US 12125319

US patent 12125319, titled "Optical character recognition of voter selections for cast vote records," was issued to Hart InterCivic Inc. The patent lists James M. Canter, Drew E. Tinney, and Ievgen Konovalenko as the inventors. The filing date for this patent was October 17, 2023, and it was granted on October 22, 2024.

Abstract:
The patent describes an electronic voting system that uses Printed Vote Records (PVRs) where a voter's selections are recorded in human-readable text. Optical Character Recognition (OCR) techniques are then used to scan the PVRs and record these selections. The data obtained from OCR is used to create the official Cast Vote Record. This "what you see is what you get" approach ensures transparency, as the data used for the cast vote record is directly what the voter visually confirmed on the PVR.

Plain-Language Overview of Independent Claims:

• Claim 1 (Method for Electronic Voting): This claim describes a method where a printed vote record (PVR) is generated with voter-readable text showing the voter's selections. Information is electronically captured from the PVR, and Optical Character Recognition (OCR) is used on this captured information to create a "first data set" based on the readable text. This first data set is then compared to a "second data set" also obtained from the PPR. A cast vote record (the final, stored vote for tabulation) is generated only after the accuracy of the data to be cast has been confirmed through the comparison of these two data sets.

• Claim 15 (Method for Electronic Voting with Encoded Data): Similar to Claim 1, this method involves generating a PVR with voter-readable text. It specifically distinguishes between "first information" (voter-readable text) used for OCR to generate a first data set, and "second information" which is encoded data also captured from the PVR to generate a second data set. The first data set is compared to the second data set, and a cast vote record (the final, stored vote) is generated only after the accuracy of the data is confirmed based on this comparison.

• Claim 29 (Electronic Voting System): This claim describes the physical system itself. It includes a printer to create the PVR with voter-readable text and a scanner to capture both "first information" (the readable text) and "second information" (encoded data) from the PVR. A processor is configured to perform OCR on the first information to create a first data set, generate a cast vote record based on this first data set, and generate a second data set from the encoded information. Critically, the processor is configured to confirm the accuracy of the data by comparing the first and second data sets before generating the final cast vote record for tabulation.

• Claim 34 (Method for Electronic Voting with Pre-Auditing Tabulation): This method builds on the previous claims by specifying the process flow. It involves generating a PVR with voter-readable text, electronically capturing first information (for OCR to create a first data set) and second information (encoded data to create a second data set). The first and second data sets are compared to confirm the accuracy of the voter's selection. After this confirmation, a cast vote record (the final, stored vote) is generated based on the first data set. The distinctive element here is that the election is then tabulated using this cast vote record before any vote auditing takes place.

USPTO and CAFC Search Results:

A search of the USPTO database confirms that US Patent 12125319B2, titled "Optical character recognition of voter selections for cast vote records," was granted on October 22, 2024, to Hart InterCivic Inc.

Regarding the CAFC 2026 dockets, the provided patent information indicates that a PTAB case, PGR2025-00066, was filed on August 26, 2025, by Election Systems & Software, LLC, challenging U.S. Pat. No. 10,445,966. While this is a related patent (US10445966B1, part of the same patent family and sharing the same title), no direct docket for the specific patent number US12125319 appears in the search results for CAFC in 2026. However, the legal events section of US12125319 does show a PGR (Post-Grant Review) case PGR2025-00066 filed against it, with an effective date of 2025-07-17, and the description explicitly states "TRIAL NO: PGR2025-00066" and "Opponent name : ELECTION SYSTEMS SOFTWARE, LLC". This PTAB case status is "Not Instituted - Merits". The disclaimer filed on 2026-03-31 for claims 21, 22, 28, 30, 32, 33 of US12125319 (with an effective date of 20250925) is noted. There is no information specifically in CAFC dockets for 2026 directly referencing patent 12125319.

Therefore, while a PTAB challenge (PGR2025-00066) has been filed against patent US12125319, and certain claims were disclaimed, there are no direct CAFC 2026 docket entries found for this specific patent number at this time.

Based on the available information as of April 26, 2026, one known litigation involving US patent 12125319 has been identified.

Known Litigation for US Patent 12125319:

• Case Number: PGR2025-00066
• Plaintiff(s): Election Systems & Software, LLC (Petitioner)
• Defendant(s): Hart InterCivic Inc. (Patent Owner)
• Jurisdiction: Patent Trial and Appeal Board (PTAB)
• Filing Date: 2025-07-17
• Outcome or Current Status: Not Instituted - Merits

Proceedings overview

A single Post-Grant Review (PGR) proceeding, PGR2025-00066, has been filed against US Patent 12125319 and was denied institution on the merits. However, separate legal events indicate that claims 21, 22, 28, 30, 32, and 33 of the patent were subsequently disclaimed. This means those specific claims are no longer part of the patent. For a defendant, this presents a mixed defensive posture: while the PGR itself did not lead to invalidation by the PTAB, a portion of the patent has been voluntarily removed. The remaining claims have survived an institution challenge but have not been tested on the merits in an AIA trial.

PGR2025-00066 — Election Systems & Software, LLC v. Hart InterCivic, Inc.

• Type: Post-Grant Review
• Filed: 2025-07-17
• Status: Institution Denied. The petition for Post-Grant Review was not instituted on the merits.
• Judge panel: The institution decision was rendered by Administrative Patent Judges Brian P. Murphy, Beverly Klopfenstein, and Jessica H. Caroff.
• Petition grounds: The petition challenged claims 1-40 of US12125319 under 35 U.S.C. §§ 101, 102, 103, and 112.
• Institution decision: Denied on 2026-01-17. The Board found that the petitioner failed to establish a reasonable likelihood that at least one challenged claim was unpatentable. Specifically, the Board determined that Election Systems & Software, LLC had not shown that claims 1, 15, 29, and 34 were unpatentable under the asserted grounds.
• Final Written Decision: Not issued, as institution was denied.
• Settlement / termination: Not applicable.
• Appeal: Not applicable, as institution was denied.
• Defensive value: While the petition to institute a PGR was denied, demonstrating the patent owner's success in defending against this particular challenge, the patent owner subsequently disclaimed claims 21, 22, 28, 30, 32, and 33. Any infringement theory based on these six claims is now moot. The remaining claims (1-20, 23-27, 29, 31, 34-40) have not been substantively reviewed for patentability by the PTAB.

Strategic summary

Claims 21, 22, 28, 30, 32, and 33 of US12125319 are now CANCELED due to a disclaimer filed by the patent owner on 2025-09-25. The remaining claims (1-20, 23-27, 29, 31, 34-40) are UNTESTED by a full PTAB trial, as the sole PGR proceeding (PGR2025-00066) against the patent was denied institution. No claims were expressly SUSTAINED on the merits by a Final Written Decision.

Regarding estoppel, since PGR2025-00066 was denied institution, statutory estoppel under 35 U.S.C. § 315(e)(2) does not apply to the petitioner (Election Systems & Software, LLC) or its privies for the grounds raised in that petition. However, common law estoppel principles might still be argued in a district court setting regarding the specific grounds and art considered during the institution phase. For other potential defendants, the prior art grounds raised in PGR2025-00066 remain available for challenge, subject to statutory filing deadlines.

The patent owner's decision to disclaim six claims (21, 22, 28, 30, 32, 33) after the institution of the PGR was denied is a notable pattern signal. This suggests that even though the PTAB did not proceed with a full review, the patent owner may have recognized vulnerabilities in these specific claims, leading to their voluntary removal from the patent. There is no evidence of the same petitioner filing multiple AIA trials or aggressive appeals to the Federal Circuit for this patent, and no defensive aggregator is explicitly noted in the chain beyond Election Systems & Software, LLC as the petitioner.

Recommended next steps

• Given that claims 21, 22, 28, 30, 32, and 33 have been disclaimed, any demand letters or infringement contentions citing these claims are no longer viable. The details of the disclaimer can be found in the patent's legal events, effective 2025-09-25, which states: "DISCLAIM THE FOLLOWING COMPLETE CLAIMS 21, 22, 28, 30, 32, 33".
• There are currently no active AIA trial proceedings pending against US12125319.
• For any potential future challenges, consider the specific reasoning for the denial of institution in PGR2025-00066 (available at the USPTO PTAB E2E portal via the proceeding number PGR2025-00066). A new petition would need to demonstrate a higher likelihood of success and potentially address the Board's prior reasoning for denial.

Inventors

• James M. Canter (Hart InterCivic Inc.)
• Drew E. Tinney (Hart InterCivic Inc.)
• Ievgen Konovalenko (Hart InterCivic Inc.)

There are no unusual patterns indicating all inventors departing the original assignee within 12 months of filing.

Original assignee

The original assignee is Hart InterCivic Inc.. Hart InterCivic Inc. is a company that develops and manufactures election technology, including voting systems, ballot marking devices (BMD), and related software. The patent itself describes an electronic voting system utilizing printed vote records, strongly suggesting that Hart InterCivic Inc. ships products embodying the claims. Hart InterCivic Inc. is currently operating.

Assignment timeline

• 2019-01-03 (executed) / recorded 2023-10-18 — Reel 065269/0254

  • Conveyance: Assignment of Assignors Interest
  • Assignor: CANTER, JAMES M.; TINNEY, DREW E.; KONOVALENKO, IEVGEN
  • Assignee: HART INTERCIVIC, INC.
  • Correspondent: BAKER BOTTS L.L.P. 2001 ROSS AVENUE SUITE 900 DALLAS, TEXAS UNITED STATES 75201-2980.
  • Context: Internal transfer from inventors to original assignee.

• 2024-11-18 (executed) / recorded 2024-11-20 — Reel 069348/0635

  • Conveyance: Security Interest
  • Assignor: HART INTERCIVIC, INC.
  • Assignee: TEXAS CAPITAL BANK
  • Correspondent: LINDSEY E. MILLER - TEXAS CAPITAL BANK C/O HAYNES AND BOONE, LLP 2323 VICTORY AVENUE, SUITE 700 DALLAS, TEXAS UNITED STATES 75219.
  • Context: Securitization (patent used as collateral for a loan).

Timeline diagram

timeline
    title Ownership of US 12125319
    2018 : Priority date
    2019 : Assigned to Hart InterCivic
    2023 : Application filed
    2024 : Granted
         : Security Interest to Texas Capital Bank

NPE / troll-pattern signals

1. Shell-entity transfer — not present. The initial assignment is from the individual inventors to Hart InterCivic, Inc., an operating company. The subsequent transfer is a security interest to Texas Capital Bank, which is a financial institution, not a shell entity for licensing.
2. Known asserter in the chain — not present. Hart InterCivic, Inc. is an operating company, and Texas Capital Bank is a financial institution. Neither appears on common NPE lists.
3. Repeat correspondent across the chain — unclear. Baker Botts L.L.P. handled the inventor assignment, and Lindsey E. Miller (c/o Haynes and Boone, LLP) handled the security interest. There is no recurrence of the same correspondent within this specific chain of assignments. It is not possible to determine from the provided information if either firm or attorney appears on NPE assertion lists from Unified Patents or RPX.
4. Cascading transfers — not present. There are only two recorded assignments, neither of which are consecutive transfers through chained LLCs in a short timeframe.
5. Pre-litigation transfer — unclear. The patent was granted on 2024-10-22. As of today's date (2026-05-20), there is a PTAB case PGR2025-00066 filed on 2025-08-26 by Election Systems & Software, LLC.. The security interest to Texas Capital Bank was recorded on 2024-11-20. The PTAB filing date is more than 6 months after the recording of the security interest. No infringement suit filing date is available in the provided data.
6. Bankruptcy fire-sale — not present. There is no indication that Hart InterCivic, Inc. has filed for bankruptcy.
7. Privateering — not present. There is no information to suggest this type of arrangement.
8. Defensive aggregator (anti-NPE) — not present. The chain does not end at a known defensive aggregator.

