Intelligent recommendation system

A computer system to generate vector matrixes for determining recommendations for objects and subjects, the system comprising:

• a processor;
• and
• a memory in association with the processor storing instructions related to:
• iteratively evaluating purchase data to determine mappings associating one or more objects from the objects with one or more subjects from the subjects;
• based on the determined mappings, generating an object vector matrix based on average measures over subject vectors included in an initial subject vector matrix;
• generating an updated subject vector matrix including updated subject vectors based on the generated object vector matrix and the purchase data;
• iteratively generating a set of consequent versions of the object vector matrix and a set of consequently generated versions of the updated subject vector matrix, wherein the iterative generation starts with the object vector matrix and the updated subject vector matrix as an input, and wherein generation of a consequent version of the object vector matrix from the set of consequent versions of the object vector matrix is based on a previously generated version of the subject vector matrix from the set;
• based on applying a validation criteria over the iteratively generated version of the object vector matrix and the versions of the subject vector matrix, defining a final object vector matrix and a final subject vector matrix to be provided for determining a recommendation for the objects and the subjects;
• computing vector distance measures between a first object vector and a set of other object vectors at the final object vector matrix, wherein a first object associated with the first object vector is associated with a mapping to a first subject;
• comparing the computed distances to determine a result object vector from the set of other object vectors to be a closest object vector, when a distance between the result object vector and the first object vector is lowest from the vector distances;
• and
• providing an identification of a result object associated with the result object vector as the recommendation, the recommendation being provided to one or more mapped subjects to the result object in the determined mappings.

A non-transitory computer-readable medium storing instructions, which when executed cause a computer system to:

• analyze purchase data to determine mappings associating one or more objects from the objects with one or more subjects from the subjects;
• based on the determined mappings, generate an object vector matrix based on average measures over subject vectors included in an initial subject vector matrix;
• generate an updated subject vector matrix including updated subject vectors based on the generated object vector matrix and the purchase data;
• iteratively generate a set of consequent versions of the object vector matrix and a set of consequently generated versions of the updated subject vector matrix, wherein the iterative generation starts with the object vector matrix and the updated subject vector matrix as an input, and wherein generation of a consequent version of the object vector matrix from the set of consequent versions of the object vector matrix is based on a previously generated version of the subject vector matrix from the set;
• and
• based on applying a validation criteria over the iteratively generated version of the object vector matrix and the versions of the subject vector matrix, define a final object vector matrix and a final subject vector matrix to be provided for determining a recommendation for the objects and the subjects;
• compute vector distance measures between a first object vector and a set of other object vectors at the final object vector matrix, wherein a first object associated with the first object vector is associated with a mapping to a first subject;
• compare the computed distances to determine a result object vector from the set of other object vectors to be a closest object vector, when a distance between the result object vector and the first object vector is lowest from the vector distances;
• and
• provide an identification of a result object associated with the result object vector as the recommendation, the recommendation being provided to one or more mapped subjects to the result object in the determined mappings.