Method and system for utilizing domain knowledge to automatically identify solution to a problem

A system (100) for utilizing domain knowledge to identify solution to a problem, the system comprises:

• an input module (102) for capturing domain knowledge using one or more of the following sources:
• a domain expert using a knowledge based questionnaire,
• parsing the web and enterprise repository based on relevant concepts and keywords, and
• a plurality of external sources;
• a memory (104);
• and
• a processor (106) in communication with the memory, the processor further comprising:
• a transformation module (108) for transforming the captured domain knowledge in form of ontologies and instances of knowledge by applying natural language processing and template matching techniques;
• a lookup table creation module (110) for creating a contextual domain knowledge (CDK) look up table for a set of existing problems to store a mapping of contextual knowledge and meta-data features around each existing problem to corresponding solution pipelines, wherein the solution pipelines are a plurality of subtasks required for solution implementation, with each subtask comprising a related knowledge and associated algorithm list stored as the mapping in a knowledge store;
• a problem definition module (112) for obtaining the problem information from a user in the form of problem files, wherein the problem information comprises information mapped to the CDK lookup table;
• a recommendation module (114) for recommending a sequence of the solution pipelines by matching the problem with set of existing problems in the CDK look up table and fetching corresponding pipeline solution steps, wherein the recommendation is performed by:
• a sematic based recommendation;
• and
• a situation and sequence aware similarity matching by projecting features around the problem and solution in a n-dimensional feature space and recommending the solution pipelines by applying a vector based similarity, wherein a union among the solution pipeline steps is taken if the similarity is found in more than one solution feature;
• a PDDL transformer module (116) for converting the domain knowledge and the problem files into planning domain definition language (PDDL) files;
• and
• a planning module (118) for applying a planning algorithm on the PDDL files to order the recommended solution pipeline steps and provide an ideal solution path among many solution paths in search space by matching constraints with a predefined set of conditions.

One or more non-transitory machine readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause managing a plurality of events, the instructions cause:

• capturing via one or more hardware processors, domain knowledge using one or more of the following sources:
• a domain expert using a knowledge based questionnaire,
• parsing the web and enterprise repository based on relevant concepts and keywords, and
• a plurality of external sources;
• transforming via one or more hardware processors, the captured domain knowledge in form of ontologies and instances of knowledge by applying natural language processing and template matching techniques;
• creating via one or more hardware processors, a contextual domain knowledge (CDK) look up table for a set of existing problems to store a mapping of contextual knowledge and meta-data features around each existing problem to corresponding solution pipelines, wherein the solution pipelines are a plurality of subtasks required for solution implementation, with each subtask comprising a related knowledge and associated algorithm list stored as the mapping in a knowledge store;
• obtaining via one or more hardware processors, the problem information from a user in the form of problem files, wherein the problem information comprises information mapped to the CDK lookup table;
• recommending via one or more hardware processors, a sequence of the solution pipelines by matching the problem with set of existing problems in the CDK look up table and fetching corresponding solution pipeline steps, wherein the recommendation is performed by:
• a sematic based recommendation;
• and
• a situation and sequence aware similarity matching by projecting features around the problem and solution in a n-dimensional feature space and recommending the solution pipelines by applying a vector based similarity, wherein a union among the solution pipeline steps is taken if the similarity is found in more than one solution feature;
• converting via one or more hardware processors, the domain knowledge and the problem files into planning domain definition language (PDDL) files;
• and
• applying via one or more hardware processors, a planning algorithm on the PDDL files to order the recommended solution pipeline steps and provide an ideal solution path among many solution paths in search space by matching constraints with a predefined set of conditions.