RoboMatic Documentation

X
  • Introduction
  • Getting Started
  • Non-developers Guide
  • Developers Guide
  • Embedded Script
  • API - REST
  • API - Websocket
  • Entities
  • Entity Relations
  • User Inputs
  • Extractors
  • Dynamic Entity Relations using Extractors
  • Chatbot Outputs
  • Output Groups
  • RoboMatic Thoughts
  • Abilities & Operations
  • Short-term & Long-term Memory
  • Insights
  • RoboMatic Habits
  • Chatbots' Intelligence
  • Intelligence Marketplace
  • Chatlogs
  • Unsupervised Learning
  • Third-party Integrations
  • About RoboMatic
  • Privacy Policy
  • Terms & Conditions
  • Technical Troubleshooting
  • Frequently Asked Questions
  • Known Issues
  • Sentence Components
  • Acceptance & Rejection Thoughts
  • Short-term Thoughts
  • Long-term Thoughts
  • Inline Thoughts
  • Users' Sessions Thoughts
  • Constructive Thoughts
  • Suggested Inputs
  • RoboMatic Interjections
RoboMatic has the ability to observe and learn from unlabeled data of different sources using algorithms that extract, classify and cluster data.

There are several concerns about Unsupervised Learning that still needs to be addressed before allowing any chatbot to use this technique safely.

Output Relevancy

No doubt a good algorithms should be able to generate a relevant responses to user's request. While some learned responses might be generically used to many user's requests, however the goal is to be able to produce a response that's very relevant to the subject and the context of the conversation.

Output Accuracy

A chatbot should be able to generate responses that are correct, authentic and acceptable by humans. This is very challenging because the data can be easily manipulated by the public, therefore a system needs to be set to carefully verify the information autonomously.

Ethical Supervision

A chatbot should be able to generate responses that doesn't offend or harm any human in any form. While the generated response could be relevant and accurate to the user's request, but still should not be presented to a user. This can't be done without setting a code of conduct rules that supervise the chatbot behavior. These rules are hard coded in the unsupervised learning engine.

Tracing

Understanding how a chatbot generated a response is critical. The algorithms are evolving each time a chatbot learn new rules. It gets very complicated to understand how responses are generated, therefore a system needs to be set to monitor the learning methodology, which should be useful to improve the overall quality of the responses.

Fail-safe Protocol

In the event of unexpected responses or actions done by a chatbot, a fail-safe protocol should be in place to reverse, minimize or eliminate any harm caused and to avoid the consequences of the system's failure. As RoboMatic designed to control software and hardware, supervised rules should be set to request permissions from users before proceeding to take certain actions. Autonomous backups or actions can be managed by a system before letting a chatbot proceeding with a certain action.
Powered by:
     

The Next Generation Bot • v4.1 (Beta)
RoboMatic is the next generation bot system that can assist people in their daily life. It can be integrated in desktop, web, mobile or IoT applications to be used in businesses or personal purposes. RoboMatic intelligence is improving every second by teachers all over the world. A free API for developers and much more!

User Login

Forgot password?
Create free account.