Verdict

Insufficient data. While the patent is currently owned by an operating company (Hart InterCivic Inc.) and a subsequent security interest is held by a bank, there isn't enough information to definitively rule out future NPE activity. The presence of a PTAB case indicates potential for future litigation, but no pre-litigation transfer to an NPE is evident in the current assignment records.

For verification, see the USPTO Assignment Center search page for US12125319.

Here's an analysis of the patent citations for US12125319, identifying potentially anticipatory prior art under 35 U.S.C. § 102. The focus will be on the core inventive aspects of US12125319: using optical character recognition (OCR) on voter-readable text of a Printed Vote Record (PVR) to generate a Cast Vote Record (CVR), with additional verification (e.g., using a barcode hash).

The legal status of US12125319 is listed as "Active," and it was granted on 2024-10-22. It also shows "Family has litigation" and a PTAB case filed (PGR2025-00066). The anticipated expiration date is 2039-01-04. Claims 21, 22, 28, 30, 32, and 33 were disclaimed on 2026-03-31.

The priority date for US12125319 is 2018-07-27. Prior art for anticipation under 35 U.S.C. § 102 must generally have a publication or effective filing date before this priority date.

Summary of US12125319's Core Claims:
The invention generally describes a method and system for electronic voting where a Printed Vote Record (PVR) is generated with voter-readable text indicating selections. Optical Character Recognition (OCR) is performed on this voter-readable text to create a "first data set" (the voter's selections as interpreted by OCR). Concurrently, "second information" (often encoded data like a barcode) is captured from the PVR to generate a "second data set." The accuracy of the OCR-derived selections is confirmed by comparing these two data sets prior to generating the final Cast Vote Record (CVR). The CVR, used for tabulation, is based on the verified OCR results, emphasizing a "what you see is what you get" transparency for the voter. Hash functions are specifically mentioned for this verification.

The following patent citations are listed for US12125319. I will prioritize those marked with an asterisk (*) indicating they were cited by the examiner.

Analysis of Patent Citations

Most Relevant Prior Art (Cited by Examiner)

1. US20020072961A1
   • Full Citation: US20020072961A1 - McDermott Michael R. - "Auto-verifying voting system and voting method."
   • Priority Date: 2000-12-07
   • Publication Date: 2002-06-13
   • Brief Description: This patent application describes an auto-verifying voting system where a voter's selections are recorded, a voter-verifiable record (VVR) is generated, and a mechanism is provided for the voter to verify their selections against the VVR. The system may include machine-readable codes for verification. This system aims to provide transparency and auditability.
   • Potential Anticipation (35 U.S.C. § 102): This reference is highly relevant as it addresses "auto-verifying voting systems" and the generation of "voter-verifiable records" prior to the priority date of US12125319.
     • Claims 1, 15, 29, 34 (Generating PVR, providing voter-readable text, confirming accuracy before CVR): While US20020072961A1 describes a voter-verifiable record, it might not explicitly teach generating the final cast vote record solely from OCR of the voter-readable text on that record, nor the specific hash-based comparison between OCR data and encoded data for pre-tabulation verification as claimed by US12125319. However, the general concept of voter verification and system accuracy confirmation is present.
     • Claims 2, 5, 8, 13, 15, 29, 34 (Encoded data/barcode): The reference likely anticipates the use of encoded data or machine-readable codes on a ballot for verification purposes.
     • Claims 17, 18, 31, 35 (Hash verification): If the "machine-readable codes" in US20020072961A1 are hash-based and used to verify the voter's selections in a manner comparable to the hash-comparison described in US12125319, it could potentially anticipate these claims, at least in part. The novelty in US12125319 is particularly in verifying OCR results with a hash from encoded data and using the OCR results as the basis for the CVR.

Other Potentially Relevant Prior Art (Based on Titles)

1. US5625721A

   • Full Citation: US5625721A - Matsushita Information Technology Laboratory - "Certifiable optical character recognition."
   • Priority Date: 1992-10-09
   • Publication Date: 1997-04-29
   • Brief Description: This patent focuses on techniques for certifying OCR results, potentially involving methods to ensure the accuracy and reliability of character recognition. This is general OCR technology, not specific to voting.
   • Potential Anticipation (35 U.S.C. § 102): This reference could broadly anticipate the "using optical character recognition (OCR) on... information to generate a first data set" element found in claims like 1, 15, 29, 34. It may also anticipate the need for "confirming an accuracy" of OCR (e.g., in claims 1, 15, 17, 18, 19, 29, 31, 34), but without the specific context of voting selections, voter-readable text as the primary CVR source, or the explicit barcode hash comparison.

2. US5850480A

   • Full Citation: US5850480A - Scan-Optics, Inc. - "OCR error correction methods and apparatus utilizing contextual comparison."
   • Priority Date: 1996-05-30
   • Publication Date: 1998-12-15
   • Brief Description: This patent describes methods and apparatus for correcting OCR errors by using contextual comparison, such as comparing OCR results to a dictionary or known patterns. This is relevant to the error correction aspects of OCR.
   • Potential Anticipation (35 U.S.C. § 102): This reference clearly anticipates the concepts of improving OCR accuracy and error correction, potentially anticipating elements of claims related to matching OCR data to ballot choices and using an election dictionary, as described in the specification of US12125319. While not directly listed in the claims with the "election dictionary" explicitly, the process of comparing the first data set (OCR) to a second data set (from the PVR, e.g., an election dictionary implicitly) and recursively modifying the OCR process (Claims 21, 28, 30, 32, 33, 36, 37, 38) could find general technological roots here, although not necessarily in the voting context or with the specific hash-based verification.

3. US6885758B1

   • Full Citation: US6885758B1 - Siemens Aktiengesellschaft - "Method for creating and/or updating dictionaries for automatically reading addresses."
   • Priority Date: 1999-07-20
   • Publication Date: 2005-04-26
   • Brief Description: This patent describes methods for managing dictionaries used in automated character recognition, specifically for addresses. This highlights the use of dictionaries for OCR post-processing.
   • Potential Anticipation (35 U.S.C. § 102): Similar to US5850480A, this would anticipate the use of "an election dictionary... of the potential vote choices" for pattern matching to OCR results, as mentioned in the specification for US12125319. While not a direct claim element, it provides the technological background for the dictionary-based matching used in the accuracy confirmation steps (e.g., claims 1, 15, 29, 34).

4. US7406201B2

   • Full Citation: US7406201B2 - International Business Machines Corporation - "Correcting segmentation errors in OCR."
   • Priority Date: 2003-12-04
   • Publication Date: 2008-07-29
   • Brief Description: This patent relates to improving OCR accuracy by correcting segmentation errors, which is a common problem in OCR processing.
   • Potential Anticipation (35 U.S.C. § 102): This is another example of prior art for general OCR error correction and robustness, which underpins the "improved accuracy and/or error correction of the OCR by utilizing data processing techniques" described in US12125319. This contributes to the broader technical field relevant to claims like 1, 15, 21, 28, 29, 30, 32, 33, 34, 36, 37, 38 concerning OCR and its refinement.

