`571-272-7822
`
`Paper 29
`Entered: September 1, 2021
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`ELECTION SYSTEMS & SOFTWARE, LLC,
`Petitioner,
`
`v.
`
`HART INTERCIVIC, INC.,
`Patent Owner.
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`
`
`
`
`
`
`
`
`Before CHARLES J. BOUDREAU, AMANDA F. WIEKER, and
`JASON M. REPKO, Administrative Patent Judges.
`
`REPKO, Administrative Patent Judge.
`
`JUDGMENT
`Final Written Decision
`Determining All Challenged Claims Unpatentable
`35 U.S.C. § 328(a)
`
`
`
`
`
`
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`I.
`
`INTRODUCTION
`
`Election Systems & Software, LLC (“Petitioner”) filed a petition to
`
`institute a post-grant review of claims 1–30 of U.S. Patent No. 10,445,966
`
`B1 (Ex. 1001, “the ’966 patent”). Paper 2 (“Pet.”). Hart InterCivic, Inc.
`
`(“Patent Owner”) did not file a Preliminary Response.
`
`On September 2, 2020, we instituted an inter partes review of all
`
`challenged claims based on all grounds in the Petition. Paper 6
`
`(“Inst. Dec.”). Patent Owner filed a Response. Paper 14 (“PO Resp.”).
`
`Petitioner filed a Reply. Paper 19 (“Pet. Reply”). Patent Owner filed a Sur-
`
`reply. Paper 20 (“Sur-reply”). An oral hearing was held on July 6, 2021. A
`
`transcript of that hearing has been entered into the record. Paper 28 (“Tr.”).
`
`We have jurisdiction under 35 U.S.C. § 6. This Final Written Decision
`
`is issued under 35 U.S.C. § 328(a). For the reasons that follow, Petitioner
`
`has shown by a preponderance of the evidence that claims 1–30 are
`
`unpatentable.
`
`A.
`
`Related Matters
`
`Neither party identifies any related matters. Pet. 1; Paper 4, 2
`
`(Mandatory Notice).
`
`B.
`
`The ’966 Patent
`
`The ’966 patent describes an electronic voting system that uses
`
`printed vote records (PVRs). Ex. 1001, 1:54–57. PVRs create an official
`
`record for later auditing. Id. at 1:29–31, 36. Electronic voting systems that
`
`create a paper record are called “hybrid voting” systems because they are a
`
`cross between hand-marked paper ballots and electronic systems.
`
`Id. at 1:31–34. In these hybrid systems, PVRs typically contain a bar code
`
`that encodes the voter’s selections. Id. at 1:40–42. The hybrid voting system
`
`then digitally scans the PVR’s bar code to create an electronic cast vote
`
`2
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`record. Id. at 1:42–43. But, according to the patent, voters find that this
`
`system lacks transparency because they cannot read the bar code to confirm
`
`that their votes match the printed record and their intended vote selections.
`
`Id. at 1:43–47.
`
`To improve transparency, the ’966 patent’s PVRs record the voter’s
`
`selections using voter-readable characters. See id. at 1:48–57. For example,
`
`the top of a PVR generated by a ballot-marking device is shown below. See
`
`id., Fig. 3.
`
`
`
`Figure 3 shows the PVR’s header region 310 with instructions to the
`voter, voter selection section 340, which is readable by the voter, and
`bar code 320. Id. at 6:25–40.
`
`The system scans the PVR using optical character recognition (OCR) to
`
`record the votes. Id. at 1:57–59. That is, the voter can read the same data
`
`from which the system generates the cast vote record. Id. at 1:60–65. This
`
`allows the voter to confirm that the PVR matches the selections entered into
`
`the ballot-marking device. Id. at 6:40–43.
`
`The system uses a bar code to detect counterfeit or altered PVRs.
`
`Id. at 2:31–32. The bar code includes a digitally signed hash. Id. at 2:33–36.
`
`During scanning, the system generates the OCR results and verifies them
`
`3
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`with another hash. Id. at 2:36–38. In particular, the system produces a hash
`
`from the OCR-generated characters and compares them to the hash provided
`
`in the bar code. Id. at 2:38–40. According to the patent, this technique
`
`confirms that a known source printed the PVR and detects post-printing
`
`alterations. Id. at 2:40–44.
