topics:	main-page   everything   99things   things-to-do   software   space   future   exercise & health   faith
	thought   web   movies+TV   music   mymusic   food   curiosity   tidbits   I remember   wishlist   misc   links

This section lists all blog posts, regardless of topic.

Exercise 30: Web UI
October 18, 2008

Summary

An in-house AI's main interface is voice communication, but it should also support remote interaction via the web.

Create a web UI to view, add to, and clear the current grocery list. Add a web service to Grace to allow the ASP.NET page to submit a textual query and get a string response back.

Solution

Since Grace is a standalone Windows application (not an ASP.NET web site), adding a web service isn't quite as straightforward. It requires the Web Service Extension 3.0 and a URI such as soap.tcp://99.236.137.8/graceservice.

Aside from the web service quirks, implementing this functionality was actually quite straight forward!

Click here

Web UI

Click here

NOTE: This will only work if my PC is running.


Exercise 29: Queries
October 18, 2008

Summary

Implement the following queries:

1.	What time is it?

2.	What is {first_name}'s phone number?

3.	We need more {food_item}.

4.	What's on the grocery list?

5.	Clear the grocery list.

Support at least 10 phone numbers and 25 grocery items.

Implementing the "What's on the grocery list?" query will involve enhancing the query system so that it can return multiple results. Implementing the "Clear the grocery list" command will involve some new syntax to allow a relationship to be cleared.

Web UI

Click here

Grace application

Click here

NOTE: The Grace application requires that you have Expressivo installed for text-to-speech.


Exercise 28: Integrating Plato with Grace
October 17, 2008

Summary

The goal of this exercise is to integrate the Plato engine with the Grace UI, and to demonstrate the temperature() functionality.

Combine the Grace and Plato projects into a single solution.

Create data files (.txt) to represent each of Plato's data structures, and have Grace load those into memory when it initializes.

Every 5 seconds, have Grace examine the file sizes and reload them as needed.

Create a new data structure for Grace which will represent the phrases that will be recognized. They may use entity types, such as:

what is the current temperature in {city}?

Optional words are denoted with round brackets, and vertical bars can be used to denote several options. For required words that have several options, use square brackets. For example:

[what is|what's] the (current) temperature in {city}?

Transform these specifications into an SR grammar, write it to disk, compile it, and then load it.

When a statement is recognized, pass it on to Plato.

If Plato can't parse the statement/question, abort all operations. Simply do nothing.

If Plato's output is a value, display it and speak it.

If Plato's output is an assignment or relationship, briefly display a full-screen green window. (1 second)

If Plato's output is a TransOutput_Program, display a full-screen gray window while the query executes. If the query completes properly, display and speak the answer. If it fails, display a full-screen red window for 1 second.

Test the setup in a kitchen with a laptop, speakers, and a Bluetooth headset.

...

older >>