Purchase Transaction Data Retrieval System With Unobtrusive Side Channel Data Recovery

Abstract

A server system automatically extracts sets of purchase transaction related data values for target purchase-related field types from purchase transaction records. The server system automatically flags respective ones of the sets of purchase transaction related data values with one or more respective incomplete purchase transaction related data values for one or more of the target purchase-related field types, and stores the flagged sets of purchase transaction related data values and ones of the other sets of purchase transaction related data values determined to be complete in a data storage system. For each of respective ones of the flagged sets of purchase transaction related data values, the server system unobtrusively recovers missing purchase transaction related information from one or more side channels based on one or more query terms and one or more query result selection criteria determined at least in part from the extracted purchase transaction related data values.

1 . A method implemented by one or more computing devices specifically programmed to perform operations comprising:
 for each of one or more sets of text strings corresponding to respective target purchase-related field types comprising a product description field type and one or more price field types and extracted from respective purchase transactions and comprising a respective partial description text string for a first one of the products and one or more total price text strings for the respective purchase transactions but missing an individual price text string for the first product, by one or more of the computing devices,
 building a respective query as a function of the partial description text string for the first product, 
 determining respective upper and lower price bounds for the first product as a function of one or more of the total price text strings for the respective purchase transaction, 
 sending at least one request comprising the query to at least one network address that triggers at least one server network node to execute a search engine to return at least one electronic document comprising a respective dynamically generated ranked list of product-related items matching the query and comprising respective sets of descriptions and individual prices for respective products, 
 selecting a product-related item in the ranked list of product-related items by evaluating one or more of the individual product prices in one or more of the product-related items against the respective upper and lower price bounds for the first product and one or more heuristics that preferentially select higher ranked product-related items over lower ranked product-related items, 
 excerpting a respective complete product description text string and a respective individual product price text string from the selected product-related item, and 
 augmenting the extracted set of structured text strings for the respective purchase transaction with the excerpted complete product description text string and the respective individual product price text string. 
   
     
     
         2 . The method of  claim 1 , wherein the product-related items in the list are derived from records of purchase transactions other than the respective purchase transaction. 
     
     
         3 . The method of  claim 1 , wherein the product-related items in the list are derived from records of products offered for purchase. 
     
     
         4 . The method of  claim 1 , wherein the one or more sets of text strings comprise an extracted a subtotal price text string for a subtotal price field type and an extracted order total price text string for an order total price field type; and the determining comprises, by one or more of the computing devices, automatically determining the upper and lower price bounds for selecting the product-related item in the ranked list based on the subtotal price text string and the order total price text string. 
     
     
         5 . The method of  claim 4 , wherein the one or more sets of text strings comprise one or more extracted item quantity text strings; and the determining comprises, by one or more of the computing devices, automatically determining the upper and lower price bounds for the first product based on the one or more item quantity text strings extracted for the respective purchase transaction. 
     
     
         6 . The method of  claim 5 , wherein the selecting comprises: in response to a determination that none of the product-related items in the ranked list has a price lower than the upper price bound, selecting a highest ranked one of the product-related items in the ranked list. 
     
     
         7 . The method of  claim 5 , wherein the selecting comprises: in response to a determination that the order total price text string is zero, selecting a highest ranked one of the product-related items in the ranked list. 
     
     
         8 . The method of  claim 1 , wherein the product related items in the dynamically generated ranked list are unconnected to the respective purchase transaction and are derived from records of purchase transactions ( 31 ,  33 ) other than the respective purchase transaction and that comprise complete ones of the sets of structured text strings extracted for purchase transactions other than the respective purchase transaction. 
     
     
         9 . The method of  claim 1 , wherein the selecting is conditioned on satisfaction of a predetermined timeliness requirement that the selected product-related item is associated with a date that is within a specified timeframe of a date associated with the respective purchase transaction. 
     
     
         10 . The method of  claim 1 , wherein the sending comprises applying the query to records of respective ones of the extracted sets of structured text strings determined to be complete, and the parsing comprises extracting the respective sets of structured text strings from purchase transaction records in respective electronic messages transmitted between network nodes ( 28 ), and the sending comprises applying the query to records of complete ones of the sets of extracted structured text strings. 
     
     
         11 . The method of  claim 1 , wherein:
 the one or more sets of text strings comprise an extracted order subtotal price text string for an order subtotal price field type and one or more extracted item quantity text strings; and   the determining comprises deriving the lower price bound for the first product by scaling the order subtotal price text string by a factor dependent on a linear function of the one or more item quantity data values extracted from the respective purchase transaction.   
     