`
`The system’s OCR process uses several techniques to improve
`
`recognition accuracy. See id. at 1:66–2:30. In one embodiment, the OCR
`
`process uses a dictionary of potential vote choices (such as candidate names)
`
`for pattern matching. Id. at 1:67–2:4. The OCR process may also use a
`
`Levenshtein distance algorithm for error detection. Id. at 2:20–22. This
`
`algorithm calculates a distance between the OCR data and each key in the
`
`dictionary. Id. at 2:22–24. If the distance is below a threshold, the data is
`
`matched to the dictionary entry. Id. at 2:24–26.
`
`C.
`
`Claims
`
`Claims 1, 6, 11, 15, 20, 23, 28, and 29 are independent. Claim 1 is
`
`reproduced below.
`
`1. A method for electronic voting using printed vote records,
`comprising:
`
`recording a voter’s vote selections;
`
`generating a printed vote record of the voter’s vote
`selections, the printed vote record containing voter
`readable text indicating the voter’s vote selections;
`
`scanning the printed vote record, the scanning including
`scanning the voter readable text;
`
`utilizing optical character recognition (OCR) on the
`scanned voter readable text to create a data set
`which identifies the voter’s vote selections; and
`
`4
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`generating a cast vote record from the data set so that voter
`selections in the cast vote record are based on the
`voter readable text,
`
`wherein identifying the voter’s vote selection includes
`comparing the data set to an election dictionary, the
`election dictionary containing potential vote
`choices,
`
`wherein the comparing the data set to the election
`dictionary is utilized to confirm an accuracy of the
`OCR to limit potential OCR errors.
`
`Ex. 1001, 8:43–62.
`
`D.
`
`Evidence
`
`Reference
`Name
`Brockhouse US 2014/0231513 A1, published Aug. 21,
`2014
`US 7,406,201 B2, issued July 29, 2008
`BR 10 2013 018558-2 A2, published Sept. 2,
`2014
`
`Heilper
`Nadaf
`
`US 9,401,059 B2, issued July 26, 2016
`Backert
`Herskowitz US 6,971,574 B1, issued Dec. 6, 2005
`
`
`Exhibit
`1004
`
`1005
`1006
`(original),
`1007
`(translation)
`1008
`1009
`
`Petitioner also relies on the declarations of Michael I. Shamos, Ph.D.,
`
`J.D. Ex. 1003 (“Shamos Decl.”); Ex. 1055 (“Shamos Suppl. Decl.”). Patent
`
`Owner relies on the declaration of Daniel P. Lopresti, Ph.D. Ex. 2006
`
`(“Lopresti Decl.”).
`
`
`
`
`
`5
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`E.
`
`Asserted Grounds
`
`Petitioner asserts that claims 1–30 are unpatentable on the following
`
`grounds. Pet. 3.
`
`Claim(s) Challenged
`1–30
`1–5, 15–19
`6–10, 23, 24, 29, 30
`6–10, 23, 24, 29, 30
`25
`25
`11–14, 28
`
`35 U.S.C. §
`101
`103
`102
`103
`103
`103
`103
`
`20–22
`
`4, 5, 26, 27
`
`103
`
`112(b)
`
`Reference(s)/Basis
`Eligibility
`Brockhouse, Heilper
`Nadaf
`Nadaf
`Nadaf, Heilper
`Nadaf
`Nadaf, Backert, Herskowitz
`Nadaf, Heilper, Backert,
`Herskowitz
`Indefiniteness
`
`II. ANALYSIS
`
`A.
`
`Eligibility for Post-Grant Review
`
`Post-grant review is available only for patents that, at one point,
`
`contained at least one claim with an effective filing date, as defined by
`
`35 U.S.C. § 100(i), on or after March 16, 2013. Petitioner asserts that the
`
`’966 patent is available for post-grant review. Pet. 2. We agree. The filing
`
`date for the ’966 patent is January 4, 2019, and the patent claims benefit
`
`only to a provisional application filed on July 27, 2018. Ex. 1001,
`
`codes (22), (60).
`
`B.
`
`Level of Ordinary Skill in the Art
`
`According to Petitioner,
`
`A person of ordinary skill in the art at the time the
`‘966 patent was filed (“POSITA”) would have had at least a
`Bachelor’s degree in electrical engineering or computer science
`or an equivalent field, or equivalent work experience and, in
`addition, at least two years’ experience with electronic voting
`machines, including familiarity with ballot marking devices
`
`6
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`(“BMDs”), optical character recognition, bar codes and hash
`functions.
