Method and apparatus for generating causal explanations using models derived from data

An apparatus that is configured to transform an input causal model into a format that can be expressed causally in one or more phrases with a narrative representation of the input causal model data, the apparatus comprising at least one processor and at least one memory including computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to at least:

• obtain an input causal model representing a data set composed of a plurality of nodes, a plurality of edges, and a plurality of paths, wherein the input causal model is created by machine learning and data mining algorithms, and wherein each node of the plurality of nodes representative of numerical or symbolic content and each edge of the plurality of edges representative of a causal relationship between nodes of the plurality of nodes, wherein a causal relationship identifies a relation between a cause and an effect;
• identify, by traversing each path of a plurality of paths of causally related nodes of the input causal model and determining an importance level for each path of the plurality of paths, a path of maximal significance of the plurality of paths;
• identify, by performing one or more of pruning or collapsing the path of maximal significance, a subset of nodes from the plurality of nodes and a subset of edges from the plurality of edges associated with the path of maximal significance, wherein pruning the path of maximal significance results in removal of nodes from the path of maximal significance and collapsing the path of maximal significance results in only those nodes of the plurality of nodes remaining in the subset of nodes that are deemed significant and relate to a causal explanation;
• and
• generate, using a natural language generation system that is configured to execute on a processor, one or more phrases with a narrative representation describing the subset of nodes and the subset of edges associated with the path of maximal significance.

A computer program product comprising at least one computer readable non-transitory memory medium having program code instructions stored thereon, the program code instructions which when executed by an apparatus causes the apparatus to at least:

• obtain an input causal model representing a data set composed of a plurality of nodes, a plurality of edges, and a plurality of paths, wherein the input causal model is created by machine learning and data mining algorithms, and wherein each node of the plurality of nodes representative of numerical or symbolic content and each edge of the plurality of edges representative of a causal relationship between nodes of the plurality of nodes, wherein a causal relationship identifies a relation between a cause and an effect;
• identify, by traversing each path of a plurality of paths of causally related nodes of the input causal model and determining an importance level for each path of the plurality of paths, a path of maximal significance of the plurality of paths;
• identify, by performing one or more of pruning or collapsing the path of maximal significance, a subset of nodes from the plurality of nodes and a subset of edges from the plurality of edges associated with the path of maximal significance, wherein pruning the path of maximal significance results in removal of nodes from the path of maximal significance and collapsing the path of maximal significance results in only those nodes of the plurality of nodes remaining in the subset of nodes that are deemed significant and relate to a causal explanation;
• and
• generate, using a natural language generation system that is configured to execute on a processor, one or more phrases with a narrative representation describing the subset of nodes and the subset of edges associated with the path of maximal significance.