Knowledge extraction and abstraction

A system for extracting and abstracting knowledge embedded in an input data stream, comprising:

• a processor;
• a non-transitory computer readable storage medium;
• a contextual taxonomy store segment configured to store patterns;
• a pattern abstraction engine configured to receive the source code in a programming language;
• a pattern hatcher engine configured to create a dynamic pattern using fuzzy-neural rules and dynamic rules when no predefined pattern match a part of the source code;
• a pattern filtering engine configured to determine whether to store the dynamic pattern in the contextual taxonomy store segment;
• a pattern conversion engine configured to extract a plurality of knowledge elements from the source code based on predefined patterns and to identify the plurality of knowledge elements in an abstract representation;
• a contextual pattern decoder engine configured to determine knowledge attributes for the extracted knowledge elements by extracting keywords and descriptions from the source code, and to identify the knowledge attributes associated with the plurality of knowledge elements in the abstract representation;
• and
• a pattern classification engine configured to classify each of the extracted knowledge elements into one of a data entity, a variable, and a business rule, using classification patterns and associated knowledge attributes, wherein each of the classification patterns comprises a predefined pattern or a dynamic pattern, wherein the pattern abstraction engine is further configured to generate the abstract representation of the source code in a standard format independent of the programming language of the source code, and wherein the abstract representation includes the plurality of knowledge elements.

A non-transitory computer readable storage medium structured to store instructions executable by the processor to extract and abstract knowledge embedded in source code of an application the instructions when executed by the processor cause the processor to:

• receive the source code in a programming language;
• extract a plurality of knowledge elements from the source code based on predefined patterns
• responsive to no predefined pattern matching a part of the source code, creating a dynamic pattern using fuzzy-neural rules and dynamic rules and determine whether to store the dynamic pattern in a store;
• determine knowledge attributes for the extracted knowledge elements by extracting keywords and descriptions from the source code
• classify each of the extracted knowledge elements into one of a data entity, a variable, and a business rule, using classification patterns and associated knowledge attributes, wherein each of the classification patterns comprises a predefined pattern or a dynamic pattern;
• generate an abstract representation of the source code in a standard format independent of the programming language of the source code, wherein the abstract representation includes the plurality of knowledge elements;
• and
• identify the plurality of knowledge elements with associated knowledge attributes in the abstract representation.