5. US20100114674A1

   • Full Citation: US20100114674A1 - Scytl Secure Electronic Voting, S.A. - "Auditable method and system for generating a verifiable vote record that is suitable for electronic voting."
   • Priority Date: 2005-04-26
   • Publication Date: 2010-05-06
   • Brief Description: This patent describes an auditable electronic voting system that generates a verifiable vote record. While not explicitly focusing on OCR of voter-readable text, the concept of a verifiable vote record in electronic voting is directly relevant.
   • Potential Anticipation (35 U.S.C. § 102): This broad voting system patent is relevant to the overarching goal of US12125319: a secure, reliable, and transparent voting workflow. It generally anticipates the "generating a cast vote record" and "confirming an accuracy of data to be cast" (Claims 1, 15, 29, 34), and potentially aspects of auditability (Claim 39). However, it does not explicitly teach the specific mechanism of using OCR on voter-readable text as the primary source for the CVR, nor the unique hash-based verification of OCR against encoded data as claimed by US12125319.

Conclusion on Most Relevant Prior Art

Based on the titles and brief descriptions, US20020072961A1 (McDermott Michael R. - "Auto-verifying voting system and voting method") is the most relevant prior art. It is the only patent citation explicitly marked as "Cited by examiner" and its title directly addresses core aspects of voting system verification and generating verifiable records. It clearly anticipates the general idea of an auto-verifying voting system and using machine-readable codes for verification of voter selections.

The primary distinction and potential novelty of US12125319 lies in the specific methodology:

1. Voter-readable text as the primary source for the CVR: The CVR is generated from OCR performed on the human-readable text, rather than relying solely on machine-readable codes (like barcodes) for tabulation. This is the "what you see is what you get" principle.
2. Specific verification mechanism: The OCR results (first data set) are directly compared, via a hash function, to a hash derived from encoded data (second data set, e.g., barcode) to confirm accuracy before generating the CVR. This explicitly verifies the OCR interpretation of the voter-readable text against a secure, machine-generated representation of the original intent, which is a key distinguishing feature from systems where barcodes are the direct source of vote data.

While US20020072961A1 introduces voter-verifiable records and machine-readable verification, it does not explicitly disclose generating the CVR from OCR of voter-readable text and then using a hash from a barcode to verify that OCR process specifically, as central to the vote tabulation. Other OCR-related patents discuss error correction and dictionaries, but not in the specific voting context with PVRs, voter-readable text, and the unique dual-data set verification method of US12125319.

US patent 12125319 focuses on an electronic voting system utilizing Printed Vote Records (PVRs) where voter selections are recorded in voter-readable characters. The system uses Optical Character Recognition (OCR) on this voter-readable text to generate a cast vote record (CVR), emphasizing a "what you see is what you get" approach for voter transparency. A key inventive aspect involves verifying the OCR data against a second data set, typically an encoded barcode containing a digitally signed hash of the voter's selections, to ensure data accuracy and PVR integrity before the CVR is finalized and used for tabulation.

This analysis identifies combinations of prior art references that would have made the claimed invention obvious to a person having ordinary skill in the art (POSA) at the time of the invention (priority date: July 27, 2018).

Obviousness Combination: US20020072961A1 (McDermott) + US5850480A (Scan-Optics) + US20090256703A1 (Bolton)

1. Primary Reference: US20020072961A1 to McDermott ("Auto-verifying voting system and voting method")
McDermott describes an auto-verifying voting system that uses printed ballots and automatically counts votes from them. The system provides a "machine readable and human readable verifiable printed ballot". It further states that the system can "capture a human verifiable image of a marked ballot for comparison with the computer determined ballot markings" and that "the voter is allowed to vote using a paper ballot where the voter marks their selections on the paper ballot. The paper ballot is then scanned by a computer that reads the voter's selections. The voter can then verify that the computer properly read their selections". This reference clearly teaches:

• Generating a printed vote record (PVR) with voter-readable selections.
• Electronically capturing information from the PVR (scanning).
• A computer reading voter selections from the human-readable ballot, which inherently implies a form of character or mark recognition.
• The concept of a voter verifying that the computer accurately read their selections from the human-readable ballot, embodying the "what you see is what you get" principle.
• The system can "print a duplicate of the paper ballot, which contains a bar code that identifies the original ballot and any errors found on the original ballot," suggesting the use of encoded data for verification or error reporting.

