`
`Paper No. ___
`Filed: September 24, 2021
`
`
`
`
`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`_____________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`_____________________________
`
`AMAZON.COM, INC.; AMAZON.COM LLC; AMAZON WEB
`SERVICES, INC.; A2Z DEVELOPMENT CENTER, INC. D/B/A LAB126;
`RAWLES, LLC; AMZN MOBILE LLC; AMZN MOBILE 2 LLC;
`AMAZON.COM SERVICES, INC. F/K/A AMAZON FULFILLMENT
`SERVICES, INC.; AND AMAZON.COM SERVICES LLC (formerly
`AMAZON DIGITAL SERVICE LLC),
`Petitioners,
`
`v.
`
`VB ASSETS, LLC,
`Patent Owner.
`_____________________________
`
`Case IPR2020-01367
`Patent No. 8,073,681
`_____________________________
`
`
`PATENT OWNER’S SURREPLY
`
`
`
`TABLE OF CONTENTS
`
`B.
`
`C.
`
`I.
`II.
`
`INTRODUCTION ........................................................................................... 1
`PETITIONERS FAIL TO ESTABLISH THAT KENNEWICK RENDERS
`OBVIOUS “ONE OR MORE WORDS HAVING DIFFERENT
`MEANINGS IN DIFFERENT CONTEXTS”................................................. 2
`A.
`Petitioners’ new claim construction argument broadly construes the
`limitation to include any ambiguity in the user’s request ..................... 2
`Kennewick’s “temperature” example does not disclose a word with
`“different meanings in different contexts” ............................................ 9
`Kennewick’s “flight one hundred and twenty too” example does not
`disclose words with “different meanings in different contexts” ......... 11
`PETITIONERS FAIL TO REBUT PATENT OWNER’S ARGUMENT
`THAT SHORT-TERM SHARED KNOWLEDGE REQUIRES
`INTERPRETED UTTERANCES ................................................................. 15
`IV. CONCLUSION .............................................................................................. 21
`V. APPENDIX .................................................................................................... 23
`
`
`III.
`
`i
`
`
`
`I.
`
`INTRODUCTION
`The Board’s institution decision correctly rejected the petition’s arguments
`
`that Kennewick’s “temperature” and “flight” examples render obvious “one or
`
`more words having different meanings in different contexts.” Paper 7, 13-15.
`
`Having failed the first time, Petitioners now come back with new arguments and
`
`expert testimony regarding how a POSA would purportedly understand the claim
`
`limitation, including its component terms “meaning” and “context.” However,
`
`nothing Petitioners belatedly argue at the reply stage rehabilitates their flawed
`
`petition case, because the fact remains that the cited Kennewick examples do not
`
`involve words with different meanings in different contexts.
`
`Moreover, because the specification describes the claimed short-term shared
`
`knowledge as knowledge that facilitates shared assumptions and expectations
`
`between speakers, recognized speech that has not been interpreted cannot
`
`constitute such knowledge. It is undisputed that Kennewick’s “dialog history”
`
`contains nothing but tagged, recognized words which have not yet been
`
`interpreted. Petitioners’ reply (Paper 16, “Reply”) must now rely on a misreading
`
`of the ’681 specification, arguing its Session Input Accumulator performs no
`
`interpretation and is therefore equivalent to Kennewick’s speech recognition
`
`engine that generates the dialog history. This is simply not supported by the ’681
`
`patent or record evidence.
`
`1
`
`
`
`For these reasons, the Board should reject Petitioners’ grounds of challenge.
`
`II.
`
`PETITIONERS FAIL TO ESTABLISH THAT KENNEWICK
`RENDERS OBVIOUS “ONE OR MORE WORDS HAVING
`DIFFERENT MEANINGS IN DIFFERENT CONTEXTS”
`A.
`Petitioners’ new claim construction argument broadly construes
`the limitation to include any ambiguity in the user’s request
`The meaning of the limitation “one or more words that have different
`
`meanings in different contexts” is plain on its face. It contains no terms of art nor
`
`any ambiguity that would require the Board to construe this limitation. The
`
`limitation requires an utterance containing one or more words that in one context
`
`has one meaning and in a different context has a different meaning. Thus, the
`
`Board had no trouble rejecting Petitioners’ arguments that Kennewick’s
`
`“temperature” example involves different meanings of “temperature” or that its
`
`example of “flight one hundred and twenty too” involves different contexts. Paper
`
`7, 14-15.
`
`With hindsight of this deficiency in their petition, Petitioners for the first
`
`time now seek to construe the limitation so broadly as to salvage their invalidity
`
`challenge. Reply, 1 (“[T]he limitation ‘one or more words that have different
`
`meanings in different contexts’ must be considered in light of the specification.”);
`
`see also id., 2-6, 11. Yet beyond pointing to the “Portland” example, Petitioners
`
`never actually explain how they believe this term should be interpreted.
`
`Nevertheless, their own expert’s testimony, both direct and on cross-examination,
`
`2
`
`
`
`demonstrates the flaws in Petitioners unstated, and belated, construction as to the
`
`terms “meaning” and “context.”
`
`Petitioners never explain their interpretation of the “different meanings”
`
`portion of the limitation. However, they do cite their expert Dr. Smyth’s
`
`supplemental declaration for the proposition that “the conversation speech engine
`
`of the patent must establish ‘the intended meaning within the identified context’ in
`
`order to disambiguate an intent that the user had in speaking … so that the system
`
`can fulfill the user’s request.” Reply, 5 (citing EX1033, ¶¶7, 9). This is as close as
`
`Petitioners get to acknowledging the unstated premise underlying their obviousness
`
`theory: that the claim term “meaning” should not be given its plain and ordinary
`
`definition but rather should be broadened to encompass the intent behind the user’s
`
`request as a whole, not just the meaning of a spoken word. See also, e.g., Reply 7
`
`(“[T]he existence of a generic definition for a word in an utterance … does nothing
`
`to resolve the user’s intent and cannot override the concept of “different meanings’
`
`in the claim.”).
`
`Dr. Smyth is somewhat more direct. In his supplemental declaration he
`
`asserts that the claim term “meaning” does not refer to a word’s “generic
`
`definition” because that is not enough to “resolve the user’s intent.” EX1033, ¶11.
`
`During cross-examination, he doubled down on this claim construction theory,
`
`testifying that in his view, a POSA would interpret the claimed “meaning” “as not
`
`3
`
`
`
`the dictionary definition, but a more specific interpretation that it’s information
`
`that satisfies the user’s request.” EX2006, 13:9-14:3; see also id., 11:11-25
`
`(distinguishing claimed “meaning” from a dictionary definition); 20:8-12 (“[M]y
`
`interpretation of ‘meaning’ is that it certainly goes beyond the dictionary definition
`
`and needs to take into account specific information necessary to satisfy a user’s
`
`request.”).
`
`Yet despite applying this special understanding of “meaning” in his analysis,
`
`Dr. Smyth testified “I don’t recall” whether the ’681 patent itself explained that is
`
`how the term should be interpreted. Id., 20:13-19. What’s more, he asserted that
`
`“meaning” has a specific definition in the field of computer systems. Id., 20:19-
`
`21:1 (“The term meaning when you’re building a computer system has a more
`
`precise definition than a dictionary generic definition for a word like ‘temperature.’
`
`This would have been well known in the field at the time.”). However, Dr. Smyth
`
`could identify no corroborating evidence for this proposition other than a vague
`
`reference to an unidentified chapter of the Huang book that he admitted he did not
`
`cite in either declaration. Id., 21:2-19 (discussing EX1011). Thus, while
`
`Petitioners never admit it outright in their reply, their theory rests on applying a
`
`special definition of “meaning” that finds no support in the ’681 patent or any
`
`record evidence beyond conclusory, uncorroborated expert testimony.
`
`4
`
`
`
`With respect to the term “context,” Petitioners now seek to define it as “a
`
`domain that can be further built ‘based on user profiles, environmental profiles,
`
`historical profiles, or other information,’ allowing the system to disambiguate the
`
`words of an utterance and return an appropriate response.” Reply, 5-6. However,
`
`such a construction is at odds with their own expert’s view on the meaning of the
`
`term, not to mention similar to their district court construction that has already
`
`been rejected.
`
`While Petitioners construe context as “a domain that can be further built…”
`
`their expert Dr. Smyth takes a contrary view, testifying that context and domains
`
`are distinct from each other. For example, his supplemental declaration asserts
`
`“context may be associated with a domain”—not that context is a domain as
`
`Petitioners construe the term.1 Compare EX1033, ¶7 with Reply, 5-6. Dr. Smyth
`
`confirmed this distinction during cross-examination, testifying that in his view the
`
`’681 patent refers to “context and domain as two separate concepts.” EX2006,
`
`19:14-23. As the sole exhibit Petitioners cite for their “context” construction is Dr.
`
`Smyth’s contradictory testimony—with no citation to the ‘681 patent itself—their
`
`belated construction lacks even a shred of supporting evidence. See Reply, 5-6
`
`(citing EX1033, ¶7).
`
`
`1 All emphasis added unless otherwise noted.
`
`5
`
`
`
`As further reason to reject Petitioners’ “context “ construction, and the
`
`obviousness arguments that depend on it, their construction resembles Petitioners’
`
`proposed construction in the district court litigation which the court has already
`
`rejected. EX1032, 9 (“Defendants propose that the term means ‘subject domain(s)
`
`that apply to a user utterance.’”). Instead, the district court found it unnecessary to
`
`construe “context.” Id., 2. The Board should do the same here, especially since
`
`Petitioners never proposed construing it in the petition. What’s more, Petitioners’
`
`construction is also flawed, as Petitioners never explain what it means for a domain
`
`to be “further built” (e.g. “further built” from what?), and in any case, Petitioners’
`
`construction introduces ambiguity not found in the word “context” itself.
`
`Beyond the specific definitions of “meaning” and “context” underlying
`
`Petitioners’ theory, their interpretation of the larger term “one or more words that
`
`have different meanings in different contexts” remains unclear. However, it would
`
`appear that Petitioners wish to include any ambiguity in the user’s intent. Thus,
`
`Petitioners use the ’681 patent’s “Portland” example to conclude that the limitation
`
`“would have included words [that] absent context, would still be ambiguous.”
`
`Reply, 5. Petitioners also state that “temperature” is a word with different
`
`meanings in different contexts merely because “simply recognizing that
`
`temperature is a ‘measurement of warmth or coolness’ is not enough information
`
`to ‘disambiguate an intent that the user had in speaking the one or more words’ and
`
`6
`
`
`
`respond to the user.’” Reply, 7 (quoting another limitation in claim 1); see also id.
`
`(“[T]he existence of a generic definition for a word in an utterance, like
`
`‘temperature,’ does nothing to resolve the user’s intent and cannot override the
`
`concept of ‘different meanings’ in the claim.”).
`
`Petitioners arguments lack merit. It is undisputed that the ’681 patent’s
`
`“Portland” example presents a word that involves a different meaning depending
`
`on context. E.g., Reply, 2-3. The ’681 patent explains that the word “Portland”
`
`has multiple meanings, as it could at least mean Portland, Maine or Portland,
`
`Oregon. EX1001, 11:62-12:18. The system chooses which of the different
`
`meanings the user intends by looking at such things like the context of a previous
`
`question (e.g. whether the user previously asked about Portsmouth or Seattle). Id.
`
`But there is no reason to extend that example to encompass any ambiguity,
`
`including for a word that does not have multiple meanings like “temperature.” As
`
`explained in further detail below, “temperature” as used in Kennewick is not a
`
`word that has multiple meanings—instead, the ambiguity lies in what object the
`
`user intends to obtain the temperature of. This is beyond what the ’681 patent
`
`describes in its Portland example where the word “Portland” itself can refer to
`
`multiple cities and therefore have multiple meanings.
`
`Moreover, Petitioners essentially seek to use a different limitation in claim 1,
`
`“disambiguate an intent that the user had in speaking the one or more words,” to
`
`7
`
`
`
`render superfluous the limitation that the one or more words have different
`
`meanings in different contexts. But that makes little sense. Claim 1 requires that
`
`the one or more words have a specific kind of ambiguity—that they have “different
`
`meanings in different contexts”—and further requires that the system disambiguate
`
`the user’s intent. The latter limitation does not nullify the former limitation.
`
`To the extent Petitioners’ vague, conclusory arguments about the meaning of
`
`the limitation “one or more words that have different meanings in different
`
`contexts” can be discerned, it appears to rely on the newly identified interpretations
`
`of its constituent components “meaning” and “context.” Since these are entirely
`
`new arguments—relying on new evidence—that should have been made in the
`
`petition, they are improper and should be disregarded. See 37 C.F.R. § 42.23(b);
`
`PTAB Consolidated Trial Practice Guide November 2019 at 74 (“It is also
`
`improper for a reply to present new evidence … that could have presented in a
`
`prior filing, for example newly cited prior art reference intended to ‘gap-fill’ ...”);
`
`Unified Patents Inc. v. DivX, LLC, IPR2019-01379, Paper 52 at 23 (“We agree
`
`with Patent Owner that the Petition does not detail Petitioner’s newly proposed
`
`interpretation of this claim limitation. … Under these circumstances, we consider
`
`Petitioner’s new argument waived.”). Beyond that, they are unsupported by the
`
`evidence of record (including Petitioners’ own cited expert testimony) and
`
`logically flawed. The Board should reject them in favor of plain meaning.
`
`8
`
`
`
`Regardless, as discussed below, Petitioners have failed to establish Kennewick
`
`renders the claims obvious under any construction.
`
`B. Kennewick’s “temperature” example does not disclose a word
`with “different meanings in different contexts”
`As the Board properly recognized in its institution decision, “the word
`
`‘temperature’ has the same meaning in both of these examples [disclosed in
`
`Kennewick]. In both instances, the word ‘temperature’ refers to a measurement of
`
`warmth or coolness.” Paper 7, 14; see also EX2006, 23:13-24:3 (Petitioners’
`
`expert admitting same); EX1033, ¶7 (“temperature generally refers to a degree of
`
`hotness or coldness”). Accordingly, the cited portion of Kennewick fails to
`
`disclose a word having different meanings in different contexts. Paper 7, 14.
`
`Petitioners attempt to circumvent this plainly correct understanding by
`
`comparing the word “temperature” with the word “Portland,” arguing that, like
`
`“Portland,” “temperature” alone is “not enough information to ‘disambiguate an
`
`intent that the user had.’” Reply, 7. Yet the mere existence of ambiguity is
`
`insufficient to meet the limitation—what matters is whether the word has a
`
`different meaning in a different context. “Portland” has multiple meanings in that
`
`it can mean different cities—the ambiguity is with the meaning of the word. By
`
`contrast, Kennewick’s example of “temperature” refers to a property of some
`
`object or thing (e.g. body or weather), but the ambiguity is not with respect to the
`
`meaning of the word, but rather the target object or thing. EX2006, 25:19-26:3
`
`9
`
`
`
`(Dr. Smyth conceding “certainly a common interpretation of temperature is that
`
`it’s measuring … the hotness or coldness of a particular thing, or person, or
`
`location”). Indeed, Kennewick recognizes this distinction in that it states that
`
`“temperature” can “imply” a “context value of weather” or “a context for a
`
`measurement.” EX1003, ¶160. The target object or thing relates to the context for
`
`the temperature the user is interested in—yet “temperature” still maintains the
`
`same meaning.
`
`A simple example illustrates this difference. The question “What’s Helena’s
`
`temperature?” could be referring to the weather temperature in Helena, Montana or
`
`the body temperature of someone named Helena. But the system would need to
`
`disambiguate “Helena,” not “temperature,” because “Helena” is the word with
`
`different meanings. That is, “Helena” is the ambiguous word for which the user
`
`wants to know some specific, non-ambiguous property (i.e. temperature).
`
`The ’681 patent also supports this distinction between a word that has
`
`different meanings (like “Portland”) and a word that refers to a singular property of
`
`different objects (like “temperature”). For example, the ’681 patent provides an
`
`example where “traffic” may have different meanings. EX1001, 11:10-20. But the
`
`three different meanings the ’681 patent discloses are (1) “a Rock and Roll band,”
`
`(2) “a film directed by Steven Soderbergh,” and (3) “conditions on roads along a
`
`route to the airport.” Id. Thus, the ’681 patent teaches that “traffic” can have
`
`10
`
`
`
`“different meanings in different contexts” with respect to what the word refers to
`
`(such as a band, film, or road conditions)—but does not teach that “traffic” can
`
`have different meanings with respect to what route’s traffic the user is interested
`
`in. Thus, Petitioners’ argument that a word with one “generic definition,” like
`
`temperature or traffic (with respect to road conditions), has “different meanings”
`
`lacks any support in the ’681 patent and is nonsensical on its face. Reply, 7.
`
`Beyond the fact that Kennewick’s “temperature” example involves the same
`
`meaning of the word “temperature” regardless of context, the Patent Owner
`
`response also pointed out that Kennewick only teaches that “temperature” can
`
`imply a different context within different dialogs, and thus does not describe
`
`different meanings in different contexts for a single utterance. Paper 13 (“POR”),
`
`5. Petitioners’ reply has no response. See Reply, 6-8. Having failed to even
`
`attempt to respond, this further distinction between Kennewick and the ’681 claims
`
`is now unrebutted.
`
`For these reasons, Petitioners have failed to establish that Kennewick’s
`
`“temperature” example renders obvious “one or more words with different
`
`meanings in different contexts.”
`
`C. Kennewick’s “flight one hundred and twenty too” example does
`not disclose words with “different meanings in different contexts”
`As recognized by the Board, Kennewick’s “flight one hundred and twenty
`
`too” example “describes an ambiguity that exists within a single, previously-
`
`11
`
`
`
`identified context,” not “different meanings in different contexts.” Paper 7, 15.
`
`Having failed to provide evidence to the contrary in the petition, Petitioners now
`
`attempt to cure this deficiency with new expert testimony. Again, such evidence
`
`and corresponding arguments in the reply should be disregarded as improper,
`
`especially since Patent Owner cannot provide rebuttal testimony and is therefore
`
`prejudiced. See 37 C.F.R. § 42.23(b). Beyond that, Petitioners’ theory relies on
`
`hindsight-based conjecture about what Kennewick does not preclude, rather than
`
`what it actually describes.
`
`As an initial matter, Petitioners’ argument that “‘too’ and ‘two’ are
`
`effectively the same word” can be easily rejected. Reply, 8 n.3. They are
`
`homophones, which refers to different words having the same pronunciation.
`
`EX2006, 27:6-28:7 (Dr. Smyth conceding homophones are different words having
`
`the same pronunciation but different meanings), 29:1-3 (admitting “too” and “two”
`
`are homophones). The record contains no evidence to support Petitioners’
`
`assertion that this common understanding does not apply in a spoken language
`
`system. Reply, 8 n.3.
`
`Once “too” and “two” are properly understood to be different words, as
`
`Petitioners’ own expert conceded, the invalidity challenge falls apart. Kennewick
`
`teaches that “what about flight one hundred and twenty too?” can be interpreted as
`
`“flight 100 and flight 20 also,” “flight 100 and flight 22,” and “flight 122.”
`
`12
`
`
`
`EX1003, ¶163. However, Kennewick does not disclose that the identified phrase
`
`involves the same words with different meanings as required by the claims. The
`
`words “flight one hundred and twenty too” only has one meaning disclosed by
`
`Kennewick: “flight 100 and flight 20 also.” EX1003, ¶163. Moreover, even if the
`
`Board were to consider the phrase “flight one hundred and twenty two” (which
`
`does not appear in Kennewick), Petitioners fail to assert that “flight 122” and
`
`“flight 100 and flight 22” involve different contexts. Instead, Petitioners only
`
`argue that “flight 122” and “flight one hundred and twenty too” have different
`
`meanings in different contexts—yet as explained above, these are not the same
`
`words. See Reply, 9 (comparing “flight 122” with “‘flight 100’ and ‘flight 20’”),
`
`10 (comparing “flight 122” with “flight one hundred and twenty too”).
`
`Petitioners’ argument with respect to Kennewick’s “flight” example also
`
`fails to show “different contexts” as claimed. As explained above, the reply
`
`arguments rely on a new construction of “context” which the Board should reject.
`
`But even if the Board were to consider the “context” argument for the flight
`
`example it still fails because rather than discussing Kennewick’s actual disclosure,
`
`Petitioners and Dr. Smyth instead resort to speculating what different contexts may
`
`involve “flight 122” and “flight one hundred and twenty too.”
`
`For example, Dr. Smyth asserts that “a dialog during which the user
`
`previously reserved a seat on ‘flight 122’ would provide a context in which ‘flight
`
`13
`
`
`
`122’ exists” whereas “if the context of the dialog had included flight 100 and flight
`
`20 instead, the system would likely determine a different plausible intended
`
`meaning.” EX1033, ¶15; see also Reply, 10. But Dr. Smyth cites nothing in
`
`Kennewick that discloses such behavior, nor anything indicating that Kennewick
`
`considers these examples to be different contexts. Indeed, Dr. Smyth admitted
`
`during cross-examination that Kennewick does not disclose more than one flight
`
`information context. EX2006 18:9-19:2 (“Kennewick … does not explicitly state
`
`that.”). Rather, Dr. Gershman’s opinions are based on a reading of the reference
`
`that “[t]here’s nothing in Kennewick that rules that out.” Id. Thus, Dr.
`
`Gershman’s invalidity theory is based on a hindsight-driven premise that while
`
`Kennewick may not disclose the claimed subject matter, the reference doesn’t
`
`preclude it either. This is insufficient to prove obviousness, particularly where the
`
`petition asserts “Kennewick thus discloses receiving utterances that ‘include[] one
`
`or more words that have different meanings in different contexts.’” Pet., 17.
`
`Petitioners also take Dr. Gershman’s testimony out of context. Reply, 10.
`
`In testifying that “the flight number becomes part of that context of what they’re
`
`talking about,” Dr. Gershman was testifying with respect to a continuation of the
`
`’681 patent itself, not Kennewick. EX1031, 85:21-86:21; see also Reply 4 n.1
`
`(addressing relevance of U.S. Patent No. 9,015,049 to ’681 patent). The discussion
`
`simply had no relevance to the disclosure in Kennewick. Even then, Dr. Gershman
`
`14
`
`
`
`did not testify that a different flight number would involve a different context. He
`
`simply explained that based on the disclosure and claims of the ’049 patent a flight
`
`number would “become[] part of that context,” because the system needs to know
`
`“which flight we’re talking about.” EX1031, 85:12-14. But Dr. Gershman never
`
`testified that if a different flight number became a “part of that context,” it would
`
`constitute a different context under the ’681 patent, let alone that Kennewick says
`
`anything of the sort. For Petitioners to imply otherwise is nothing more than
`
`hindsight-based conjecture.
`
`For these reasons, Petitioners fail to establish that Kennewick’s “flight”
`
`example renders obvious “one or more words that have different meanings in
`
`different contexts.”
`
`III. PETITIONERS FAIL TO REBUT PATENT OWNER’S ARGUMENT
`THAT SHORT-TERM SHARED KNOWLEDGE REQUIRES
`INTERPRETED UTTERANCES
`As explained in the Patent Owner Response, “[c]laims must be read in view
`
`of the specification,” and the ’681 patent’s specification demonstrates that “short-
`
`term shared knowledge” refers to knowledge that allows speakers to share
`
`“assumptions and expectations” about the conversation. Phillips v. AWH Corp.,
`
`415 F.3d 1303, 1315 (Fed. Cir. 2005); POR, 12-18; EX1001, 1:22-30. This
`
`knowledge cannot be obtained without interpreting the user’s utterances. POR, 12-
`
`18. Ignoring these disclosures, Petitioners seek to improperly construe “short-term
`
`15
`
`
`
`shared knowledge” to mean any “details about what the user said” during the
`
`conversation. Reply, 15; see also Pet., 17-18. Yet Petitioners cite nothing in the
`
`specification supporting this interpretation, as their position is based on the
`
`mischaracterization that all the ’681 specification’s Session Input Accumulator
`
`(SIA) does is collect recognized words, not interpret them. Once the false
`
`equivalence between the Session Input Accumulator and Kennewick’s dialog
`
`history is rejected, Patent Owner is left without a case. It is undisputed that
`
`Kennewick’s dialog history does not contain interpreted utterances.
`
`Petitioners concede the short-term shared knowledge recited in the claims is
`
`described in the specification as being built by the Session Input Accumulator.
`
`See, e.g., Reply, 20 (“[T]he SIA … build[s] shared knowledge (including short-
`
`term knowledge) and provides it as an input to an intelligent hypothesis builder.”);
`
`21 (“The SIA … merely builds short-term knowledge by ‘accumulating’ various
`
`inputs.”). The dispute lies in whether this short-term shared knowledge requires
`
`interpretation or extends to mere recognized text. Id.
`
`Attempting to equate Kennewick’s dialog history and the data generated by
`
`the Session Input Accumulator described in the ’681 patent, the reply essentially
`
`reinterprets the ’681 patent by asserting that the Session Input Accumulator builds
`
`short-term shared knowledge by merely “accumulating” inputs without interpreting
`
`recognized text, Reply, 20-21. Petitioners go on to argue that is instead the
`
`16
`
`
`
`Intelligent Hypothesis Builder that interprets recognized text to build said
`
`assumptions and expectations (a tacit admission that assumptions and expectations
`
`do, in fact, require interpreted text). Reply, 21-22 n.7. However, Petitioners do
`
`not cite to anything in the ’681 patent that says the Session Input Accumulator
`
`merely tags and accumulates inputs. That the Intelligent Hypothesis Builder
`
`performs additional interpretation in no way dictates that the SIA performs no
`
`interpretation of its own as Petitioners contend.
`
`Indeed, the patent is quite clear what the SIA does: it accumulates inputs in
`
`order to “build shared knowledge models and more accurate adaptive responses.”
`
`EX1001, 13:39-47. And one of those inputs used to build shared knowledge is
`
`“recognition text”—whereas recognized text is the output of Kennewick’s speech
`
`recognition engine that Petitioners contend generates the dialog history. Compare
`
`EX1001, 13:47-49 with EX1003, ¶155; Pet., 17-18. While Petitioners try to equate
`
`the “conversation history” generated by the ’681 patent’s SIA and Kennewick’s
`
`dialog history, as explained in the POR, the ’681 patent lists the “conversation
`
`history” generated by the SIA itself as an example of shared assumptions and
`
`expectations, which necessitates that it is interpreted text, not merely recognized
`
`text. POR, 18; EX1001, 14:24-27. The patent also teaches that “shared knowledge
`
`may enable a user and voice user interface to share assumptions and expectations
`
`… that facilitate a cooperative conversation between human users and a system.”
`
`17
`
`
`
`EX1001, 1:22-30; see also id., 1:23-25 (“shared assumptions and expectations
`
`regarding various aspects of the conversation, such as … participant knowledge
`
`about the topic”). Petitioners argument that the SIA merely accumulates utterances
`
`rather than interpreting them is unfounded.
`
`As further evidence that the Session Input Accumulator is not limited to
`
`merely accumulating recognized text and other inputs as Petitioners allege, the
`
`’681 specification explains that the SIA populates the Intelligent Hypothesis
`
`Builder with “relevant information,” “identif[ies] information with long-term
`
`significance,” and provides Adaptive Response Builder with “relevant state and
`
`word usage information.” EX1001 13:51-58. Identification of relevant and
`
`significant information plainly requires interpretation beyond mere accumulation
`
`of recognized words, and is in no way analogous to the dialog history in
`
`Kennewick that consists of a mere list of tagged recognized words.
`
`Petitioners’ argument that “shared knowledge ‘may enable’ sharing of
`
`assumptions and expectation, with no requirement that assumptions or expectations
`
`… must be included in short-term shared knowledge” misreads the patent. Reply,
`
`20-21. The term “enable” does not refer to some data flow where shared
`
`knowledge is later processed to create shared assumptions and expectations—
`
`instead, as described in the patent and explained by Dr. Gershman, knowledge that
`
`is shared between speakers in a conversation enables the speakers to share
`
`18
`
`
`
`assumptions and expectations. EX1001, 1:23-25; EX2001, ¶¶45-47; POR, 15. Dr.
`
`Gershman provides an example where shared knowledge about tipping enables
`
`speakers to share an assumption that the tipper was at a sit-down restaurant.
`
`EX2001, ¶¶46-47.2
`
`Petitioners’ argument that short-term shared knowledge is built prior to
`
`interpretation of an utterance also finds no support in the language of the claims.
`
`Petitioners argue that claim 1’s step of identifying a context based on short-term
`
`shared knowledge occurs prior to interpretation of the utterance(“establishing the
`
`intended meaning”). Reply, 17. But this misreads the claims. Nothing in claim 1
`
`indicates that the short term knowledge about the current conversation extends to a
`
`mere list of recognized words. For example, claim 2 recites that the short-term
`
`shared knowledge includes populating a short-term context stack with information
`
`about the utterance. That is, the short-term shared knowledge recited in claim 2
`
`tracks contexts that arise during the conversation, not mere utterances. See also,
`
`
`2 Petitioners are incorrect that this example conflates “outside knowledge” with
`
`shared short-term knowledge. Reply, 22-23. The shared short-term knowledge
`
`would be the fact that the first person ate a burger and left a tip (conveyed through
`
`the conversation), which enables the shared short-term assumption that the first
`
`person ate a restaurant.
`
`19
`
`
`
`e.g., EX1001, 11:33-56 (explaining that “a context stack may track conversation
`
`topics and attempt to fit a current utterance into a most-recent context, next-most
`
`recent topic, etc.”). That claim 1 goes on to recite that the short-term shared
`
`knowledge is used to disambiguate the user’s intent in no way indicates that Claim
`
`1 states that short-term shared knowledge can be recognized utterances as
`
`Petitioners propose. The accumulation of short-term shared knowledge about the
`
`conversation (e.g., tracking context) and subsequent disambiguation of intent both
`
`require interpretation.
`
`Petitioners argue Dr. Gershman testified that tagged utterances may form
`
`short-term shared knowledge, but he did no such thing. Reply, 21. Rather, the
`
`cited testimony (and surrounding testimony Petitioners neglected to cite) show Dr.
`
`Gershman explaining that tagged utterances would be the “input” into the NLU
`
`processor that generates short-term knowledge, not the short-term knowledge
`
`itself. EX031, 80:22-82:19.
`
`Thus, the ’681 patent’s claims
Accessing this document will incur an additional charge of $.
After purchase, you can access this document again without charge.
Accept $ Charge
Still Working On It
This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.
Give it another minute or two to complete, and then try the refresh button.
A few More Minutes ... Still Working
It can take up to 5 minutes for us to download a document if the court servers are running slowly.
Thank you for your continued patience.
This document could not be displayed.
We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.
Your account does not support viewing this document.
You need a Paid Account to view this document. Click here to change your account type.
Your account does not support viewing this document.
Set your membership
status to view this document.
With a Docket Alarm membership, you'll
get a whole lot more, including:
- Up-to-date information for this case.
- Email alerts whenever there is an update.
- Full text search for other cases.
- Get email alerts whenever a new case matches your search.
One Moment Please
The filing “” is large (MB) and is being downloaded.
Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.
Your document is on its way!
If you do not receive the document in five minutes, contact support at support@docketalarm.com.
Sealed Document
We are unable to display this document, it may be under a court ordered seal.
If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.
Access Government Site
