# US 2022/0036414 A1 · Product description-based line item matching

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0036414 A1
- Document ID: US20220036414A1
- Status: Published application
- Kind: A1
- Application number: 16/943,457
- Filed: 2020-07-30
- Published: 2022-02-03
- Earliest claimed: 2020-07-30
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Sampath Dechu; Neelamadhav Gantayat; Harshit Rawat; Sivakumar Narayanan
- Web page: https://patentgrep.com/patent/US20220036414A1

## CPC

- G06Q 30/04: Billing or invoicing
- G06F 40/247: Thesauruses; Synonyms
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/30: Semantic analysis
- G06Q 30/0635: {replenishment orders; recurring orders}
- G06V 30/412: Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Methods, systems, and computer program products for product description-based line item matching are provided herein. A computer-implemented method includes obtaining a digital invoice from a vendor comprising one or more text descriptions; retrieving a purchase order corresponding to the digital invoice from a purchase order database, wherein the purchase order comprises one or more line items; applying an out-of-vendor vocabulary model to the one or more text descriptions, wherein the out-of-vendor vocabulary model is trained to remove irrelevant text from the text descriptions based at least in part on historical purchase orders stored in the purchase order database corresponding to the vendor; and matching the one or more text descriptions from the digital invoice to the one or more line items in the purchase order based at least in part on output of the out-of-vendor vocabulary model.

## Claims

### Claim 1

A computer-implemented method, comprising:
obtaining a digital invoice from a vendor comprising one or more text descriptions;
retrieving a purchase order corresponding to said digital invoice from a purchase order database, wherein said purchase order comprises one or more line items;
applying an out-of-vendor vocabulary model to said one or more text descriptions, wherein said out-of-vendor vocabulary model is trained to remove irrelevant text from said text descriptions based at least in part on historical purchase orders stored in said purchase order database corresponding to said vendor; and
matching said one or more text descriptions from said digital invoice to said one or more line items in said purchase order based at least in part on output of said out-of-vendor vocabulary model;
wherein the method is carried out by at least one computing device.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, wherein each of said one or more line items in said purchase order comprises a corresponding description.

### Claim 3 (depends on 2)

The computer-implemented method of claim 2, wherein said matching comprises:
generating a context dependent discriminative word set for each of said one or more line items in said purchase order based on said corresponding descriptions.

### Claim 4 (depends on 3)

The computer-implemented method of claim 3, wherein said matching comprises:
matching said one or more text descriptions to said one or more line items based at least in part on said context dependent discriminative word sets.

### Claim 5 (depends on 4)

The computer-implemented method of claim 4, comprising:
for a given one of the context dependent discriminative word sets, generating a set of scores based on semantic comparisons of said given context dependent discriminative word set and word sets identified from each of said text descriptions of said digital invoice.

### Claim 6 (depends on 3)

The computer-implemented method of claim 3, wherein generating the context dependent discriminative word sets comprises:
tokenizing the descriptions corresponding to the one or more line items into respective token sets; and
determining a maximal co-occurrence word subset which belongs to one and only one of the token sets.

### Claim 7 (depends on 1)

The computer-implemented method of claim 1, comprising:
extracting a purchase order number from said digital invoice, wherein the purchase order is retrieved from the purchase order database based on the extracted purchase order number.

### Claim 8

A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computing device to cause the computing device to:
obtain a digital invoice from a vendor comprising one or more text descriptions;
retrieve a purchase order corresponding to said digital invoice from a purchase order database, wherein said purchase order comprises one or more line items;
apply an out-of-vendor vocabulary model to said one or more text descriptions, wherein said out-of-vendor vocabulary model is trained to remove irrelevant text from said text descriptions based at least in part on historical purchase orders stored in said purchase order database corresponding to said vendor; and
match said one or more text descriptions from said digital invoice to said one or more line items in said purchase order based at least in part on output of said out-of-vendor vocabulary model.

### Claim 9 (depends on 8)

The computer program product of claim 8, wherein each of said one or more line items in said purchase order comprises a corresponding description.

### Claim 10 (depends on 9)

The computer program product of claim 9, wherein said matching comprises:
generating a context dependent discriminative word set for each of said one or more line items in said purchase order based on said corresponding descriptions.

### Claim 11 (depends on 10)

The computer program product of claim 10, wherein said matching comprises:
matching said one or more text descriptions to said one or more line items based at least in part on said context dependent discriminative word sets.

### Claim 12 (depends on 11)

The computer program product of claim 11, wherein the program instructions executable by a computing device further cause the computing device to:
for a given one of the context dependent discriminative word sets, generate a set of scores based on semantic comparisons of said given context dependent discriminative word set and word sets identified from each of said text descriptions of said digital invoice.

### Claim 13 (depends on 10)

The computer program product of claim 10, wherein generating the context dependent discriminative word sets comprises:
tokenizing the descriptions corresponding to the one or more line items into respective token sets; and
determining a maximal co-occurrence word subset which belongs to one and only one of the token sets.

### Claim 14 (depends on 8)

The computer program product of claim 8, wherein the program instructions executable by a computing device further cause the computing device to:
extract a purchase order number from said digital invoice, wherein the purchase order is retrieved from the purchase order database based on the extracted purchase order number.

### Claim 15

A system comprising:
a memory; and
at least one processor operably coupled to the memory and configured for:
obtaining a digital invoice from a vendor comprising one or more text descriptions;
retrieving a purchase order corresponding to said digital invoice from a purchase order database, wherein said purchase order comprises one or more line items;
applying an out-of-vendor vocabulary model to said one or more text descriptions, wherein said out-of-vendor vocabulary model is trained to remove irrelevant text from said text descriptions based at least in part on historical purchase orders stored in said purchase order database corresponding to said vendor; and
matching said one or more text descriptions from said digital invoice to said one or more line items in said purchase order based at least in part on output of said out-of-vendor vocabulary model.

### Claim 16 (depends on 15)

The system of claim 15, wherein each of said one or more line items in said purchase order comprises a corresponding description.

### Claim 17 (depends on 16)

The system of claim 16, wherein said matching comprises:
generating a context dependent discriminative word set for each of said one or more line items in said purchase order based on said corresponding descriptions.

### Claim 18 (depends on 17)

The system of claim 17, wherein said matching comprises:
matching said one or more text descriptions to said one or more line items based at least in part on said context dependent discriminative word sets.

### Claim 19 (depends on 17)

The system of claim 17, wherein generating the context dependent discriminative word sets comprises:
tokenizing the descriptions corresponding to the one or more line items into respective token sets; and
determining a maximal co-occurrence word subset which belongs to one and only one of the token sets.

### Claim 20 (depends on 15)

The system of claim 15, wherein the at least one processor is configured for:
extracting a purchase order number from said digital invoice, wherein the purchase order is retrieved from the purchase order database based on the extracted purchase order number.
