I have a set of nodes which they are parents nodes what I know about them is their mean and the standard deviation. So I assumed that they are continuous and they are normally distributed. My first question can I build node with this information with no need to discretize the node to different states. My second question if I need to have a child node from all the parents and want my the node value to be a linear regression model, so I have the value of the node to be equal to value of other nodes times the correlation coefficient.
BayesiaLab handles continuous variables by discretizing them. However, you will be able to use the equation editor to define the normal distributions of the parent nodes, as well as the relation of these parents with the child.