`
`Pet. 24 (citing Shamos Decl. ¶¶ 50–51).
`
`In the Institution Decision, we applied Petitioner’s proposed
`
`definition. Inst. Dec. 6. Patent Owner does not dispute Petitioner’s proposed
`
`level of ordinary skill in the art. (PO Resp. 3). We continue to find that the
`
`level of ordinary skill identified by Petitioner (Pet. 24) is consistent with the
`
`record. See Inst. Dec. 6. Thus, we use the same definition here that we used
`
`in the Institution Decision.
`
`C.
`
`Claim Construction
`
`We need only construe terms that are in controversy. Nidec Motor
`
`Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed.
`
`Cir. 2017) (citing Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200 F.3d 795,
`
`803 (Fed. Cir. 1999)).
`
`We construed the term “cast vote record” in the Institution Decision.
`
`Inst. Dec. 6–8. After considering the full record, we adopt the same
`
`construction in this decision. Our reasons follow.
`
`In interpreting the claims of the ’966 patent, we “us[e] the same claim
`
`construction standard that would be used to construe the claim[s] in a civil
`
`action under 35 U.S.C. [§] 282(b).” 37 C.F.R. § 42.100(b).
`
`Under that standard, claim terms are generally given their ordinary
`
`and customary meaning as would be understood by one with ordinary skill
`
`in the art in the context of the specification, the prosecution history, other
`
`claims, and even extrinsic evidence. Phillips v. AWH Corp., 415 F.3d 1303,
`
`1312–17 (Fed. Cir. 2005) (en banc). But extrinsic evidence is “less
`
`significant” than the intrinsic record. Id. at 1317. The specification is the
`
`“single best guide” to a disputed term’s meaning. Id. at 1315. It “may reveal
`
`7
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`a special definition given to a claim term.” Id. at 1316. Or it may reveal “an
`
`intentional disclaimer, or disavowal, of claim scope.” Id. at 1316. “The
`
`patentee’s lexicography must, of course, appear ‘with reasonable clarity,
`
`deliberateness, and precision’ before it can affect the claim.’” Renishaw PLC
`
`v. Marposs Societa’ per Azioni, 158 F.3d 1243, 1249 (Fed. Cir. 1998). The
`
`patentee can disavow claim scope with language in the specification or the
`
`prosecution history. Poly-Am., L.P. v. API Indus., Inc., 839 F.3d 1131, 1136
`
`(Fed. Cir. 2016). “In either case, the standard for disavowal is exacting,
`
`requiring clear and unequivocal evidence that the claimed invention includes
`
`or does not include a particular feature.” Id.
`
`In Petitioner’s view, the ’966 patent defines the term “cast vote
`
`record.” Pet. 23 (quoting Ex. 1001, 1:23–28). We agree. Here, the patentee’s
`
`lexicography appears with “reasonable clarity, deliberateness, and
`
`precision.” Renishaw, 158 F.3d at 1249. Specifically, the patent states,
`
`As used herein, an electronic cast vote record is an electronic
`record that indicates the determined voter’s choice or selection
`in a manner determined in accordance with the voting system.
`Thus, an electronic cast vote record is an electronic record of a
`voter’s cast voting selections and may be used in the vote
`tabulation process.
`
`Ex. 1001, 1:23–28 (emphasis added). We emphasize “as used herein”
`
`because it indicates the patentee’s intent to define the term as described in
`
`the sentence. The claims, however, simply recite a “cast vote record”
`
`without the word “electronic.” See, e.g., id. at 8:43–62 (claim 1). So our
`
`construction of “cast vote record” follows the quoted definition but omits the
`
`word “electronic.” Accord Pet. 23.
`
`The ’966 patent also notes that the cast vote record “may be used in
`
`the vote tabulation process.” Id. (quoting Ex. 1001, 1:23–28) (emphasis
`
`8
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`added). We agree with Petitioner that the term “may” indicates that
`
`tabulation is not required. Id.
`
`Thus, we construe “cast vote record” as a “record that indicates the
`
`determined voter’s choice or selection in a manner determined in accordance
`
`with the voting system.” See Ex. 1001, 1:23–28; Pet. 23. Because the term
`
`“cast vote record” is not required to be electronic or tabulated, we do not
`
`include the second part of the quoted definition: “an electronic record of a
`
`voter’s cast voting selections and may be used in the vote tabulation
`
`process.” See Ex. 1001, 1:23–28.
