Data structure description

Variables

variables=list()
The element type is :dict
variable
{
User-provided attributes:
'var_id': Variable ID, int type, counting from 0
'is_easy': Whether this variable is a easy instance, the value is True or False
'is_evidence': Whether this variable is an evidence variable, the value is True or False * 'label': Inferred labels: 0 is negative, 1 is positive, -1 is unknown
'true_label': The true label of this variable
'feature_set': All feature information owned by this variable
{
feature_id1: [theta1,feature_value1],
feature_id2: [theta2,feature_value2],
...
}

Attributes that may be automatically generated while the code is running
'inferenced_probability': Inferred probability
'probability': Inferred probability
'evidential_support': Evidence support
'entropy': Entropy
'approximate_weight':Approximate weight
'approximate_probability': Approximate probability
...

}

Features

features = list() #The element type is: dict:feature
feature
{
User-provided attributes
'feature_id': The id of this feature, int type, counting from 0
'feature_type': Whether this feature is a single factor feature or a dual factor feature, currently supports both unary_feature and binary_feature
'feature_name': Feature name, optional
'alpha_bound':[bound0,bound1] alpha's Upper and lower bounds
'tau_bound':[bound0,bound1] tau's Upper and lower bounds
'parameterize':type of int , indicating whether the feature of this type is parameterized
'weight': Information about all relevant variables of this feature
{
var_id1: [weight_value1,feature_value1],
(varid3,varid4): [weight_value2,feature_value2],
...
}

Attributes that may be automatically generated while the code is running
'tau': tau value
'alpha':alpha value
'regerssion'： Linear regression results

}