| CPC G06Q 30/0201 (2013.01) [G06F 16/285 (2019.01); G06F 16/9554 (2019.01); G06K 19/06028 (2013.01); G06N 20/00 (2019.01); G06Q 10/087 (2013.01); G06Q 10/10 (2013.01); G06Q 30/0185 (2013.01); G06Q 30/0204 (2013.01)] | 20 Claims |
|
1. A method for predicting and providing, in real time, a certain similar item for a certain item in a certain transaction to a particular transaction system that is processing the certain transaction, comprising:
mapping, by a processor, barcoded item codes for multiple retailers into a culture-specific vector space based on transactions processed by multiple transaction systems and associated with the multiple retailers within a given geographical region, wherein the culture-specific vector space represented in a multidimensional space for item purchases processed by the multiple transaction system within the given geographical region, each item corresponds to a unique barcoded item code and each item represented and plotted as an item vector within the culture-specific vector space to defining contexts of purchases made for the corresponding item based on transaction data provided by the multiple transaction systems, wherein cultures associated with the culture-specific vector space are groupings or segments of customers selected based on customer geographic constraints, customer income constraints, or customer nationality constraints, wherein the culture-specific vector space includes aggregated transaction data for the customers that spans the multiple transaction systems for a given culture, wherein the mapping improves accuracy and efficiency of product categorization;
mapping, by the processor, retailer-specific item codes for a given retailer into a given retailer-specific vector space by defining the given retailer-specific vector space within the culture-specific vector space for the item purchases of the given retailer;
identifying, by the processor, a non-barcoded item code from the retailer-specific item codes mapped within the given retailer-specific vector space using a unique identification process that improves the reliability of matching products in a retail environment;
determining, by the processor, a select retailer-specific item code that is most similar to the non-barcoded item code within the given retailer-specific vector space based on distances associated with probabilities between plotted retailer-specific items within the given retailer-specific vector space, wherein determining further includes applying a probabilistic model that enhances predictions of consumer behavior;
linking, by the processor, the non-barcoded item code to the select retailer-specific item code, wherein the select retailer-specific item code is a particular barcoded item code that is mapped in the culture-specific vector space;
clustering, by the processor, the barcoded item codes within the culture-specific vector space into product categories using a clustering algorithm that operates in the defined multidimensional space; and
providing, by the processor, the non-barcoded item code to a certain transaction system associated with the given retailer based on the select retailer-specific item code being provided by the certain transaction system during a given transaction with the certain transaction system, wherein the providing further includes communicating the non-barcoded item code to the certain transaction system to enhance real-time transaction communication during the given transaction.
|