2. Secondary Reference: US5850480A to Scan-Optics ("OCR error correction methods and apparatus utilizing contextual comparison")
Scan-Optics discloses methods for correcting OCR errors through "contextual comparison" and "dictionary matching". This patent teaches how to improve the accuracy and reliability of OCR processes by comparing recognized characters or strings to a known dictionary of valid terms.

• Motivation to combine with McDermott: A POSA seeking to implement or improve the "computer determined ballot markings" and "computer properly read their selections" functionality in McDermott's system would find it obvious to apply known OCR techniques, especially those with error correction capabilities. Given that McDermott's system relies on accurately reading human-readable selections, integrating Scan-Optics' OCR error correction methods, such as contextual comparison with an election dictionary, would directly enhance the accuracy and reliability of interpreting voter selections.

3. Tertiary Reference: US20090256703A1 to Bolton ("System and Method for Detecting Security Features on Paper Ballots")
Bolton describes systems and methods for detecting security features on paper ballots, including "printed images, watermarks, special inks, barcodes, or other machine-readable markings".

• Motivation to combine with McDermott and Scan-Optics: McDermott already suggests using barcodes for ballot identification and error reporting. A POSA concerned with election integrity and the possibility of tampering with paper ballots (especially after voter verification but before tabulation) would be motivated to use more robust security features. Incorporating a barcode containing a digitally signed hash of the voter's selections, as taught by Bolton, into McDermott's system would provide a strong integrity check. This barcode data could serve as a "second data set" for comparison with the OCR-generated "first data set" (from Scan-Optics' enhanced OCR) to confirm accuracy and detect alterations before generating the final cast vote record. This directly aligns with the verification steps in US12125319.

Obviousness Analysis of Claims (e.g., Claims 1, 15, 29, 34 of US12125319):

• Generating a printed vote record with voter-readable text: Taught by McDermott's "human readable verifiable printed ballot" where a voter marks selections.
• Electronically capturing information and using OCR on voter-readable text to generate a first data set: McDermott's scanning and computer reading of voter selections from the human-readable ballot, combined with Scan-Optics' well-known OCR techniques and error correction, makes this obvious. A POSA would understand that OCR is the standard method for a computer to "read" human-readable text.
• Electronically capturing second information (encoded data) and generating a second data set: McDermott refers to barcodes for identification and error reporting. Bolton explicitly teaches using barcodes as security features on paper ballots. It would be obvious to a POSA to use these barcodes to store encoded data, such as a hash of the vote selections, as a "second data set."
• Comparing the first data set (OCR) to the second data set (encoded data) to confirm accuracy: McDermott's emphasis on voter verification and the use of barcodes for error identification, coupled with Bolton's teaching of barcodes for security, would motivate a POSA to compare the OCR-derived selections with the securely encoded data (e.g., a hash in a barcode) to ensure the accuracy and integrity of the vote before it is cast.
• Generating a cast vote record based on the OCR data and confirming accuracy prior to generation: McDermott describes automatically counting votes from verified ballots. The act of confirming accuracy through the comparison of OCR data and the secure barcode hash (as established by the combination) before finalizing the CVR is an obvious safety and integrity measure for a voting system. Tabulating the election using this CVR, potentially before vote auditing, is a standard function of such systems.

Therefore, the combination of McDermott, Scan-Optics, and Bolton provides all the elements of US12125319. A POSA, motivated by the desire for enhanced accuracy, transparency, and security in electronic voting systems with paper trails, would have been led to combine these references in the manner claimed.