`
`Petitioner cites a definition for “cast vote record” from the Electronic
`
`Voting Glossary: “permanent record of all votes produced by a single voter
`
`whether in electronic, paper or other form. Also referred to as BALLOT
`
`IMAGE when used to refer to electronic ballots. … An ‘archival record of
`
`all votes produced by a single voter.” Pet. 23–24 (quoting Ex. 1045, 7)
`
`(alteration in original). Petitioner, however, does not appear to rely on this
`
`definition in its unpatentability analysis. See generally Id. In particular,
`
`Petitioner’s rationale does not depend on the permanence of the record. Id.
`
`In contrast with Petitioner, Patent Owner argues that the cast vote
`
`record must be permanent. PO Resp. 4. Patent Owner argues that the
`
`definition in the patent is consistent with the definition in the Electronic
`
`Voting Glossary. Id. In Patent Owner’s view, the glossary defines the cast
`
`vote record as a “permanent record” or “archival record” of a voter’s vote
`
`selections. Sur-reply 18 (citing Ex. 1045, 7–8). According to Patent Owner,
`
`the claim excludes transient records. Id.
`
`But, here, the patentee created a special and particular definition for
`
`“cast vote record.” Ex. 1001, 1:23–28. That definition does not include any
`
`requirement that the record be permanently stored. See id. “[A] common
`
`9
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`meaning, such as one expressed in a relevant dictionary, that flies in the face
`
`of the patent disclosure is undeserving of fealty.” Renishaw, 158 F.3d at
`
`1249. Because the definition in the ’966 patent disclosure does not require a
`
`permanent record or even a record of tabulation, the construction of “cast
`
`vote record” in this decision does not require the record to be permanently
`
`stored.
`
`Apart from cast vote record, neither party offers or argues for any
`
`additional constructions. Because no other terms are in controversy, we do
`
`not construe any other claim terms in this decision.
`
`D.
`
`Principles of Law
`
`In this decision, we must determine whether Petitioner has established
`
`by a preponderance of the evidence that the specified claims are
`
`unpatentable over the cited prior art. 35 U.S.C. § 326(e). Petitioner “has the
`
`burden from the onset to show with particularity why the patent it challenges
`
`is unpatentable.” Harmonic Inc. v. Avid Tech., Inc., 815 F.3d 1356, 1363
`
`(Fed. Cir. 2016). This burden never shifts to Patent Owner. See Dynamic
`
`Drinkware, LLC v. Nat’l Graphics, Inc., 800 F.3d 1375, 1378 (Fed. Cir.
`
`2015).
`
`We previously instructed Patent Owner that “any arguments for
`
`patentability not raised in the [Patent Owner Response] may be deemed
`
`waived.” Paper 7, 8; see also In re NuVasive, Inc., 842 F.3d 1376, 1381
`
`(Fed. Cir. 2016) (holding that patent owner’s failure to make arguments at
`
`trial as instructed in the scheduling order constitutes waiver). Also, the
`
`Board’s Trial Practice Guide states that the Patent Owner Response “should
`
`identify all the involved claims that are believed to be patentable and state
`
`the basis for that belief.” Patent Trial and Appeal Board Consolidated Trial
`
`10
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`Practice Guide 66 (Nov. 20, 2019), available at http://www.uspto.gov/
`
`TrialPracticeGuideConsolidated (“CTPG”).
`
`E. Obviousness over Brockhouse and Heilper
`
`1.
`
`Claims 1 and 15
`
`Petitioner asserts that the subject matter of independent claims 1 and
`
`15 would have been obvious over Brockhouse and Heilper. Pet. 49–56. As
`
`discussed in the following sections, Petitioner asserts that Brockhouse
`
`teaches all limitations of claims 1 and 15, except for using a dictionary to
`
`improve the accuracy of the OCR process. Id. For that remaining limitation,
`
`Petitioner asserts that one of ordinary skill in the art would have used a
`
`dictionary, as taught by Heilper, for error correction. Id. at 56.
`
`a.
`
`Brockhouse
`
`Brockhouse describes a voting system that prints a voter’s selections
`
`on a card. See generally Ex. 1004, Abstract, Fig. 6. In one embodiment, the
`
`voter selects candidates on a touch-screen device. Id. ¶¶ 41–42. After the
`
`voter finishes selecting the candidates, the voting system prints those
`
`selections on an activation card. Id. ¶ 53. The activation card includes other
`
`information, such as a bar code. Id. In one mode, the tabulation device
`
`decodes voter selections on the activation card using OCR and generates the
`
`cast vote record from the resulting data set. Id. ¶¶ 130, 134.
`
`b.
`
`Heilper
`
`Heilper generally relates to OCR error correction. Ex. 1005, 1:6–9.
`
`Heilper explains that an OCR process is typically followed by an error-
`
`correction step. Id. at 1:20–21. For example, Heilper describes an algorithm
`
`for dictionary-based OCR error correction. Id. at 1:26–41.
`
`11
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`c.
`
`Preamble
`
`The preamble of claim 1 recites “A method for electronic voting using
`
`printed vote records.” Ex. 1001, 8:43–44. Claim 15 recites the same
`
`preamble. Id. at 9:64–65.
`
`To the extent that the preamble is limiting, Brockhouse teaches the
`
`subject matter recited in the preamble. Pet. 49 (citing Ex. 1004, Abstract;
`
`Shamos Decl. ¶¶ 208–209). In particular, the cited evidence shows that
`
`Brockhouse’s electronic voting station uses a printed activation card.
`
`Ex. 1004, Abstract. Brockhouse’s printed activation card shows the voter’s
`
`selections. Id. Thus, Petitioner shows that Brockhouse teaches a method for
`
`electronic voting using printed vote records, as recited in claims 1 and 15.
`
`Patent Owner does not substantively address these assertions or
`
`otherwise contend that the art lacks any of the features. We are persuaded by
`
`Petitioner’s cited evidence and analysis, which Patent Owner does not
`
`contest.
`
`d.
`
`Recording Vote Selections
`
`Claim 1 recites, in part, “recording a voter’s vote selections.”
`
`Ex. 1001, 8:45. Claim 15 recites a similar step but adds that the selection is
`
`“with a ballot marking device.” Id. at 9:66–67. The ’966 patent discloses that
`
`a “wide range” of BMDs may be used. Id. at 4:27–30. In one example, the
`
`voter enters selections via the BMD’s touch screen, and the BMD prints a
`
`PVR with an internal printer. Id. at 4:39–45.
`
`Petitioner asserts that Brockhouse’s voting apparatus performs the
`
`recited recording. See Pet. 50 (citing Ex. 1004 ¶¶ 16, 41–42, 53, 104–106,
`
`117, 134). Like the BMD disclosed in the ’966 patent, Brockhouse’s voting
`
`apparatus records a voter’s selections via a touch-screen display or a keypad.
`
`Ex. 1004 ¶¶ 41–42, cited in Pet. 50. After these selections, Brockhouse’s
`
`12
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`voting apparatus prints a PVR, i.e., an activation card, to create the cast vote
`
`record. Id. ¶ 53, cited in Pet. 50. Thus, Petitioner shows that Brockhouse
`
`records a voter’s vote selections with a BMD.
`
`Patent Owner does not substantively address these assertions or
`
`otherwise contend that the art lacks any of the features. We are persuaded by
`
`Petitioner’s cited evidence and analysis, which Patent Owner does not
`
`contest.
`
`e.
`
`Generating a Printed Vote Record
`
`Claim 1 recites, in part, “generating a printed vote record of the
`
`voter’s vote selections, the printed vote record containing voter readable text
`
`indicating the voter’s vote selections.” Ex. 1001, 8:46–48. Claim 15 recites a
`
`similar limitation but adds “a candidate name of a voter’s vote selection.”
`
`Id. at 10:1–4.
`
`Petitioner asserts that Brockhouse prints the recited PVR. Pet. 52.
`
`Brockhouse states that “after a voter has made his/her voting selections with
`
`universal voting station 12, printer 72 prints a master bar code 110, vote
`
`selection bar codes 112, and vote selections 114 on activation card 20.”
`
`Ex. 1004 ¶ 53, quoted in Pet. 52. One embodiment of Brockhouse’s
`
`activation card is shown below with Petitioner’s annotations. Pet. 52.
`
`13
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`
`
`The figure above shows Brockhouse’s activation card with Petitioner’s
`annotations. Pet. 52 (citing Ex. 1004, Fig. 6, ¶ 16).
