to optimize it.

First adapt the schema of the database to welcome the ngrams list instead of saving it as file.

After this connection, we will work on the way to interface the in-memory data structure store (NodeStory) and progressively save it in database in a continous way (the same way Reddis is coupled with others databases; here we use the NodeList structure which is much more complex than a simple key value).