Welcome to the Knowledge Inter-Operative Coputing (KIOC) project site.
KIOC project aims to construct an expert system specialized in causal analysis with the following features.
- Performs inference over distributed knowledge. Hence, each knowledege owner or group can maintain own domain knowledge, and consolidation and inference/expanstion of knowledge are done by the system.
- Performs causal analysis in both directions; (1) backward chaining (identifying possible causes of an event) and (2) forward chaining (find what would happen based on a given event)
- Performs inference under uncertainty (missing/wrong info, various strength of symptom manifestations, and various levels of belief for giving observations)
- Automatically adjusts belife strengh of given knowledge based on observations.
- Inputs can be made by human or external systems synchronously or asynchoronously. In many cases, not all information is readly avaialble. Extracting some information may take time or costly while other information can be obtained in real-time. As a new evidence is supplied to the system, it re-evaluate the event and adjust possilibities assigned to causes.
Please note that in this site KIOC may refer to any of the following:
- Project name
- A set of software/protocal specifications
- A server or a cluster of servers hosting explicit/inferred knowledge and exchanging knowledge with peer servers.
What technologies is this system based on?
The system maps given and inferrred knowledge into semantic network and bayesian network. The distribution model is influenced by Sematic Web.
What technologies this system is NOT based on?
The inference is not based on first order logic. Hence, it's not capable of producing a result in a form of ture/false. It's always produce a result with any numeric value between -1 and 1 (e.g., -0.15 represents weak disbelif on a givien fact)
What are possible usages of this system?
KIOC was originally designed for diagnosing events/symptoms observed in hyper complex and distributed systems carrying a high level of uncertainty where expert knowledge is also highly distributed. If your organization has multiple expert teams working together, then their knowledge may be hosted in KIOS.
What are not possible usages of this system?
If you have a small-to-medium set of knowledge or rules, then using other rule engines or inferencers may suffice.
Who is the owner of this project?
Aki Sanagi is the initiator and owner of this project.
What is the status of this project?
Currently laying out possible algorithm of performing the target inference and working on PoC codes.