Katoptron (chatbot)

Overview:

This is an Artifical Intelligence experiment in order to demonstrate learning through interaction. It is on the level of a toddler and can be "stubborn".

This entity learns your way to respond to challenges - and after some time begins to mimic your behaviour. - It will reason based on your information, but its reasoning mechanisms will be different from yours. It will slowly learn. "Canned text" is NOT its modus operandi, as in common chatbots.

This is NOT yet another variation of the classic "Eliza" chatbot from the 1960s. Eliza DOES NOT LEARN. The Katoptron DOES. NOT using "canned text" means it can sometimes repeat in a less polished way than Eliza; but it will do better as you converse with it. It comes pre-delivered with a very little bit of learning, about 6-7% of its general capacity.

It has been written in RFO-Basic - kind thanks to Paul Laughton for having created it and to Nicolas Mougin for the easy-to-use app-creator.

Usage:

I tried to make it look "retro" - I aimed at 1970s/1980s Science Fiction movies... So it is a bit "command-linish":

1. Touch the screen or a volume button to trigger the prompt.

2. Type your text into the dialog box and press OK.

3. Wait for the response.

4. To terminate the conversation, just press "OK" in an empty dialog box without having entered any text or press the backbutton.

Used files:

In the folder "Katoptron" in your sdcard-directory, you will find a sub-folder "data" - it contains "svmemory.txt", which contains the saved memory of what the device has learned from you, as well as "chatlog.txt", which contains your conversation so far. - To reset the Katoptron, simply empty each file with a text editor of your choice.

Both files are exclusively used for internal operation of the program and are NOT used to transmit any data whatsoever about you.

Privacy Policy:

This app is not involved in the transmission of any data to or from your device. It does not collect or operate on any data except (a) such data which is internal to it as well as (b) such data that it learns from you during its interaction with you.

Exclusion of Liability:

To the extent permitted by law, I exclude all liability for any claim, liability, harm or damage, whether direct or indirect, and however caused by the use of this program. Usage is provided "as is" - use at your own risk, this is free software.

- Nino

Recent changes:
UPDATE: Handled a bug where some replies could be particularly repetitive.

Touch the typed text on the screen to chat; press the BACK-button to exit. Giving more context, up to 240 chars, lets it respond better. Type correctly, in one line.

This is an AI experiment, not a "chatbot"! Keep that in mind - its replies will be less "polished" than "canned text" and may be repetitive, but it will LEARN (as opposed to a "canned text" bot). Have patience - you are handling a toddler! Enjoy! :)