Project Knoware

SYNOPSIS

An interactive application which relies on user-volunteered system statistics and bug/behavior reports to identify patterns between the two using inductive logic. The goal is to help developers and normal users solve system-related issues easily and rapidly, while involving them in the community.

MOTIVATION

Imagine: An end to forum-based tech support that will revolutionize bug fixing and problem solving.

You open an application that networks you with dozens, hundreds, or thousands of people, but you don't know any of them. You're related by various pieces of your Linux System configuration -- hardware and software information you voluntarily contributed. You're also related because you've subscribed to similar Problems -- bugs or issues of unknown origin that you'd like to fix. Based on statistics and user input, Knoware has determined that one of your Problems is video card related. For example, data collection shows that only people who have ATI cards subscribe to a specific Problem, which may not appear to have any relation to graphics cards. As more users subscribe, it can be narrowed down to a specific generation of cards, or even a specific driver setting for that generation. Knoware analyzes thousands of System configurations and Problem subscriptions and, using inductive reasoning, finds the most likely causes of your Problems. Since Knoware's reach is limited by user input and knowledge of the system, it has also has connected you with the users necessary to communicate, tweak, and log the settings, and gives you the means necessary to post a solution once found. Then, everyone subscribed to that Problem (and even people Knoware has determined *probably do* have the Problem, based on their System) are notified of the solution, and all rejoice!

GOALS

Linux is a platform commonly plagued with hardware and configuration issues, and this is intimidating to many users. With so many possible system configurations, it's easy to see how this has become an issue. Linux can empower a user more than any other platform, but it often does just the opposite. Knoware unveils the root of complex problems and gives users a starting point for solving them. KDE already does a lot to make Linux viable for people -- Knoware expands upon this area. And where the programmed logic fails, human communication prevails. Knoware will:

Conceptually:

• Connect users who face similar issues based on their system configuration
• Use system and problem statistics with inductive logic to solve problems
• Track problems and notify users of solutions

Deliverables:

• A library for gathering system information and 'normalizing' its format to be useful as input to Knoware's classification system (might already exist)
• A library that uses classification hierarchy and statistics to recognize patterns (Bayes nets?) between problem areas and system configuration
• A database layer for managing statistics and user input (configuration records, keyworded problem descriptions, etc.)
• A client interface that unites the above with live 'data browsing' and communication geared towards solving the user's problems

PROJECT DETAILS

This is an ambitious project which relies heavily on user participation in order to succeed. Requirements and open issues include:

Server:

• A beefy server and well-managed database to handle all the statistics and input
• A web interface to the problem database and related statistics

Logic: