`
`——————————
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`——————————
`
`GOOGLE LLC,
`
`Petitioner,
`
`v.
`
`UNILOC 2017 LLC,
`
`Patent Owner.
`
`——————————
`
`Case No. IPR2020-00755
`U.S. Patent No. 6,366,908
`Filing Date: December 30, 1999
`Issue Date: April 2, 2002
`
`PETITION FOR INTER PARTES REVIEW
`
`
`
`
`
`I.
`
`II.
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`TABLE OF CONTENT
`
`Introduction ...................................................................................................... 1
`
`Overview of Prior Art ...................................................................................... 2
`
`A. Braden-Harder ......................................................................................... 2
`
`1.
`
`Incorporation of the Heidorn and Messerly patent applications ...... 5
`
`2. Heidorn ............................................................................................. 7
`
`3. Messerly ........................................................................................... 9
`
`B. Grossman ............................................................................................... 12
`
`C. Kucera .................................................................................................... 15
`
`D. Miller ...................................................................................................... 16
`
`III. The ’908 Patent .............................................................................................. 17
`
`A. Overview ................................................................................................ 17
`
`B. Challenged Claims ................................................................................. 21
`
`IV. Level of Ordinary Skill .................................................................................. 22
`
`V.
`
`Claim Construction ........................................................................................ 23
`
`A. Prior Art Renders Claims Obvious Under Both Parties’ Proposed
`District Court Constructions .................................................................. 23
`
`B. Step-Plus-Function Terms ..................................................................... 25
`
`1.
`
`2.
`
`3.
`
`“keyfact extracting step for…” (Functions [6a1], [6a2], [6a3]) ....25
`
`“keyfact indexing step for…” (Functions [6b1], [6b2]) ................27
`
`“keyfact retrieving step for…” (Functions [6c1], [6c2], [6c3]) .....29
`
`i
`
`
`
`
`VI. Claims 6-12 Should Be Cancelled ................................................................. 32
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`A. Claims 6-12 Would Have Been Obvious over Braden-Harder in
`view of Grossman (Ground 1) or over Braden-Harder in view of
`Heidorn, Messerly, and Grossman (Ground 2) ..................................... 33
`
`1.
`
`Independent claim 6 .......................................................................33
`
`2. Claim 7 ...........................................................................................63
`
`3. Claim 8 ...........................................................................................65
`
`4. Claim 9 ...........................................................................................69
`
`5. Claim 10 .........................................................................................70
`
`6. Claim 11 .........................................................................................79
`
`7. Claim 12 .........................................................................................83
`
`B. Claims 7-9 Would Have Been Obvious over Braden-Harder in
`View of Grossman, as Set Forth in Ground 1, and Further in
`View of Kucera (Ground 3) or over Braden-Harder in View of
`Heidorn, Messerly, and Grossman, as Set Forth in Ground 2, and
`Further in View of Kucera (Ground 4) .................................................. 85
`
`1. Claim 7 ...........................................................................................85
`
`2. Claims 8-9 ......................................................................................87
`
`C. Claim 9 Would Have Been Obvious over Braden-Harder in
`View of Grossman, as Set Forth in Ground 1, and Further in
`View of Miller (Ground 5) or over Braden-Harder in View of
`Heidorn, Messerly, and Grossman, as Set Forth in Ground 2, and
`Further in View of Miller (Ground 6).................................................... 87
`
`ii
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`D. Claim 9 Would Have Been Obvious over Braden-Harder in
`View of Grossman and Kucera, as Detailed in Ground 3, and
`Further in View of Miller (Ground 7) or over Braden-Harder in
`View of Heidorn, Messerly, Grossman, and Kucera, as Detailed
`in Ground 4, and Further in View of Miller (Ground 8) ....................... 90
`
`VII. Google Raises New Unpatentability Grounds ............................................... 90
`
`VIII. Mandatory Notices......................................................................................... 93
`
`A. Real Party-in-Interest ............................................................................. 93
`
`B. Related Matters (37 C.F.R. §42.8(b)(2)) ............................................... 93
`
`C. Counsel and Service Information .......................................................... 94
`
`IX. Grounds for Standing ..................................................................................... 94
`
`X.
`
`Conclusion ..................................................................................................... 95
`
`
`
`
`
`
`
`iii
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`TABLE OF AUTHORITIES
`
` Page(s)
`
`
`
`Cases
`
`Advanced Display Sys., Inc. v. Kent State Univ.,
`212 F.3d 1272 (Fed. Cir. 2000) ............................................................................ 6
`
`Blackboard, Inc. v. Desire2Learn, Inc.,
`574 F.3d 1371 (Fed. Cir. 2009) .............................................................. 25, 27, 29
`
`Commonwealth Sci. & Indus. Research Org. v. Buffalo Tech. (USA), Inc.,
`542 F.3d 1363 (Fed. Cir. 2008) .......................................................................... 33
`
`In re Google LLC,
`No. 2019-126, 2020 U.S. App. LEXIS 4588 (Fed. Cir. Feb. 13, 2020) ............. 92
`
`Harari v. Lee,
`656 F.3d 1331 (Fed. Cir. 2011) ............................................................................ 6
`
`Masco Corp. v. United States,
`303 F.3d 1316 (Fed. Cir. 2002) .......................................................................... 25
`
`Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co.,
`868 F.3d 1013 (Fed. Cir. 2017) .......................................................................... 23
`
`Okajima v. Bourdeau,
`261 F.3d 1350 (Fed. Cir. 2001) .......................................................................... 22
`
`Paice LLC v. Ford Motor Co.,
`881 F.3d 894 (Fed. Cir. 2018) .............................................................................. 6
`
`Phillips v. AWH Corp.,
`415 F.3d 1303 (Fed. Cir. 2005) (en banc) .......................................................... 23
`
`iv
`
`
`
`
`Seal-Flex, Inc. v. Athletic Track & Court Const.,
`172 F.3d 836 (Fed. Cir. 1999) (Rader, J., concurring) ....................................... 25
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc.,
`200 F.3d 795 (Fed. Cir. 1999) ............................................................................ 23
`
`Uniloc 2017 LLC v. Google LLC,
`2:18-cv-00553 (E.D. Tex.) .................................................................................. 93
`
`PTAB Cases
`
`Advanced Bionics, LLC, v. Med-El Elektromedizinische Geräte Gmbh,
`IPR2019-01469, Paper 6 (PTAB February 13, 2020) (precedential) ................. 90
`
`Becton Dickinson & Co. v. B. Braun Melsungen AG,
`IPR2017-01586, Paper 8 (PTAB Dec. 15, 2017) ......................................... 90, 91
`
`Facebook, Inc. v. Blackberry Ltd.,
`IPR2019-00899, Paper 15 (PTAB Oct. 8, 2019) ................................................ 92
`
`NHK Spring Co., Ltd. v. Intri-Plex Techs. Inc.,
`IPR2018-00752, Paper 8 (PTAB Sept. 12, 2018) (precedential) ....................... 92
`
`Precision Planting, LLC et al. v. Deere & Co.,
`IPR2019-01044, Paper 17 (PTAB Dec. 2, 2019) ............................................... 92
`
`Resideo Techs., Inc. v. Innov. Scis., LLC,
`IPR2019-01306, Paper 19 (PTAB Jan. 27, 2020) .............................................. 92
`
`Statutes
`
`35 U.S.C. §102(a) .................................................................................................... 12
`
`35 U.S.C. §102(b) .............................................................................................. 15, 16
`
`35 U.S.C. §102(e) .............................................................................................. 2, 7, 9
`
`v
`
`
`
`
`35 U.S.C. §103 ................................................................................................... 30, 31
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`35 U.S.C. §112 ..................................................................................................passim
`
`Other Authorities
`
`37 C.F.R. §42.8(b)(2) ............................................................................................... 93
`
`37 C.F.R. §314 ................................................................................................... 92, 93
`
`37 C.F.R. §325 ................................................................................................... 90, 92
`
`PTAB Tr. Prac. Guide at 48 (November 2019) ....................................................... 24
`
`Changes to the Claim Construction Standard for Interpreting Claims
`in Trial Proceedings Before the Patent Trial and Appeal Board,
`83 Fed. Reg. 51340 (Oct. 11, 2018) ................................................................... 23
`
`vi
`
`
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`TABLE OF EXHIBITS
`
`Exhibit
`Ex. 1001
`
`Description
`U.S. Patent No. 6,366,908 to Chong et al.
