BNet™.Builder

Category Intelligent Software>Bayesian Network Systems/Tools

Abstract BNet.Builder is a desktop application for rapidly creating Belief Networks, entering information, and getting results.

Bayesian belief networks --

Bayesian belief networks, or just “belief networks”, are advanced tools for reasoning about causes and effects in a wide variety of situations.

Bayesian belief networks are very effective for predicting outcomes (deductive reasoning) or diagnosing causes (abductive reasoning), given some prior information about the situation of interest.

They are presented visually as graphs, with nodes representing variables, and links representing the effect of one variable on another. They capture the relationships among variables as probabilities based on historical information about those relationships.

They are particularly well suited to modeling situations where information is incomplete, vague, conflicting, and/or uncertain.

Examples of how belief networks can be used –

Some real-life examples of how belief networks have been used include systems to help doctors choose proper tests and diagnosis illnesses, to aid banks in assessing whether or Not a loan applicant is likely to default, and to help anti-spam software determine which incoming email is junk.

BNet.Builder can be used to --

1) Quickly create graphical models of cause and effect.

2) Easily incorporate past experiences.

3) See the results of entering evidence about the current situation or hypothetical cases as you enter it.

4) See how incoming data affects results in real-time.

5) Clearly communicate how your model works.

BNet.Builder additional features & benefits --

1) BNet.Builder is one of the first products to bring true ease-of-use to Belief Network modeling.

2) An easy to use interface with tools to make your work go faster.

3) Dynamic Update -- See beliefs as soon as you change your network in any way, including: entering Conditional Probability Table (CPT) data, posting evidence, and deleting nodes or links. No need to compile, run, or update.

4) Easy CPTs -- See the CPT as soon as you select a node. Enter data in probability tables:

• a) Set multiple values in multiple rows at once using sliders.
• b) Drag and drop columns to group states with similar values.
• c) Copy and paste values from spreadsheets.

5) Single Mode Interface -- Build your network and see calculated beliefs in one single window, Not in separate "build" and "run" modes.

6) Expandable Nodes - Use nodes to show information and to add information:

• a) Show beliefs.
• b) Show evidence.
• c) Post and retract evidence.
• d) Add or delete states.

7) Extensive help --

• a) Detailed User Guide.
• b) Quick Start Guide for experienced users.
• c) Introduction to Bayesian Belief Networks.

System Requirements

BNet.Builder Hardware Recommendations

• 1.5 GHz processor
• 256 MB RAM
• 40 MB hard-drive space
• Internet connection for receiving updates (optional)

Operating System Requirements

BNet is unique in that its Java platform is easily adaptable to most operating environments.

	OS		Hugin		Netica		BNet.Builder
	Windows 95/98/NT/2k/XP		Y		Y		Y
	Mac OS X		N		N		Y
	Linux		Y		N		Y

Note: the processor speed and memory capacity specified above are only recommended values, not minimum required values. BNet.Builder should work well with most networks with the above processor speed and memory capacity. Slower processor speed and smaller memory capacity will result in degraded performance. Faster processor speed and larger memory capacity will increase performance of BNet.Builder and allow for easier handling of larger networks.

Manufacturer

• Charles River Analytics Inc.
• 625 Mt. Auburn St.
• Cambridge, MA 02138 USA
• Tel: (617) 491-3474
• Fax: (617) 868-0780

Manufacturer Web Site BNet.Builder

Price $650; Support Contract $165/year

G6G Abstract Number 20214

G6G Manufacturer Number 100512