A Simple Dynamic Categorizer for Chandler (CATY)

AboutMyMotivation20031223

Summary

Would it not be nice if all your electronic stuff (documents, presentations, spreadsheet and email) would be also sorted and related automatically for you?

While sophisticated scientific approach tries to capture �user-context� for knowledge management or research in �dynamic reasoning�, I am going to outline �no-white-smoke� approach here, which relies on the paradigm of �Multi-Dimensional Categorization� trough a �Dynamic Taxonomic Indexer� to relate to the users current context through a combination of simple tools. Well there is a semantic component there, but we will come to that later. I call my idea CATY for no specific reason, just to give it a nice name.

Most algorithms CATY is made from are already available and known as count, sort, diff, grep, software-summarizer and the like. Think of it as a toolbox of already commonly known things. Without being too high-flying, there is some innovation in the logic these things are put together to produce the desired results.

With this in Mind, CATY should be fast to implement into what I can see of Chandler today (0.2 Release), delivering the �cool features� to the Canoga targeted audience in time. It is no matter of how much RDF will be in Canoga. It is more a question of what code will be shipped with Canoga, which exposes knowledge concentric functionality. My bottom line is here: very little, but essential RDF � as much rich functionality through user-configurable common algorithms, as we can get reliably implemented.

What exactly would automatic categorization do for you? First of all it would provide another way of looking at documents and information, through introducing a reliable categorical paradigm. Along with the new look, there comes another way of organizing your work, which is more category- and content- concentric.

All right � this sounds nice - but where is the extra bells and whistles with �automated categorization�?

Automated Categorization, while working with Chandler�

• will instantly relate any information in the users interest to existing categories, whether these categories are found on your own machine, or found through the network.
• will keep a history of what you worked on sorted by content and �meaning�.
• will help to find meaningful references to related information.
• will allow you to share and promote any categorized information with everybody across the network.
• will allow you to find and network people by interest and meaning

Chandler also implements the concept of Agents, this means in a defined situation a piece of software will wake up and start working. It is easy to underestimate the powers in this idea by just putting it into simple words. Consequentially, the agent interface will have total access to all user- and system-events, while chandler is running. Possibly in the end chandler does not do run after all, but I am not so deep into OSAF�s ideas on Agents so far.

Automated Categorization, while Chandler is working for you�

• will act on any events happening, by looking into the �meaning� of accompanying information by categorization.
• will do more predefined complex assistance in the context of your current work.

In fact, the way these things are told to interact and their specific results are interpreted should be highly user-configurable. The possibility to define and interact with the logic arranging the general composition of a higher-value function and possible complex categorizing tool is essential. It enables the user to take a predefined logic, which produces generally desirable results and change them to match more specific needs. An individual optimization of the categorizations �behavior� through �adjustment� would be possible.

If you feel, these ideas as compelling as I do and I have raised your interest please be invited to continue reading into the next section for more detail of what I mean. I�d be also delighted to receive any feedback on these ideas.

About this Document�

This document has three main Sections. In Section 1 I will show some general requirements from my own experience, which will outline the reach of my approach. Then in Section 2 I will go into the explanation of the necessary environment and toolbox-functions and finally present in Section 3 some examples of how I imagine these things to be installed in Chandler.

CatySectionOne20031223 � What can Categorization do for me?

CatySectionTwo20031223 � What is the environment and toolbox for Categorization?

CatySectionThree20031228 � How will Automated Categorization be implemented in Chandler?

CatyDiscussion^? � What do you think?

-- BernhardGroehl - 27 Dec 2003