Can't feedback learning

[SpicyPen1992]

Dear gunthercox: My English is not very good. If the statement is not smooth, please forgive me.

After updating the new version 1.0.0, I found that ChatterBot could not complete the feedback learning. By looking https://github.com/gunthercox/ChatterBot/issues/1553, I also found that this issue may not only in the new version. After chatbot train first, i run my code:

Begin feedback learning: User: What is the game of Saint Setya? Bot: Is "What is the game of Saint Setya?" "What is Saint Setya?"'s real response? User:n Please input real response: User(real):The Saint Seiya is a game released by Leju Company. Bot: study+1 User:What is the game of Saint Setya? Bot: Is "What is the game of Saint Setya?" "What is Saint Setya?"'s real response?

---

As a newbie, I hope the author can improve the ChatterBot or give me some guidance. Thank you very much.

---

My code as follow: (PS:saintseiya.yml likes ai.yml)

from chatterbot import ChatBot
from chatterbot.conversation import Statement
from chatterbot.trainers import ChatterBotCorpusTrainer

chatbot = ChatBot(
    'Thomas',
    storage_adapter='chatterbot.storage.SQLStorageAdapter',
    database="botData.sqlite3"
)

trainer = ChatterBotCorpusTrainer(chatbot)
trainer.train(
    #'chatterbot.corpus.chinese',
    'data/saintseiya.yml'
)

def get_feedback():
    text = input("User:")
    if 'y' in text.lower():
        return False
    return True

print('Begin feedback learning：')

while True:
    try:
        text=input('User:')
        if text=='stop':
            break
        input_statement = Statement(text)
        response = chatbot.generate_response(input_statement)

        print('Bot：Is "{}" "{}"\'s real response? \n'.format(
            response.text,
            input_statement.text
        ))
        print('Please input y/Yes or other/No: ')

        if get_feedback():
            print('Please input real response：')
            correct_response = Statement(text=input("User(real)："))
            chatbot.learn_response(correct_response, input_statement)
            print('Bot：study+1！')
        else:
            print('Bot:'+response.text)
    except (KeyboardInterrupt, EOFError, SystemExit):
        break

[jkjunior]

@gunthercox I have the same issue. Did you check it already?

[gunthercox]

Sorry, I haven't had a chance to look into this yet. It's amazing how fast time goes by!

[maxlxl]

I have the same Issue here aswell. I had a look into the database the bot uses and it adds the information of the learned statement correctly. But somehow the search doesn't work correctly. Additionally I'm always getting a confidence value of 0. If you need further details, I'll would like provide them to help to fix the bug. I've already checked the source code but except for #1589 and #1608 I couldn't find anything propably related to this problem.

[Wumba]

Hello Everybody, I have a simple workaround how somebody still can train the bot on demand:

trainer= ListTrainer(bot)

def get_feedback():
    if 'stop' == str(input_statement):
        return False

last_input_statement=[]

 while True:
    try:
        input_statement = (input(bcolors.bold + "You: " + bcolors.endc))
        response = bot.get_response(input_statement)

        if get_feedback() is False:
            print("Please enter something more funny")
            correct_response = input()
            last_input_statement.append(correct_response)
            trainer.train(last_input_statement)
            print("Response added to Chatbot")
            last_input_statement.clear()

        else:
            last_input_statement.clear()
            last_input_statement.append(str(input_statement))
            print(bcolors.bold + "Chatty: "+ bcolors.endc + str(response))

    # Press ctrl-c or ctrl-d on the keyboard to exit
    except (KeyboardInterrupt, EOFError, SystemExit):
        print(bcolors.fail + " Bye!" + bcolors.endc)
        break