`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`BloomReach, Inc.,
`Petitioner,
`
`v.
`
`Guada Technologies LLC,
`Patent Owner.
`
`
`Case IPR2019-01304
`Patent No. 7,231,379
`
`
`GUADA’S PRELIMINARY RESPONSE
`
`Dated: October 25, 2019
`
`Isaac Rabicoff
`Reg. No. 74,147
`RABICOFF LAW LLC
`Lead Counsel for Patent Owner
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`
`1. Introduction .......................................................................................................... 1
`
`2. Wesemann fails to teach key limitations of each of the challenged claims,
`leaving the board with no basis for institution (Grounds 1 and 2). ........................... 1
`
`2.1 Wesemann teaches how to automatically traverse connected nodes of telephone
`service systems, because these legacy systems cannot jump nodes. ......................... 1
`
`3. Fratkina also fails to teach the jumping limitation recited by each of the
`challenged claims, leaving the board with no basis for institution (Grounds 3 and
`4). 8
`
`3.1 Fratkina’s menu embodiments fail to teach the jumping limitation. ................... 8
`
`3.2 Bloomreach cannot combine the alleged “jumping” feature of
`autocontextualization to the menu embodiment. .....................................................11
`
`4. Conclusion .........................................................................................................17
`
`
`
`
`
`
`
`
`
`— ii —
`
`
`
`1.
`
`Introduction
`
`When evaluating BloomReach’s petition for inter partes
`
`review, the Board should consider these key points:
`
`• Wesemann (the first of two primary references) teaches
`how to automatically traverse connected nodes of telephone
`service systems, because these legacy systems cannot jump
`nodes;
`
`the menu embodiment of Fratkina (the second of two
`•
`primary references) also fails to teach the jumping limitation;
`and
`
`Bloomreach cannot combine Fratkina’s alleged
`•
`“jumping” feature of autocontextualization to its menu
`embodiment.
`BloomReach has therefore failed to show “that it is more likely
`
`than not that at least 1 of the claims challenged in the petition is
`
`unpatentable” and therefore institution of inter partes review must be
`
`denied. 35 U.S.C. § 324(a).
`
`2. Wesemann fails to teach key limitations of each of the
`challenged claims, leaving the board with no basis for
`institution (Grounds 1 and 2).
`
`2.1 Wesemann teaches how to automatically traverse
`connected nodes of telephone service systems, because
`these legacy systems cannot jump nodes.
`
`“Jumping” nodes—as undisputedly construed here—is contrary
`
`
`
`— 1 —
`
`
`
`to the teachings of Wesemann.1 Element 1(b)2 recites: identifying at
`
`least one node, other than the first node, that is not directly connected
`
`to the first node but is associated with the at least one keyword, and
`
`jumping to the at least one node. Ex. 1001, claim 1 (emphasis
`
`added) (also referred to as the “jumping limitation”).
`
`Wesemann provides a user interface for navigating legacy
`
`telephone service systems that can only receive dual tone multi-
`
`frequency signals as inputs. See Ex.1004, Abstract. These legacy
`
`telephone systems cannot jump to non-connected menu states (i.e.
`
`nodes):
`
`
`
`
`1 Bloomreach adopted Guada’s construction of “jumping”, citing the prosecution
`history and the Patent Owner Opposition to Motion to Dismiss. Presumably,
`Bloomreach thereby also adopts the construction of “Jumping to the At Least One
`Node” and “Jumping to the Vertex”, which makes clear that the subject doing the
`jumping is the “system”. See Ex. 1003, 18-19 (where these terms are construed as
`“the system jumping to the at least one node” and “the system jumping to the
`vertex”, respectively; emphasis added).
`2 Bloomreach acknowledges that all of the challenged claims possess an
`equivalent jumping limitation, whether jumping across nodes (independent claim
`1 with corresponding dependent claims 2-6) or jumping across vertices
`(independent claim 7). See Petition, 39 (Bloomreach opines that the jumping
`limitation of Claim 1(b) as equivalent to the same limitation in Claim 7(c)).
`— 2 —
`
`
`
`
`
`Wesemann teaches moving between connected menu states only
`because of the inherent limitations of telephone service systems
`
`“To cause the telephone service system to jump from a first menu
`state to a second menu state, the voice-enabled user interface
`generates and transmits output to the telephone service system that
`causes the telephone service system to traverse the one or more
`menu states it is jumping over, steps 536, 538. For example, if the
`telephone service system is in the menu state of business laptop
`sales 662 when the user speaks “home laptop sales,” then the
`voice-enabled user interface generates and transmits output to
`the telephone service system that causes the telephone service
`system to first transition to business computer sales 660, then to
`sales 630, then to home computer sales 650, and then finally to
`home laptop sales 652. It should be appreciated that all of the
`communications associated with jumping from one menu state to
`another menu state are conducted without the knowledge and efforts
`of the user, which is an improvement over the prior art.” (Ex. 1004,
`12:53-65; emphasis added.)
`
`
`As an example, when a user requests home laptop sales 652
`
`while in the business laptop sales 662 menu state, Wesemann teaches
`
`how to automatically transition between connected menu states
`
`without user interaction. In this way, Weisman’s user interface (which
`
`provides inputs to, but is inherently limited by, the telephone service
`
`system) generates the following series of connected menu state
`
`transitions in order to move from business laptop sales 662
`
`(highlighted yellow) to home laptop sales 652 (highlighted purple):
`
`
`
`
`
`— 3 —
`
`
`
`
`
`Ex. 1004, Fig. 6 (annotated above); see also id., 12:53-65.
`
`As shown in the flow diagram below, Wesemann’s user
`
`interface generates the connected menu states needed to transition
`
`from home laptop sales 652 to business laptop sales 662. These
`
`connected, transition menu states are determined at step 536
`
`(highlighted blue), and outputted to the telephone service system at
`
`step 538 (highlighted green):
`
`
`
`— 4 —
`
`
`
`
`
`Ex. 1004, Fig. 5 (annotated above); see also id., 12:53-65.
`
`Wesemann therefore addresses a problem in the art that
`
`requires transitioning between connected menu states (i.e. nodes) and
`
`not non-connected nodes. As such, Wesemann defines “jumping”
`
`between menu states differently than here: transitioning across
`
`multiple connected menu states without user interaction. See Ex.
`
`1004, 12:53-65 (noting that the user interface generates the connected
`
`transition menu states needed to reach the requested menu state, and
`
`that “[i]t should be appreciated that all of the communications
`
`associated with jumping from one menu state to another menu state
`
`are conducted without the knowledge and efforts of the user, which is
`— 5 —
`
`
`
`
`
`an improvement over the prior art.”).
`
`As such, when Bloomreach suggests that Wesemann teaches
`
`“jumping” nodes, that is “jumping” according to Wesemann, not
`
`according to the parties agreed construction of jumping here. What’s
`
`worse, Bloomreach cites a passage that mentions “jumping menu
`
`states” but cuts off the lines immediately following where Wesemann
`
`explains how its form of jumping is accomplished (i.e. by traversing
`
`across connected menu states). See Petition, 28 (citing Ex. 1004,
`
`12:25-42; but omitting 12:42-65, which explains the “process” that
`
`allows “jump[ing] from one menu state to another menu state. . .”.)
`
`In fact, Bloomreach’s jumping-related passage reveals the
`
`contradiction between the definitions of “jumping” from Wesemann
`
`and the challenged patent. Wesemann spares the user from entering
`
`input while the user interface transitions between connected menu
`
`states to reach the requested menu state, but does not allow for
`
`transitions between non-connected nodes: “the invention enables a
`
`user to jump from one menu state to another menu state without
`
`having to enter input for every ‘in between’ menu state.” Id.,
`
`12:27-30.
`
`Nor does Smyth (Bloomreach’s employed expert) provide
`
`
`
`— 6 —
`
`
`
`evidence—let alone even alleges—that the system jumps across
`
`unconnected nodes. He certainly fails to explain away the contrary
`
`teaching, discussed above, that requires the system to navigate
`
`through connected nodes. Rather, his declaration confirms, in two
`
`passages, that Wesemann only spares the user from needing to make
`
`multiple requests or interactions to navigate across multiple connected
`
`nodes.
`
`Smyth’s first passage merely shows that the transfer across
`
`nodes will be automatic for the user rather than a showing a system
`
`that can jump across non-connected nodes. See Ex. 1007, ¶ 55
`
`(“Although extension 123 is not directly connected to main menu state
`
`610, upon receiving the input ‘123,’ the system will ‘identify the user
`
`input as a valid input for another state of the menu hierarchy,’ (i.e.,
`
`identifying a node other than the first node) and the user will
`
`‘automatically be transferred’ by the system to extension 123,
`
`without forcing the user to traverse through the intermediate
`
`node, ‘Directory of Personnel’ 640 (i.e., jumping the user to the other
`
`node)”; emphasis added).
`
`Faring no better, in Smyth’s second passage Smyth uses his
`
`own words when he says that the “system will jump” across nodes,
`
`
`
`— 7 —
`
`
`
`but even there he acknowledges that it is the user that is saved from
`
`traversing intermediate steps rather the system avoiding those steps by
`
`jumping. See id., ¶ 56 (“the system will ‘jump’ from home laptop
`
`sales 652 to home computer support 646 without requiring the user
`
`to traverse intermediate steps”; emphasis added).
`
`Wesemann therefore teaches transitioning only between
`
`connected menus states (i.e. nodes) without user interactions and does
`
`not teach jumping between non-connected nodes. Wesemann
`
`therefore contradicts what the jumping limitation requires and renders
`
`no challenged claim obvious.
`
`3.
`
`Fratkina also fails to teach the jumping limitation recited by
`each of the challenged claims, leaving the board with no
`basis for institution (Grounds 3 and 4).
`
`3.1 Fratkina’s menu embodiments fail to teach the
`jumping limitation.
`
`Fratkina’s menu embodiment (Figures 10-12) involve
`
`movements across connected menu-related nodes that were created by
`
`a dialog designer. The dialog offers a limited set of menu options
`
`leading to connected nodes.
`
`The annotated figures below show parent nodes outlined in red
`
`and the directly-connected child nodes outlined in green.
`
`
`
`— 8 —
`
`
`
`DIALOG ”STATE
`
`uuasmws ammo
`DURING 0mm; mu USER
`
`mwzres {SELECTIONS}
`
`
`
`ITERAIICIN N+1
`
`HOW lN'UULI) YOU LIK:
`YOUR EGGS PREPARED?
`
`ITERATIG'N N+2
`
`
`
`
`SCRAM BLED
`
`PGACH ED
`
`BENEDICT
`
`1120
`
`
`
`
`
`
`Ex. 1006, Fig. 11 (annotated) (at the breakfast node: offering directly-
`
`connected eggs or pancakes; at the eggs node: offering directly
`
`connected scrambled, poached, or benedict); see also id., 26:46-60
`
`(describing how the dialog in Figure 11 leads the user through
`
`directly-connected, predetermined node options).
`
`
`
`— 9 —
`
`
`
`
`QUESTIONS OENERATEO
`DURING DIALOO AND USER
`ANSWERS (SELECTIONS)
`
`
`DIALOO STATE
`
`ITERATJON N
`
`ARE YOU ON ANY DIET?
`
`WHICH OF THE FOLLOWING
`WOULD YOU LIKE TO GET?
`
`creoleGool
`I
`
`node=poncokes;
`
`HOW WOULD YOU
`LIKE YOUR PANCAKE?
`
`
`
`F
`BREAK AST
`
`
`
`
`
`
`WITH SYRUP WITHOUT SYRUP
`
`ITERATION N+I
`
`
`IT (breakfast, confirmed)
`
`I
`
`
`
`Id., Fig. 12 (at the menu node: offering directly-connected breakfast,
`
`lunch or dinner; at the pancakes node: offering with or without syrup).
`
`
`
`
`
`— 10 —
`
`
`
`And Fratkina specifically teaches that dialogs always lead to
`
`node-to-node traversal (as shown in the next section, the only
`
`exception is autocontextualization, which is involved with an
`
`incompatible form of goal resolution). During such dialogs, the focus
`
`node always generates answer selections that lead to that node’s
`
`directly-connected children:
`
`Focus nodes present selections leading to directly-connect child
`nodes
`
`“A. Focus Nodes/Target Nodes
`
`Each goal is comprised of one or more "target" concept nodes
`paired with one or more focus nodes. The concept nodes represent
`the node or nodes within the taxonomy at which the dialog will
`ultimately terminate, while the focus node represents the starting
`location in the taxonomy. The focus node of the goal captures the
`current status of the goal resolution process. The focus node is the
`most detailed information available to the dialog engine at each
`point in the dialog. Each question posed to a user is generated
`from the focus node of the goal, while the answer selections
`offered to the user are the focus node's children.” (Ex. 1006,
`22:54-65
`
`
`3.2 Bloomreach cannot combine the alleged “jumping”
`feature of autocontextualization to the menu
`embodiment.
