It's Alive! A Dynamic Rule-Based AI Framework Essay

810 Words 4 Pages
Everyone has heard stories of AI taking over the world, like The Matrix, but not many have heard how it can help our world. There are two main types of AI – rule-based and learning. However, learning systems require decades of effort and several million lines of code just to produce one that is semi-believable. On the other hand, current solutions for rule-based systems are clunky and slow and usually outdated. These systems also focus on content retrieval. My project aimed to create a new, scalable framework for general purpose use. The framework would allow for dynamic rules, which are rules that allow for commands to be executed. My project also aimed to create layers on top of the framework that would provide sample
…show more content…
This would emphasize authentication & trust to allow agent to access and act on personal data.
3. VIRTUAL BUDDY is a fun chatbot that is implemented across many platforms, including the iPhone and Facebook. Virtual Buddy has over 370,000 rules in his database and yet takes an average of 0.001 seconds per query, making it extremely efficient. I tested that one server can support over 50 simultaneous conversations, and likely many more.
4. ELDERLY WATCHDOG is an AI system that can detect if a person has fallen and then can take appropriate action. Using the rule-based system, it determines the appropriate level of escalation to take it to as well as to determine if it was actually a fall or just lying down. This is important as the panic buttons out there would require the user to crawl to the system to activate it. I used the Wiimote’s accelerometer and it has a range of 100m from the user’s computer, the likely radius of a house and the surrounding property. I linked the Wiimote to the computer through a Bluetooth link and used GlovePIE to interface with it. The Bluetooth link, however, with the Wiimote was a little shaky and would tend to disconnect every half of an hour.
5. NEWS FILTER. Google searches return millions of results, with only a few thousand being relevant. By breaking down the contents of each article in the RSS feed, I am able to extract word frequencies. If the keywords

Related Documents