The Chatbot class in OrionBot has a built-in training mechanism that allows you to add new training data to the chatbot and improve its accuracy over time. In this document, we’ll go over the various ways you can train your chatbot.
To add training data to the chatbot, you can use the train method. This method takes in two parameters: input_text and response_text. input_text is the text that the user inputs, and response_text is the response that the chatbot should give in return. Here’s an example:
bot = Chatbot("my_bot")
bot.train("Hello", "Hi there!")
This will add the training data “Hello” and “Hi there!” to the chatbot. The train method automatically preprocesses the text by tokenizing it, lemmatizing it, and removing stop words. The preprocessed text is then added to the chatbot’s training data.
If you have a large amount of training data that you’d like to add to the chatbot, you can use the train_from_file method. This method takes in a file path to a JSON file containing training data in the following format:
{
"input_text_1": ["response_text_1", "response_text_2", ...],
"input_text_2": ["response_text_1", "response_text_2", ...],
...
}
Here’s an example of how to use train_from_file:
bot = Chatbot("my_bot")
bot.train_from_file("training_data.json")
This will add all of the training data in the training_data.json file to the chatbot.
By default, the chatbot’s training mode is enabled, which means that if it doesn’t know how to respond to a user’s input, it will ask the user for a response and add that to its training data. However, if you’d like to disable training mode, you can use the disable_training method:
bot = Chatbot("my_bot")
bot.disable_training()
Now, if the chatbot doesn’t know how to respond to a user’s input, it will simply return a default “I’m sorry, I did not understand what you just said” message.
If you’ve disabled training mode and would like to re-enable it, you can use the enable_training method:
bot = Chatbot("my_bot")
bot.enable_training()
Now, if the chatbot doesn’t know how to respond to a user’s input, it will ask the user for a response and add that to its training data.
By default, if the chatbot doesn’t know how to respond to a user’s input and training mode is enabled, it will prompt the user for a response. If the user doesn’t provide a response, the chatbot will return a default “Okay, I’ll keep that in mind” message. However, you can customize this message by using the set_no_response_message method:
bot = Chatbot("my_bot")
bot.set_no_response_message("I'm sorry, I still don't understand. Could you please try rephrasing your question?")
Now, if the chatbot doesn’t know how to respond to a user’s input, it will send the message you specified.