`
`
`
`Ex. 1002
`
`
`
` Prosecution history of U.S. Patent No. 6,366,908
`
`Ex. 1003
`
`
`
` Declaration of Jim Jansen, Ph.D., including his Curriculum Vitae
`
`Ex. 1004
`
`Ex. 1005
`
`Ex. 1006
`
` Opening Claim Construction Brief of Uniloc 2017, Uniloc 2017
`
`LLC v. Google LLC, 2-18-cv-00553, ECF No. 128 (E.D. Tex.
`January 9, 2020)
`
` Defendant Google LLC’s Responsive Claim Construction Brief,
`
`Uniloc 2017 LLC v. Google LLC, 2-18-cv-00553, ECF No. 134
`(E.D. Tex. January 23, 2020)
`
` Uniloc 2017’s Reply Brief on Claim Construction, Uniloc 2017
`
`LLC v. Google LLC, 2-18-cv-00553, ECF No. 139 (E.D. Tex.
`January 30, 2020)
`
`Ex. 1007
`
` Proof of Service, Uniloc 2017 LLC v. Google LLC, 2-18-cv-
`
`00553, ECF No. 8 (E.D. Tex. April 3, 2019)
`
`Ex. 1008
`
`
`
` Declaration from Kelley Hayes Greenhill
`
`Ex. 1009
`
`
`
` Declaration from Kelley Hayes Greenhill
`
`Ex. 1010
`
` David A. Grossman & Ophir Frieder, Information Retrieval:
`
`Algorithms and Heuristics, Kluwer International Series in
`
`vii
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`Engineering and Computer Science (Kluwer Academic
`Publishers, Springer Science+Business Media New York prtg.)
`(1st ed. 1998) (“Grossman”)
`
`Ex. 1011
`
`
`
` U.S. Patent No. 4,868,750 to Kucera et al. (“Kucera”)
`
`Ex. 1012
`
`Ex. 1013
`
` Julio Gonzalo, Felisa Verdejo, Irina Chugur & Juan Cigarrán,
`
`Indexing with WordNet Synsets Can Improve Text Retrieval,
`Proceedings of the COLING/ACL'98 Workshop on Usage of
`WordNet for NLP, Montreal (1998) (“Gonzalo”)
`
` Ellen Riloff, Little Words Can Make a Big Difference for Text
`
`Classification, Proceedings of the 18th Annual International
`ACM SIGIR Conference on Research and Development in
`Information Retrieval, 130-36 (1995)
`
`Ex. 1014
`
`
`
` European Patent No. EP 0597630 to Addison et al.
`
`Ex. 1015
`
`Ex. 1016
`
`Ex. 1017
`
`
`
` U.S. Patent Application Publication No. 2001/0014852 to
`Tsourikov et al.
`
` Jonghee Choi, Dongsi Choi, Seyoung Park & Heekuck Oh, A
`
`Method for Improving Recall Precision on Information Retrieval
`Systems Using Multiple Terms, J. of KIISE, Vol. 25, No. 2 (1998)
`
` Mi-Seon Jun, Se-Young Park, Man-Soo Kim, The Fact
`
`Extraction Using the Keyfact, Applications of Natural Language
`to Information Systems, Proceedings of the 2nd International
`Workshop, 139-150 (1996)
`
`viii
`
`
`
`
`
`Ex. 1018
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
` Myung-Gil Jang, Sung Hyon Myaeng & Se Young Park, Using
`
`Mutual Information to Resolve Query Translation Ambiguities
`and Query Term Weighting, ACL 1999 Proceedings of the 37th
`Annual Meeting of the Association for Computational Linguistics
`on Computational Linguistics, 223-29 (June 20, 1999)
`
`Ex. 1019
`
`
`
` Declaration from Julio Gonzalo
`
`Ex. 1020
`
`Ex. 1021
`
`
`
` U.S. Patent No. 5,933,822 to Braden-Harder et al. (“Braden-
`Harder”)
`
` U.S. Patent Application No. 08/674,610 to Heidorn et al.
