How can I ensure my chatbot will give diverse answers to the same question. I am currently using the List Trainer on a yml file but it keeps giving the same response line by line.
I don't quite recall if it was embedded in a website or an installable program (I'm pretty sure it was installable). It was a project someone was working on with AI and chatterbots like A.L.I.C.E. or CleverBot. This one started with no dictionary associated, and would learn everything based on your communication with it, not distributed over everyone using it like some online ones do. Every time you would type something, it would add the words to it's dictionary and learn syntax based on your usage of syntax. Over time it would start to make complete sentences, and it was really impressive though it was kind of eerie or at the same time. I ended up deleting it, but I kind of want to check it out again to see if any progress was made.
I don't know if this was some unknown project or a currently maintained one. It was a few years ago, probably 5-6ish when I found it though I don't recall how old it was when I found it so that may not be helpful.
Any ideas?
Want a chatter bot for your phone? Get bored and want something to pass the time? Curious about artificial intelligence? Check out the Markov Chatbot Task I built! N.B. I am not the bot developer!
It will even learn from your speech!
We need Termux and the Tasker plugin for. We also need AutoVoice plugin to speak to it. We also need Markovbot
Inside Termux, we need to pkg install python
Inside Termux, download and extract the zip. I used ~/bot/markovbot-master
Tweak your python script to call from Tasker, this will allow arguments from Tasker to be passed.
import os
import time
import sys
from markovbot import MarkovBot
tweetbot = MarkovBot()
dirname = r'/sdcard/Markov'
book = os.path.join(dirname, u'brain.txt')
tweetbot.read(book)
my_first_text = tweetbot.generate_text(int(sys.argv[1]), seedword=[u'sys.argv[2]', u'sys.argv[3]'])
print(u'%s' % (my_first_text))
We can now execute the script, _python chat.py 10 word1 word2 where 10 is how many words to reply with and word1/2 seed words to illicit speech.
I created a small shellscript to be dumped into ~/.termux/tasker so we can execute this python script as well as mount Android storage..
#!/bin/sh
termux-setup-storage
python ~/bot/markovbot-master/test.py $1 $2 $3 > /sdcard/Markov/reply.txt
It uses /sdcard/Markov for the brain and replies. The file brain.txt can be almost any text file, but one that is sentence or paragraph separated works better. Some text files just don't work.
And finally, the Task..
Markov ChatBot (86)
A1: Variable Set [ Name:%spc To: Recurse Variables:Off Do Maths:Off Append:Off ]
A2: Variable Set [ Name:%rtn To:
Recurse Variables:Off Do Maths:Off Append:Off ]
A3: Array Set [ Variable Array:%no_replies Values:I have nothing,Not sure. Ask me again later,I don't have an answer for that!,Hmm. I don't know what to say! Splitter:, ]
A4: Array Set [ Variable Array:%voices Values:en-GB-Wavenet-A,en-GB-Wavenet-B,en-GB-Wavenet-C,en-GB-Wavenet-D,en-AU-Wavenet-A,en-AU-Wavenet-B,en-AU-Wavenet-C,en-AU-Wavenet-D,en-US-Wavenet-A,en-US-Wavenet-B,en-US-Wavenet-C,en-US-Wavenet-D,en-US-Wavenet-E,en-US-Wavenet-F Splitter:, ]
A5: Variable Set [ Name:%max_voices To:%voices(#) Recurse Variables:Off Do Maths:Off Append:Off ]
A6: Variable Randomize [ Name:%vrnd Min:1 Max:%max_voices ]
A7: Test File [ Type:Exists Data:MI am using ubuntu 18.x and my system has python Python 2.7.15rc1 and python Python 3.6.7.
_________________________________________________________
When I run
$python -c "import chatterbot_corpus"
$ echo $?
0 // I get this
_______________________________________________________________
But when I run
$ python3 -c "import chatterbot_corpus"
Traceback (most recent call last):
File "<string>", line 1, in <module>
ModuleNotFoundError: No module named 'chatterbot_corpus'
________________________________________________________________
I used this to install
pip install chatterbot-corpus
and got this
Collecting chatterbot-corpus
Collecting PyYAML<4.0,>=3.12 (from chatterbot-corpus)
Installing collected packages: PyYAML, chatterbot-corpus
Successfully installed PyYAML-5.1 chatterbot-corpus-1.2.0
____________________________________________________________________
Please help me how to solve this
I simply installed it and wrote lines of codes before it finally ran through and installed. The replies aren't always intelligent or accurate. What can I do make it better? Or if nobody here can help, can you at least direct me to a different subreddit for this?
Example:
You:Hey
ChatBot: - Is there a reason that I should?
You:How can I improve you?
ChatBot: - A robot cannot take offense.
You: Do you have emotions?
ChatBot: - With sufficient programming and training, I am capable of emulating some emotions. I don't know if that's the same as feeling them, though.
You:What is the meaning of life?
ChatBot: - is a literary reference to the illuminatus.
Figured I should get around to trying out chatbots, seems everyone else is doing one...
obviously there are a few different stacks to choose from, worked through tutorials, some good, some outdated, some on hold.
What stack do you use for your chat bots, and what features/services work well for you?
Not sure I'm posting to the right subreddit. If I'am not, feel free to redirect me to the right place. Thanks.
I was just wondering, with the current pseudo AI technology, is there any manageable way for me to upload my life to a "context learning" chatterbot who could then answer questions as I would do? My personality (or traits of it) are already collected by Google and Facebook and other services. If I could just upload that data into a learning chatterbot, maybe it could learn to speak like me, answer like me, interact as me. Knowing the people I know, knowing the places I've been, etc.
The ultimate goal would be to be able to come back to my relatives if I die, if possible as a "ghost" through augmented reality, a bit like Obi-Wan comes back to Luke. And to be able to talk to myself...whatever...
So, is it possible now? How do I do it? And if not.. I'll try again in 10 years.