`
`Figure 6 shows, among other things, the names “FRANK SINATRA”
`
`and “STEVIE RAY VAUGHAN.” Ex. 1004, Fig. 6. Brockhouse explains
`
`that the voter chose these candidates for the “BEST VOCAL ARTIST”
`
`contest. Id. ¶ 56. On this record, Petitioner shows that Brockhouse teaches a
`
`14
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`PVR containing voter-readable text indicating the voter’s vote selections:
`
`candidates Frank Sinatra and Stevie Ray Vaughn, among others. Pet. 52
`
`(citing Ex. 1004, Fig. 6, ¶¶ 16, 53, 106; Shamos Decl. ¶¶ 216–218).
`
`Patent Owner does not substantively address these assertions or
`
`otherwise contend that the art lacks any of the features. We are persuaded by
`
`Petitioner’s cited evidence and analysis, which Patent Owner does not
`
`contest.
`
`f.
`
`Scanning the Printed Vote Record
`
`Claim 1 recites, in part, “scanning the printed vote record, the
`
`scanning including scanning the voter readable text.” Ex. 1001, 8:49–50.
`
`Claim 15 recites, in part, “electronically scanning the printed vote record,
`
`the scanning including scanning the candidate name.” Id. at 10:5–6.
`
`Petitioner asserts that Brockhouse teaches these limitations. Pet. 53
`
`(citing Ex. 1004 ¶¶ 106, 130, 134; Shamos Decl. ¶¶ 219–221). In one of the
`
`cited paragraphs, Brockhouse teaches that, after tabulation device 14
`
`receives an activation card, ballot-scanner assembly 194 scans the card and
`
`sends the result to a processor. Ex. 1004 ¶ 130. Considering Brockhouse’s
`
`description of ballot-scanner assembly 194 here, Petitioner shows that
`
`Brockhouse teaches the limitations about scanning the PVR.
`
`Patent Owner does not substantively address these assertions or
`
`otherwise contend that the art lacks any of the features. We are persuaded by
`
`Petitioner’s cited evidence and analysis, which Patent Owner does not
`
`contest.
`
`g.
`
`Optical Character Recognition
`
`Claim 1 recites, in part, “utilizing optical character recognition (OCR)
`
`on the scanned voter readable text to create a data set which identifies the
`
`voter’s vote selections.” Ex. 1001, 8:51–53. Claim 15 has a similar
`
`15
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`limitation but recites candidate name instead of voter-readable text:
`
`“utilizing optical character recognition (OCR) on the scanned candidate
`
`name to create a first data set.” Id. at 10:7–8 (emphasis added).
`
`Petitioner asserts that Brockhouse teaches these limitations. Pet. 54
`
`(citing Ex. 1004 ¶ 134; Shamos Decl. ¶¶ 222–224). In the paragraph cited by
`
`Petitioner, Brockhouse discloses that devices 12 or 14 use OCR to “decode
`
`the voter selections 114 on the printed activation cards.” Ex. 1004 ¶ 134. As
`
`discussed above, Petitioner sufficiently shows that Brockhouse’s activation
`
`card has candidate names. Supra § II.E.1.e. Thus, Petitioner shows that
`
`Brockhouse teaches using the OCR process recited in claims 1 and 15.
`
`Patent Owner does not substantively address these assertions or
`
`otherwise contend that the art lacks any of the features. We are persuaded by
`
`Petitioner’s cited evidence and analysis, which Patent Owner does not
`
`contest.
`
`h.
`
`Generating a Cast Vote Record
`
`Claim 1 recites, in part, “generating a cast vote record from the data
`
`set so that voter selections in the cast vote record are based on the voter
`
`readable text.” Ex. 1001, 8:54–56. Claim 15 recites, in part, “generating a
`
`cast vote record based on the comparing of the first data set with the election
`
`dictionary so that voter selections in the cast vote record are based on the
`
`scanned candidate name.” Id. at 10:15–18.
`
`We construe “cast vote record” as “a record that indicates the
`
`determined voter’s choice or selection in a manner determined in accordance
`
`with the voting system.” See supra § II.C.
`
`Petitioner asserts that Brockhouse teaches generating the cast vote
`
`record, as recited. Pet. 54 (citing Ex. 1004 ¶¶ 117, 130, 134). In the cited
`
`paragraphs, Brockhouse teaches that the tabulation device uses OCR to
`
`16
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`decode voter selections on the card and generates the cast vote record from
`
`the resulting data set. Ex. 1004 ¶¶ 130, 134. Because the OCR data comes
`
`from the voter’s selections, Petitioner sufficiently shows that Brockhouse’s
`
`record indicates the determined voter’s choice—as required by the
`
`construction of cast vote record above. See supra § II.C. Petitioner shows
`
`that Brockhouse generates the cast vote record, as recited in claims 1 and 15.
`
`Patent Owner does not substantively address these assertions or
`
`otherwise contend that the art lacks any of the features. We are persuaded by
`
`Petitioner’s cited evidence and analysis, which Patent Owner does not
`
`contest.
`
`i.
`
`The Election Dictionary
`
`Claim 1 recites, in part,
`
`wherein identifying the voter’s vote selection includes
`comparing the data set to an election dictionary, the
`election dictionary containing potential vote
`choices,
`
`wherein the comparing the data set to the election
`dictionary is utilized to confirm an accuracy of the
`OCR to limit potential OCR errors.
`
`Ex. 1001, 8:57–62. Claim 15 recites a similar limitation. See id. at 10:9–14.
`
`(1) Petitioner’s Obviousness Rationale
`
`Petitioner asserts that, because of “the importance of ensuring that a
`
`cast vote record is accurate in an official election,” it would have been
`
`obvious “to incorporate the known error correction safeguards disclosed in
`
`Heilper, such as the use of a dictionary, with the common OCR procedures
`
`utilized in Brockhouse.” Pet. 56 (citing Shamos Decl. ¶¶ 231–235).
`
`Petitioner asserts that OCR processes are error prone. Id. (citing Ex. 1005,
`
`1:13–41). According to Petitioner, “One of ordinary skill in the art would
`
`17
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`have further known that the cast vote record should be based on the error-
`
`corrected OCR data set—a data set that as established above includes
`
`candidate name(s).” Id. (citing Shamos Decl. ¶ 234).
`
`Patent Owner argues that Petitioner has not shown that it would have
`
`been obvious to combine Heilper’s dictionary with Brockhouse’s voting
`
`system. PO Resp. 31–34. Patent Owner argues that Petitioner’s reasoning is
`
`“too generic and conclusory to support a finding of obviousness.” Id. at 32
`
`(quoting Pet. 56). We disagree.
`
`The record supports Petitioner’s rationale that Heilper would have
`
`provided “correction safeguards” for Brockhouse’s OCR: Heilper teaches
`
`that OCR processes typically segment the scanned document into individual
`
`characters and then apply a pattern-recognition algorithm to each character
`
`to find a match. Ex. 1005, 1:15–20. Heilper teaches that this is typically
`
`followed by an error-correction step because OCR processes are error prone.
`
`Id. at 1:20–21. Heilper’s error correction includes looking up each OCR-
`
`generated word in a dictionary and substituting the closest matching word.
`
`Id. at 1:21–25, quoted in Pet. 56.
`
`These teachings echo Dr. Shamos’s assertion that dictionaries “could
`
`and should be used to ensure the accuracy of the data set resulting from use
`
`of OCR,” which Heilper says is error prone. Shamos Decl. ¶ 231; Ex. 1005,
`
`1:20–21. Thus, we credit Dr. Shamos’s testimony on this point. See Shamos
`
`Decl. ¶ 231.
`
`Contrary to Patent Owner’s arguments, Petitioner’s assertions here are
`
`not “generic” or “conclusory.” PO Resp. 32, 34. Rather, with ample support
`
`from Dr. Shamos’s testimony and Heilper’s teachings, Petitioner proposes a
`
`specific improvement to Brockhouse: correcting the OCR-generated results
`
`by using Heilper’s dictionary. Id. Heilper’s teachings and Dr. Shamos’s
`
`18
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`testimony also support Petitioner’s conclusion that the accuracy of
`
`Brockhouse’s OCR could be improved. Pet. 56; Ex. 1005, 1:20–21
`
`(disclosing that OCR is error prone); Shamos Decl. ¶ 231.
`
`On this record, Petitioner provides sufficient “articulated reasoning
`
`with some rational underpinning to support the legal conclusion of
`
`obviousness.” In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006).