`
`(“Heidorn patent application”)
`
`Ex. 1022
`
`
`
` U.S. Patent No. 5,966,686 to Heidorn et al. (“Heidorn patent”)
`
`Ex. 1023
`
`
`
` Prosecution history of U.S. Patent No. 5,966,686
`
`Ex. 1024
`
` U.S. Patent Application No. 08/886,814 to Messerly et al.
`
`(“Messerly patent application”)
`
`Ex. 1025
`
`
`
` U.S. Patent No. 6,076,051 to Messerly et al. (“Messerly patent”)
`
`Ex. 1026
`
`
`
` Prosecution history of U.S. Patent No. 6,076,051
`
`Ex. 1027
`
` George A. Miller, Richard Beckwith, Christiane Fellbaum, Derek
`
`Gross & Katherine J. Miller, Introduction to WordNet: An On-
`line Lexical Database, International Journal of Lexicography,
`Vol. 3 No. 4, 235-244 (1990) (“Miller”)
`
`ix
`
`
`
`
`
`Ex. 1028
`
`Ex. 1029
`
`Ex. 1030
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
` Richard Beckwith & George A. Miller, Implementing a Lexical
`
`Network, International Journal of Lexicography, Vol. 3 No. 4,
`235-244 (1990) (“Beckwith”)
`
` Christopher D. Manning & Hinrich Schütze, Foundations of
`
`Statistical Natural Language Processing, MIT Press (1999)
`(“Manning”)
`
` Richard Beckwith, George A. Miller & Randee Tengi, Design
`
`and Implementation of the WordNet Lexical Database and
`Searching Software (revised version of Implementing a Lexical
`Network), Collection of Five Papers of the State of WordNet as of
`1990, 62-86 (August 1993) (“Tengi”)
`
`Ex. 1031
`
` Karen Sparck Jones, Natural Language Processing: a Historical
`
`Review (October 2011) (“Jones”)
`
`Ex. 1032
`
`Reserved
`
`
`
`Ex. 1033
`
`Ex. 1034
`
` Michael Sussna, Word Sense Disambiguation for Free-text
`
`Indexing Using a Massive Semantic Network, ACM, 67-74 (1993)
`
` Robert Krovetz & W. Bruce Croft, Lexical Ambiguity and
`
`Information Retrieval, ACM, Vol. 10, 115-141 (1992)
`
`Ex. 1035
`
`
`
` U.S. Patent No. 5,721,902 to Schultz
`
`Ex. 1036
`
`
`
` U.S. Patent No. 5,598,557 to Doner et al.
`
`Ex. 1037
`
`
`
` U.S. Patent No. 5,289,375 to Fukumochi et al.
`
`x
`
`
`
`
`
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`Ex. 1038
`
`
`
` U.S. Patent No. 5,771,378 to Holt et al.
`
`Ex. 1039
`
`
`
` U.S. Patent No. 5,541,836 Church et al.
`
`Ex. 1040
`
`Ex. 1041
`
`Ex. 1042
`
` Order to Continue Stay of Case, Uniloc 2017 LLC v. Google LLC,
`
`2-18-cv-00553, ECF No. 165 (E.D. Tex. March 20, 2020)
`
` General Docket, In re Google LLC, Case No. 2019-126 (Fed.
`
`Cir.), retrieved on March 25, 2020
`
` Defendant Google LLC’s P.R. 3-3 and 3-4 Invalidity Contentions,
`
`Uniloc 2017 LLC v. Google LLC, 2-18-cv-00553, ECF No. 128
`(E.D. Tex. Served August 26, 2019)
`
`Ex. 1043
`
`
`
` Declaration of Pamela Stansbury
`
`Ex. 1044
`
`Ex. 1045
`
` E-mail from Brett Mangrum, Counsel for Patent Owner, to Erika
`
`H. Arner, Counsel for Petitioner (Tuesday, March 24, 2020,
`12:49:33 PM) regarding Google IPR petitions
`
` E-mail from Andrew Kellogg, Supervisory Paralegal, Patent Trial
`
`and Appeal Board, to Erika H. Arner, Counsel for Petitioner
`(Wednesday, March 25, 2020, 11:25:40 AM) regarding
`Requesting Electronic IPR Service Under 37 CFR 42.5(b) in view
`of COVID-19
`
`
`
`xi
`
`
`
`
`I.
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`Introduction
`Petitioner Google LLC1 requests inter partes review (IPR) and cancellation
`
`of claims 6-12 of the ’908 patent (Ex. 1001).
`
`The ’908 patent was filed on December 30, 1999, and claims foreign priority
`
`to a Korean application filed on June 28, 1999. Ex. 1001, Cover. The patent uses
`
`conventional “natural language processing” (“NLP”) such as “syntactic analysis”
`
`to extract “keyfacts,” which express semantic relation between words, for a
`
`concept-based information retrieval (“IR”) method. Id., 1:7-19, 1:48-49, 2:10-18;
`
`4:58-60.
`
`The USPTO allowed the patent without applying any prior art. Ex. 1002,
`
`111-112. While claims 1-12 were allowed, the reasons for allowance only
`
`identified allowable subject matter in dependent claims 2 and 7. Id. As
`
`demonstrated by the prior art here, none of the claims should have been allowed.
`
`They merely recite conventional NLP and IR techniques and should be cancelled.
`
`
`Google LLC is a subsidiary of XXVI Holdings Inc., which is a subsidiary of
`
`1
`
`Alphabet Inc. XXVI Holdings Inc. and Alphabet Inc. are not real parties-in-interest
`
`to this proceeding.
`
`1
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`II. Overview of Prior Art
`A. Braden-Harder
`Braden-Harder, U.S. Patent No. 5,933,822 (Ex. 1020), assigned to
`
`Microsoft Corporation, issued from an application filed on July 22, 1997, and is
`
`prior art at least under pre-AIA 35 U.S.C. §102(e).
`
`Recognizing that existing “[k]eyword based search engines totally ignore…
`
`[a language’s] fine-grained linguistic structure,” Braden-Harder describes methods
`
`for an IR system that employs NLP to improve “accuracy” and “precision in
`
`retrieving textual information.” Ex. 1020, 3:23-24, 5:5, 7:13-14. Braden-Harder
`
`teaches both preprocessing and retrieval processes. Id., 5:1-6:3.
`
`Braden-Harder teaches preprocessing documents being indexed into the
`
`database to save “execution time whenever that document is subsequently
`
`retrieved.” Id., 5:65-6:2. During this preprocessing, Braden-Harder
`
`morphologically analyzes and parses the documents using a lexicon into words.
`
`Id., 11:62-12:5, 12:30-34. Using the lexicon, each word is assigned a part-of-
`
`speech tag. Id., 12:41-42, 43. After assigning the part-of-speech tags, Braden-
`
`Harder uses syntactic rules to resolve any part-of-speech ambiguity and to
`
`construct a syntactic-parse tree representing the grammatical structure of the text.
`
`Id., 12:37-40.
`
`2
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Id., Table 1. Finally, by applying semantic rules to the syntactic-parse tree,
`
`Braden-Harder creates a logical form graph (13:11), which is a pattern used to
`
`generate “logical form triples” that portray “semantic relationships…between
`
`important words in an input string.” Id., 11:35-61.
`
`3
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Id., Figs. 5A-5D
`
`In the retrieval process, “[a] user-supplied query is analyzed in the same
`
`manner to yield a set of corresponding logical form triples,” which is “compared to
`
`the sets of logical forms associated with each of the retrieved documents in order
`
`to ascertain a match between logical forms from the query set and logical forms
`
`from each document set.” Id., 5:33-39. The retrieved documents are scored using
`
`predefined weights based on each logical-form-triple relation type. Id., 5:9-11, Fig.
`
`8A.
`
`4
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Id., Fig. 8A. The scored documents are presented to the user in descending rank
`
`order based on their scores. Id., 8:2-6, 16:35-40.
`
`Incorporation of the Heidorn and Messerly patent applications
`1.
`In addition to its broad teachings, Braden-Harder further refers to and
`
`incorporates two co-pending U.S. patent applications assigned to Microsoft:
`
`[T]he reader is referred to co-pending United States patent
`applications entitled “Method and System for Computing Semantic
`Logical Forms from Syntax Trees”, filed Jun. 28, 1996 and assigned
`Ser. No. 08/674,610 [Ex. 1021; Heidorn patent application] and
`particularly “Information Retrieval Utilizing Semantic Representation
`of Text”, filed Mar. 7, 1997 and assigned Ser. No. 08/886,814
`[Ex. 1024; Messerly patent application.]; both of which have been
`assigned to the present assignee hereof and are incorporated by
`reference herein.
`
`Id., 14:53-61 (emphasis added). Braden-Harder clearly identifies the patent-
`
`application number and title for the two patent applications, and explicitly states
`
`5
`
`
`
`
`that the two patent applications are incorporated by reference. Because Braden-
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`Harder identifies the Heidorn and Messerly patent applications (and not just
`
`particular portions of their disclosures) and where they can be found (Ser. Nos.
`
`08/674,610 and 08/886,814), such broad and unequivocal language is plainly
`
`sufficient to incorporate both patent applications in their entirety. Paice LLC v.
`
`Ford Motor Co., 881 F.3d 894, 907 (Fed. Cir. 2018) (finding “broad and
`
`unambiguous” language is “plainly sufficient” to incorporate prior art patent in its
`
`entirety); Harari v. Lee, 656 F.3d 1331, 1335 (Fed. Cir. 2011) (finding that prior
`
`art applications were incorporated in their entirety based on “broad and
`
`unequivocal language”).
`
`Everything disclosed in the Heidorn and Messerly patent applications is
`
`“effectively part of” Braden-Harder as if these disclosures were “explicitly
`
`contained therein.” Paice, 881 F.3d at 907 (citing Advanced Display Sys., Inc. v.
`
`Kent State Univ., 212 F.3d 1272, 1282 (Fed. Cir. 2000)).
`
`6
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`2. Heidorn
`U.S. Patent No. 5,966,686 (Heidorn patent; Ex. 1022)2 issued on October
`
`12, 1999, from the Heidorn patent application filed on June 28, 1996; the Heidorn
`
`patent is prior art at least under §102(e).
`
`Heidorn teaches NLP methods for analyzing sentences using morphological,
`
`syntactical, and semantical processing. It explains that in prior-art NLP, “the
`
`morphological subsystem recognizes each word…of the input text as a separate
`
`token and constructs an attribute/value record for each part of speech of each token
`
`using the dictionary information.” Ex. 1022, 3:36-50. “These attribute/value
`
`records are then passed to the syntactic subsystem…, where they are used as the
`
`leaf nodes of the syntax parse tree that the syntactic subsystem constructs.” Id.,
`
`3:42-45. Using prior-art morphological and syntactic analyses, Heidorn transforms
`
`an input sentence—e.g., “The person who I met was my friend”—into a syntax-
`
`parse tree:
`
`
`2 Google cites to the Heidorn patent for convenience. It issued from the Heidorn
`
`patent application without any amendments to its specification during prosecution.
`
`Ex. 1023. The Heidorn patent and Heidorn patent application (collectively,
`
`Heidorn) disclose substantively the same subject matter and contain substantively
`
`identical disclosures.
`
`7
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Id., Fig. 22.
`
`This syntax-parse tree, identical to Braden-Harder’s syntactic-parse tree, is a
`
`hierarchically ordered representation of a sentence. Each lower-level node
`
`represents “a single part of speech that an input word can represent in a sentence.
`
`These parts of speech are found as attribute/value pairs in the dictionary entries.”
`
`Id., 4:44-45. For words with more than one part of speech (e.g., the and my),
`
`additional nodes are used to represent all “possible parts of speech for the word…
`
`that are found as attributes.” Id., 4:47-48. These part-of-speech ambiguities are
`
`resolved using syntax rules. Heidorn’s final syntax-parse tree—shown in Fig. 22
`
`above—extracts words without part-of-speech ambiguity.
`
`Once the syntax-parse tree is formed, Heidorn constructs a logical form
`
`graph using semantic rules.
`
`8
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Id., Fig. 43. “The logical form graph is a labeled, directed graph” that is “a
`
`semantic representation of an input sentence.” Id., 5:25-26.
`
`3. Messerly
`U.S. Patent No. 6,076,051 (Messerly patent; Ex. 1025),3 issued on June 13,
`
`2000, from the Messerly patent application filed on March 7, 1997, is prior art at
`
`least under §102(e).
`
`
`3 Google cites to the Messerly patent for convenience. It issued from the Messerly
`
`patent application with one correction to a typographical error during prosecution.
`
`Ex. 1026, 258. The Messerly patent and Messerly patent application (collectively,
`
`
`
`9
`
`
`
`Messerly relates to IR methods using semantic representation of text.
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Messerly, Abstract. As Messerly explains, conventional IR systems tokenize a
`
`sentence into a series of tokens (“characters intelligible to and differentiable by an
`
`information retrieval engine”). Ex. 1025, 1:21-23. Messerly’s improved tokenizer
`
`parses “both indexed and query text to perform lexical, syntactic, and semantic
`
`analysis of this input text” to produce “one or more logical forms, which identify
`
`words that perform primary roles in the query text and their intended senses, and
`
`that further identify the relationship between those words.” Id., 2:45-51. Messerly’s
`
`“parser preferably produces logical forms that relate the deep subject, verb, and
`
`deep object of the input text,” like Braden-Harder. Id., 2:52-53. These generated
`
`logical forms are transformed into “tokens intelligible by” the IR system that
`
`“compares the tokenized query to the index.” Id., 3:26-28.
`
`In constructing an index representing target documents, Messerly “stores the
`
`tokens…in the index with locations at which they occur.” Id., 10:19-21. Figure 13
`
`shows an index table that maps “each token to the identity of the document and
`
`location in the document at which they occur.” Id., 10:19-24.
`
`
`Messerly) disclose substantively the same subject matter and contain substantively
`
`identical disclosures.
`
`10
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Id., Fig. 13. In addition to an index table, Messerly’s index may be stored in “a
`
`number of other forms that support more efficient locations of a token in the index,
`
`such as in tree form.” Id., 10:25-28.
`
`In processing a query to identify “matches of tokens in the index,” Messerly
`
`“rank[s] the target documents…in the order of their relevance to the query.” Id.,
`
`12:27-32. Messerly teaches “a number of well-known approaches to ranking
`
`documents by relevancy” and “uses a combination of inverse document frequency
`
`and term frequency waiting [sic] to rank the matching target documents.” Id.,
`
`12:33-38.
`
`11
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`
`
`Id., Equations (1)-(2). Messerly calculates a tf-idf score for each matching token
`
`combination in each document using a dot product:
`
`
`
`Id., Equations (3). Once the document scores are calculated, Messerly further
`
`“augment[s] these scores” or “calculates a normalized score for each document”
`
`by, for example, accounting for “the size of the document.” Id., 13:22-37. The
`
`matching documents are ranked in the order of the scores. Id., 13:42-48.
`
`B. Grossman
`Grossman, Information Retrieval: Algorithms and Heuristics, Kluwer
`
`International Series in Engineering and Computer Science (Kluwer Academic
`
`Publishers, Springer Science+Business Media New York prtg.) (1st ed. 1998)
`
`(Ex. 1010) is a textbook published in 1998. A copy of textbook was cataloged and
`
`available at the UCB Library by November 17, 1998. Ex. 1009. Grossman is prior
`
`art at least under §102(a).
`
`12
`
`
`
`Grossman teaches the structure and operation of conventional IR systems.
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`Grossman teaches “[s]ince many document collections are reasonably static, it is
`
`feasible to build an inverted index to quickly find terms in the document
`
`collection.” Ex. 1010, 134. Grossman teaches creating an Index file, Document
`
`file, and Weight file for a document collection. Id., 134-37. Grossman also teaches
`
`“preprocessing” index information using a set of DOC, INDEX, and TERM tables
`
`in a relational database. Id., 154-55, 169-76. These tables contain information
`
`about the documents in the collection, specific terms in those documents, and
`
`statistical information (such as tf, idf frequencies) for each of the collection’s
`
`indexed terms.
`
`Each document (DOC) contains structured and unstructured data represented
`
`in various fields in the following exemplary document:
`
`Id., 171. Information from these document fields is stored in the DOC table, such
`
`as in Grossman’s Table 5.5:
`
`
`
`13
`
`
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`Id., 172. Other index information, including terms, term frequencies, and inverse
`
`document frequencies, is stored in INDEX and TERM tables of the relational
`
`database, as shown below:
`
`
`
`
`
`Id., 172, 175.
`
`Grossman teaches certain common “STOP” terms (“a,” “the,” and “and”)
`
`that are ignored in parsing sentences in documents and user queries. Id., 174-75,
`
`Table 5.10 (“STOP_TERM”).
`
`14
`
`
`
`Grossman teaches a vector-space model for calculating a weight (d) for a
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`
`term in a document by multiplying its tf and idf frequencies and calculating a
`
`similarity coefficient (SC) that measures the correspondence of a user-query vector
`
`and a document vector. Id., 11, 13-17. Documents that “correspond most closely to
`
`the content of the query are judged to be the most relevant.” Id., 13.
`
`C. Kucera
`Kucera, U.S. Patent No. 4,868,750 (Ex. 1011), issued September 19, 1989,
`
`is prior art under at least §102(b).
`
`Kucera teaches a “collocational grammar system” that annotates each
`
`received word with a set of tags indicative of its possible grammatical or syntactic
`
`uses. Ex. 1011, Cover. If a word is ambiguous, it may be tagged with multiple part-
`
`of-speech tags. Id., 11:1-35, FIG. 2. Kucera’s disambiguator determines the most
`
`probable part-of-speech tag for the ambiguous word. Id., FIG. 1 (reference 10a),
`
`Abstract.
`
`Kucera teaches “disambiguation processing” using “probabilistic means to
`
`determine, for each maximal ambiguously tagged string of words, a ‘most probable
`
`parse’ (denoted MPP).” Id., 11:36-41. Kucera further teaches generating candidate
`
`sequences of part-of-speech tags where “[o]f the v possible tag sequences, a single
`
`sequence Xc is selected as most probably correct by defining a local probability-
`
`15
`
`
`
`
`like distribution (called a φ function) on pairs of adjacent tags….” Id., 11:1-12:8;
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`see also id., 39:14-40:4.
`
` Miller
`D.
`Miller, Introduction to WordNet: An On-line Lexical Database, International
`
`Journal of Lexicography (1990) (Ex. 1027) is a paper published by Oxford
`
`University Press. A copy was received by the University of Wisconsin-Madison
`
`Libraries on January 4, 1991, and catalogued and made available to library patrons
`
`in 1991. Id. Miller is prior art at least under §102(b).
`
`Miller teaches an on-line lexical reference system, WordNet, that organizes
`
`its entries based on synonym sets “to provide an aid to use in searching dictionaries
`
`conceptually, rather than merely alphabetically.” Ex. 1027, 236. Specifically,
`
`“WordNet divides the lexicon into four categories: nouns, verbs, modifiers, and
`
`function words” and contains “nouns, verbs, and adjectives.” Id., 237. Each of
`
`these parts of speech are partitioned by its concepts into “synsets.” Id., 241.
`
`Further, “nouns are organized…as topical hierarchies, adjectives are organized as
`
`N-dimensional hyperspaces, and verbs are oganized [sic] by a variety of entailment
`
`relations.” Id., 237. As such, each of these lexical structures (indexes) in WordNet
`
`reflects its own separate sub-dictionary.
`
`16
`
`
`
`IPR2020-00755 Petition
`U.S. Patent No. 6,366,908
`
`
`III. The ’908 Patent
`A. Overview
`Like the prior art (Ex. 1003, ¶¶45-120), the ’908 patent describes an NLP-
`
`based method for indexing and retrieval of documents based on semantic
`
`relationships. But it uses different terminology. For example, instead of “logical
`
`form triples,” the patent uses “keyfacts,” explaining that a “keyfact means an
`
`important fact contained in sentences which constitute a document” (Ex. 1001,
`
`1:15-16), and “a part of text that represent[s] the same meaning is described as a
`
`keyfa