     
         12 . The method of  claim 1 , wherein:
 the one or more sets of text strings comprise an extracted order total price text string for an order total price field type, an extracted order subtotal price text string for an order subtotal price field type, and one or more extracted item quantity text strings; and   in response to a determination that the order subtotal price text string is zero, the determining comprises deriving the lower price bound for the first product by scaling the order total price text string by a factor dependent on a linear function of the one or more item quantity data values extracted from the respective purchase transaction.   
     
     
         13 . The method of  claim 1 , wherein the dynamically generated list of product related items is ranked by product popularity. 
     
     
         14 . The method of  claim 1 , wherein the product related items in the dynamically generated list are associated with respective dates and are ranked by degree of closeness of their respective dates to an order date associated with the respective the purchase transaction. 
     
     
         15 . The method of  claim 1 , wherein the selecting comprises selecting a highest ranked one of the product-related items in the ranked list that satisfies the respective upper and lower price bounds for the first product. 
     
     
         16 . The method of  claim 1 , wherein the records of purchase transactions other than the respective purchase transaction comprise records of complete ones of the sets of structured text strings extracted for purchase transactions other than the respective purchase transaction. 
     
     
         17 . The method of  claim 1 , wherein for a particular one of the confirmation messages associated with a particular merchant, the sending comprises sending a request comprising the query to at least one network address that triggers at least one server network node to query records of products offered for purchase by the particular merchant. 
     
     
         18 . The method of  claim 15 , wherein the sending comprises automatically, by one or more of the computing devices, entering the query into a graphical control element of a web page generated by a web server associated with the particular merchant, and the at least one electronic document comprising the respective dynamically generated ranked list of product-related items is served in a web page by the web server associated with the particular merchant. 
     
     
         19 . The method of  claim 1 , wherein the sending comprises applying the query to records of respective ones of the extracted sets of structured text strings determined to be complete. 
     
     
         20 . The method of  claim 1 , wherein the product related items in the dynamically generated list are associated with respective dates and are ranked by degree of closeness of their respective dates to an order date associated with the respective the purchase transaction. 
     
     
         21 . Apparatus comprising a memory storing processor-readable instructions, and a processor coupled to the memory, operable to execute the instructions, and based at least in part on the execution of the instructions operable to perform operations comprising:
 for each of one or more sets of text strings corresponding to respective target purchase-related field types comprising a product description field type and one or more price field types, extracted from respective purchase transactions comprising a respective partial description text string for a first one of the products and one or more total price text strings for the respective purchase transactions but missing an individual price text string for the first product, by one or more of the computing devices,
 building a respective query as a function of the partial description text string for the first product, 
 determining respective upper and lower price bounds for the first product as a function of one or more of the total price text strings for the respective purchase transaction, 
 sending at least one request comprising the query to at least one network address that triggers at least one server network node to execute a search engine to return at least one electronic document comprising a respective dynamically generated ranked list of product-related items matching the query and comprising respective sets of descriptions and individual prices for respective products, 
 selecting a product-related item in the ranked list of product-related items by evaluating one or more of the individual product prices in one or more of the product-related items against the respective upper and lower price bounds for the first product and one or more heuristics that preferentially select higher ranked product-related items over lower ranked product-related items, 
 excerpting a respective complete product description text string and a respective individual product price text string from the selected product-related item, and 
 augmenting the extracted set of structured text strings for the respective purchase transaction with the excerpted complete product description text string and the respective individual product price text string. 
   
     
     
         22 . A computer-readable data storage apparatus comprising a memory component storing executable instructions that are operable to be executed by a computer, wherein the memory component comprises:
 one or more sets of text strings corresponding to respective target purchase-related field types comprising a product description field type and one or more price field types, extracted from respective purchase transactions comprising a respective partial description text string for a first one of the products and one or more total price text strings for the respective purchase transaction but missing an individual price text string for the first product;   executable instructions to build a respective query as a function of the partial description text string for the first product;   executable instructions to determine respective upper and lower price bounds for the first product as a function of one or more of the total price text strings for the respective purchase transaction;   executable instructions to send at least one request comprising the query to at least one network address that triggers at least one server network node to execute a search engine to return at least one electronic document comprising a respective dynamically generated ranked list of product-related items matching the query and comprising respective sets of descriptions and individual prices for respective products;   executable instructions to select a product-related item in the ranked list of product-related items by evaluating one or more of the individual product prices in one or more of the product-related items against the respective upper and lower price bounds for the first product and one or more heuristics that preferentially select higher ranked product-related items over lower ranked product-related items;
 executable instructions to excerpt a respective complete product description text string and a respective individual product price text string from the selected product-related item; and 
 executable instructions to augment the extracted set of structured text strings for the respective purchase transaction with the excerpted complete product description text string and the respective individual product price text string.