You are on page 1of 28

UNIT 5

PROBLEM DECOMPOSITION
• SAINT
• Dendral
• Goal Trees
• Rule Based Systems
• XCON
• Rule Based Expert Systems
1.AND Edges:
1. In the context of AI or computational models, AND edges represent a logical
AND operation. This means that for a signal or information to pass from one
node to another connected by an AND edge, all the incoming edges must be
active or carry a signal. It's analogous to the logical AND gate in digital logic.
2.OR Edges:
1. Conversely, OR edges represent a logical OR operation. In this case, for
information to flow from one node to another connected by an OR edge, at
least one of the incoming edges must be active. It's analogous to the logical
OR gate in digital logic.
• A B
• \ /
• AND
Node C has an AND edge coming from A and B, meaning
• |
both A and B must be true for the signal to pass through
• C C. The signal from C then goes to a node with an OR
• | edge, where either D or E being true would allow the
signal to continue.
• OR
• /\
• D E
DENDRAL
DENDRAL is one of the earliest expert systems, developed in the 1960s at
Stanford University by Edward A. Feigenbaum, Joshua Lederberg, and
their colleagues.

DENDRAL stands for "DENDRitic ALgorithm," and it was designed to


interpret mass spectrometry data for molecular structure determination,
particularly in the field of organic chemistry.
• Key features and contributions of DENDRAL include:

1.Domain-Specific Expertise: DENDRAL was tailored for a specific domain—


analyzing and interpreting mass spectrometry data to identify the structure of
organic compounds.
2.Rule-Based System: It used a rule-based approach, where a set of rules
represented the knowledge of human experts in the field of chemistry. These
rules allowed the system to make inferences about the structure of molecules
based on the input data.
3.Inference Engine: DENDRAL had an inference engine that used forward
chaining, a reasoning approach where the system starts with known facts and
applies rules to generate new conclusions.
4.Heuristic Reasoning: The system employed heuristic reasoning, allowing it to
make educated guesses and prioritize hypotheses based on likelihood.
EXPERT SYSTEM
An expert system is a computer program that is designed to solve complex problems
and to provide decision-making ability like a human expert.

It performs this by extracting knowledge from its knowledge base using the reasoning
and inference rules according to the user queries.
Components of knowledge base
XCON
• XCON (short for "Expert CONfigurer")
is another classic expert system that was
developed at Digital Equipment
Corporation (DEC) in the late 1970s
and early 1980s.
• XCON is one of the most well-known
examples of an expert system used for
configuration tasks.
1.Configuration Expertise: XCON was designed to configure computer
systems based on customer requirements. It could determine the appropriate
combination of hardware and software components to meet specific user
needs.
2.Rule-Based System: Like DENDRAL, XCON was a rule-based system. It
utilized a knowledge base consisting of rules that encoded the expertise of
human configurators.
3.Forward Chaining Inference Engine: XCON used forward chaining, where
the system starts with known facts and applies rules to reach conclusions. This
approach is particularly suitable for systems that involve making decisions
based on available information.
4.Scalability: XCON demonstrated the scalability of expert systems by
handling complex configuration tasks for DEC's VAX computer systems. It
could efficiently navigate through a large solution space to arrive at a valid
and optimal configuration.
GOAL TREES
• In the context of artificial intelligence (AI), goal trees can be used as a
tool for problem decomposition and planning, similar to their
application in systems engineering.
• When working on AI projects, goal trees help organize and break
down the overarching objectives into more manageable sub-goals and
tasks.
1.Define the AI Objective:
1. Start by clearly defining the main objective of your AI project. This could be a high-
level goal such as developing a recommendation system, natural language processing
application, or image recognition model.
2.Identify Sub-Goals:
1. Break down the main AI objective into more specific sub-goals. For example, if the
main goal is to build a recommendation system, sub-goals could include data
collection, feature engineering, model training, and user interface development.
3.Decompose Further:
1. Continue breaking down sub-goals into smaller tasks until you have a detailed plan.
For instance, under "model training," you might have sub-tasks like data
preprocessing, model selection, hyperparameter tuning, and model evaluation.
1.Hierarchy of Tasks:
1. Organize the sub-goals and tasks in a hierarchical structure. This hierarchy helps
visualize the relationships between different components of the AI project and ensures
that each task contributes to the overall success of the project.
2.Dependencies and Order:
1. Identify dependencies between tasks. Some tasks may need to be completed before
others can begin. Understanding these dependencies is crucial for effective project
planning.
3.Assign Responsibilities:
1. Once the goal tree is established, it becomes easier to assign responsibilities. Different
team members or groups may be responsible for specific sub-goals or tasks.
4.Monitor and Adjust:
1. Use the goal tree to monitor progress throughout the project. As tasks are completed,
mark them off on the tree. If adjustments need to be made due to changes in
requirements or unexpected challenges, the goal tree provides a visual guide for
adapting the plan.

You might also like