# US 5,704,018 A · Generating improved belief networks > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 5,704,018 A - Document ID: US5704018A - Status: Granted - Kind: A - Application number: 08/240,019 - Filed: 1994-05-09 - Granted: 1997-12-30 - Earliest claimed: 1994-05-09 - Assignees: Microsoft Corporation - Inventors: David E. Heckerman; Dan Geiger; David M. Chickering - Web page: https://patentgrep.com/patent/US5704018A ## CPC - G06N 5/022: Knowledge engineering; Knowledge acquisition - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract An improved belief network generator is provided. In a preferred embodiment of the present invention, a belief network is generated utilizing expert knowledge retrieved from an expert in a given field of expertise and empirical data reflecting observations made in the given field of the expert. In addition to utilizing expert knowledge and empirical data, the belief network generator of the preferred embodiment provides for the use of continuous variables in the generated belief network and missing data in the empirical data. ## Claims ### Claim 1 In a computer system, a method for generating a belief network for assisting in making a decision, comprising the steps of: receiving expert knowledge specifying factors that causally influence the decision, an indication of relationships between the factors, and an indication of a strength of the relationships between the factors; receiving empirical data, the empirical data comprising cases of observations of the factors; creating a belief network using the received expert knowledge; and modifying the created belief network utilizing the received empirical data. ### Claim 2 (depends on 1) The method of claim 1 wherein the expert knowledge contains variables corresponding to the factors, one type of variable being a discrete type of variable having a value selected from a set of discrete values, and wherein the step of receiving expert knowledge includes the step of receiving the expert knowledge containing the discrete type of variable. ### Claim 3 (depends on 1) The method of claim 1 wherein the expert knowledge contains variables corresponding to the factors, one type of variable being a continuous type of variable having a value selected from a set of continuous values and wherein the step of receiving expert knowledge includes the step of receiving the expert knowledge containing the continuous type of variable. ### Claim 4 (depends on 1) The method of claim 1 wherein the expert knowledge contains variables corresponding to the factors, one type of variable being a variable of a discrete type having a value selected from a set of discrete values, one type of variable being a variable of a continuous type having a value selected from a set of continuous values, wherein the step of receiving expert knowledge includes the step of receiving the expert knowledge containing both the discrete type of variable and the continuous type of variable. ### Claim 5 (depends on 1) The method of claim 1 wherein at least one case in the empirical data is a missing data case in which at least one factor does not have an observation, and wherein the step of creating the belief network includes the step of approximating an observation for the factor that does not have an observation. ### Claim 6 (depends on 1) The method of claim 1 wherein the step of receiving expert knowledge includes the step of receiving the expert knowledge from a human expert. ### Claim 7 A computer system for assisting in making a decision comprising: means for providing expert knowledge containing factors that causally influence the decision, an indication of relationships between the factors, and an indication of a strength of the relationships between the factors; means for providing empirical data, wherein empirical data contains cases of observations of the factors; and a belief network generator for generating a belief network utilizing the expert knowledge and for modifying the generated belief network utilizing the empirical data. ### Claim 8 (depends on 6) The computer system of claim 6 wherein the expert knowledge is received as part of a second belief network. ### Claim 9 (depends on 8) The computer system of claim 8 wherein the second belief network has a plurality of variables, one type of variable being a variable of a discrete type having a value selected from a set of discrete values. ### Claim 10 (depends on 8) The computer system of claim 8 wherein the second belief network has a plurality of variables, one type of variable being a variable of a continuous type having a value selected from a set of continuous values. ### Claim 11 (depends on 7) The computer system of claim 7 wherein at least one case is a missing data case containing a missing observation for at least one factor, further including: means for approximating an observation for the factor in the missing data case having the missing observation. ### Claim 12 (depends on 7) The computer system of claim 7 wherein the generated belief network contains a plurality of variables, one type of variable being a continuous type of variable having a value selected from a set of continuous values. ### Claim 13 (depends on 7) The computer system of claim 7 wherein the generated belief network contains a plurality of variables, one type of variable being a discrete type of variable having a value selected from a set of discrete values. ### Claim 14 A computer-readable medium whose contents cause a computer system to generate a belief network for assisting in making a decision, by performing the steps of: receiving expert knowledge comprising factors that causally influence the decision, an identification of relationships between the factors, and an indication of a strength of the relationships between the factors; receiving empirical data, the empirical data comprising cases of observations of the factors; creating a belief network using the received expert knowledge; and modifying the created belief network utilizing the received empirical data.