`
`(2) Patent Owner’s arguments about a “generic dictionary”
`
`Patent Owner argues that Heilper’s dictionary is not an “election
`
`dictionary,” as recited in the claim. PO Resp. 28–31; see also id. at 34
`
`(arguing that the combination does not teach this limitation). Patent Owner
`
`argues that the recited election dictionary’s purpose is to “provide reliable
`
`matching of the OCR results with the proper ballot choice.” Id. at 29 (citing
`
`Ex. 1001, 8:19–23). According to Patent Owner, Petitioner only argues that
`
`Heilper discloses a generic dictionary, which does not contain potential voter
`
`choices. Id. Patent Owner argues that Heilper’s generic dictionary does not
`
`fulfill the purpose of an election dictionary because most voter selections
`
`will be proper nouns, which do not appear in an ordinary dictionary.
`
`Id. at 29–30.
`
`The record, though, does not support Patent Owner’s narrow
`
`characterization of Heilper’s teachings as only applying to a particular type
`
`of OCR-generated content. Id. at 28–31, 34. Specifically, Heilper explains
`
`that the computer looks up the OCR-generated word in the dictionary to
`
`obtain the corrected word. Ex. 1005, 1:13–41, quoted in Pet. 56. This means
`
`that Heilper’s dictionary is an appropriate dictionary that is expected to
`
`contain the correct words. Id.
`
`The correct words are not limited to generic English words. Rather,
`
`the correct words are determined by the content to be corrected. For
`
`19
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`example, the appropriate dictionary and correct words could represent DNA
`
`sequences, as taught by Heilper, if DNA sequences were at issue. See
`
`Pet. Reply 15; see also PO Resp. 27 (discussing Ex. 1005, 3:10–13, 7:50–
`
`53). According to Dr. Lopresti, one would not expect to find DNA
`
`sequences in a generic English dictionary. Ex. 1053, 125:20–22 (Dr.
`
`Lopresti’s Deposition). Likewise, in the deposition, Dr. Lopresti
`
`acknowledged that Heilper could be used for non-English words.
`
`Id. at 113:19–114:19.
`
`In fact, Dr. Lopresti’s testimony indicates that the dictionary would be
`
`used for every OCR-generated word:
`
`[Y]ou would assume that if you’re going to do dictionary
`comparison, that every word that you would want to recognize in
`the document would be compared to that dictionary[’s] terms.
`
`Id. at 81:16–82:3, cited in Pet. Reply 16–17. Here, Brockhouse’s OCR-
`
`generated words include the candidate names. See Ex. 1004 ¶ 134; supra
`
`§ II.E.1.e. A person of ordinary skill in the art would understand that the
`
`“correct” words in the context of the combination would be candidates’
`
`names. See Ex. 1053, 81:16–82:3 (explaining that dictionaries contain the
`
`words that you want to recognize); Ex. 1005, 1:13–41 (explaining that
`
`dictionaries are used for correcting OCR-generated words). “A person of
`
`ordinary skill is also a person of ordinary creativity, not an automaton.” KSR
`
`Int’l Co. v. Teleflex Inc., 550 U.S. 389, 421 (2007). Thus, the record
`
`supports Petitioner’s position that, in the proposed combination, Heilper’s
`
`dictionary would contain the correct candidate names when applied to
`
`Brockhouse’s OCR-generated candidate names. See Pet. 56 (citing Shamos
`
`Decl. ¶ 234).
`
`20
`
`
`
`PGR2020-00031
`Patent 10,445,966 B1
`
`Using a dictionary that does not contain the words that you would
`
`expect to find, as described in Patent Owner’s characterization (PO
`
`Resp. 28–31), would be pointless and contradicts Heilper’s express teaching
`
`that the dictionary contains the corrected words. Ex. 1005, 1:13–41. If a
`
`dictionary did not contain the words that could be substituted for the OCR-
`
`generated words, then the dictionary would be unable to make the
`
`“correction” that Heilper describes. See id.; accord Ex. 1053, 81:16–82:3.
`
`Thus, the record does not support Patent Owner’s characterization of
`
`Heilper’s dictionary as generic. PO Resp. 28–31. Because Dr. Lopresti’s
`
`testimony on this issue relies on Patent Owner’s narrow characterization of
`
`Heilper, we give the testimony little weight. Lopresti Decl. ¶¶ 55–59.
`
`We disagree with Patent Owner that Petitioner’s Reply introduces a
`
`new theory about a “specialized dictionary.” Sur-reply 8–13. Rather, the
`
`Reply responds to Patent Owner’s mischaracterization of Heilper’s