will pick a character and then select data that starts with that character. However, this means that the bot needs to learn sentences that start with any of the 27 possible tokens before it stops saying "There is not enough learned information yet.".
This can definitely be improved.
Obviously a select on all MarkovStart tables as opposed to just one will work, but I would love to keep the program somewhat efficient even if it has enormous amounts of information.
This can also be fixed by altering the structure of the database, which currently exists of very many tables to avoid the seemingly exponential increase in query time relative to the row count.
Perhaps switching to a different database structure will help too.
The lines https://github.com/CubieDev/TwitchMarkovChain/blob/c7b8639bf85ae129f4a1e3f0dd757ecdef75f098/Database.py#L226-L230
will pick a character and then select data that starts with that character. However, this means that the bot needs to learn sentences that start with any of the 27 possible tokens before it stops saying
"There is not enough learned information yet.". This can definitely be improved. Obviously a select on allMarkovStarttables as opposed to just one will work, but I would love to keep the program somewhat efficient even if it has enormous amounts of information.This can also be fixed by altering the structure of the database, which currently exists of very many tables to avoid the seemingly exponential increase in query time relative to the row count. Perhaps switching to a different database structure will help too.