将聊天记录存储在关系数据库中

Question

I am writing a chat bot that uses past conversations to generate its responses. Currently I use text files to store all the data but I want to use a database instead so that multiple instances of the bot can use it at the same time.

How should I structure this database?

My first idea was to keep a main table like create table Sessions (startTime INT,ip INT, botVersion REAL, length INT, tableName TEXT) . Then for each conversation I create table <generated name>(timestamp INT, message TEXT) with all the messages that were sent or received during that conversation. When the conversation is over, I insert the name of the new table into Sessions(tableName) . Is it ok to programmatically create tables in this manner? I am asking because most SQL tutorials seem to suggest that tables are created when the program is initialized.

Another way to do this is to have a huge create table Messages(id INT, message TEXT) table that stores every message that was sent or received. When a conversation is over, I can add a new entry to Sessions that includes the id used during that conversation so that I can look up all the messages sent during a certain conversation. I guess one advantage of this is that I don't need to have hundreds or thousands of tables.

I am planning on using SQLite despite its low concurrency since each instance of the bot may make thousands of reads before generating a response (which will result in one write). Still, if another relational database is better suited for this task, please comment.

Note: There are other questions on SO about storing chat logs in databases but I am specifically looking for how it should be structured and feedback on the above ideas.

Answer 1

Don't use a different table for each conversation. Instead add a "conversation" column to your single table.