`
`For Fratkina’s menu embodiments, or any taxonomy with
`
`predetermined nodes with associated keywords (i.e. the only node
`
`
`
`— 11 —
`
`
`
`arrangement that could also satisfied Element 1(a)), a user’s response
`
`to a dialog prompt will return a confirmed node selection:
`
`When a user selects a predetermined menu option, the node selected
`is a confirmed node
`
`“[I]n 1120 when the user answers "scrambled" 55 in response to the
`question "How would you like your eggs prepared," the goal of the
`dialog proceeds from the "eggs" node to the "scrambled" node 1130.
`In this example, the nodes selected are confirmed nodes since
`they represent nodes whose relevance to the user's information
`need has 60 been established.” (Ex. 1006, 26:54-60.)
`
`
`And yet autocontextualization never returns confirmed nodes,
`
`making it unusable in a system where confirmed nodes with
`
`associated keywords are traversed:
`
`Autocontextualization never returns confirmed nodes, unlike the
`menu embodiments
`
`“All the text the user provides to the system is autocontextualized
`against the taxonomies in the knowledge map. This results in topic
`spotter nodes that represent the system's understanding of the user's
`input. Unlike confirmed nodes (discussed above), topic spotter
`nodes cannot automatically accepted [sic] as true since there is
`always the possibility that dialog engine 232 misunderstood the
`user's input.” (Ex. 1006, 29:5-11; emphasis added.)
`
`“Since it is assumed the underlying information need of the user
`does not change in the course of a single dialog, the dialog
`engine can always use confirmed nodes to advance goals.
`However sometimes other information cannot be fully trusted.
`
`
`
`— 12 —
`
`
`
`In particular, autocontextualization, being an automatic
`process, can make mistakes. Therefore it may not be safe to
`assume that correct concept tags have been extracted from the
`query.” (Ex. 1006, 33:49-56.)
`
`
`What’s more, Bloomreach fails to mention that the unique tags
`
`(the alleged “keywords”) associated with menu nodes created by a
`
`dialog designer are not the same tags as those created by
`
`autocontextualization3:
`
`Autocontextualization creates tags that cannot be the same tags
`associated with menu nodes
`
`“The taxonomy tag 60 also includes an attribution (not shown)
`which records whether the tag was created by a person [i.e. a dialog
`designer], or automatically by the system using
`autocontextualization.” (Ex. 1006, 14:31-34; citations omitted.)
`
`
`Smyth merely parrots Bloomreach’s analysis, and offers no
`
`evidence as to why the teachings of autocontextualization and the
`
`menu embodiments would collectively disclose the jumping
`
`
`3 Bloomreach suggests—without support and contrary to Fratkina’s teachings—
`that the tags (i.e. alleged “keywords) in the menu nodes (which are created by a
`dialog desire) are the same as the tags returned by autocontexualization. See
`Petition 58-59 (“a user may use “keyword or natural language” queries to request
`actions or make choices, and a “dialog engine” converts the user’s input into tags
`to be processed by the system using an ‘autocontextualization’ process.”)
`— 13 —
`
`
`
`
`
`limitation. Instead, Smyth concludes—without explanation—that
`
`autocontextualization would allow a user to jump over nodes like
`
`those in the menu embodiments: “The system is able to identify the
`
`non-adjacent node to which the user would like to go through
`
`“autocontexualization.” Ex. 1007, ¶ 84. But as shown above, the
`
`nodes returned by autocontextualization are topic spotter nodes, not
`
`the confirmed nodes that would result from a user selecting menu-type
`
`node options constructed by a dialog designer. Smyth leaps to the
`
`desired conclusion with no supportive analysis or reference citation.
`
`The Board should therefore reject it.
`
`3.3 Merely uttering the word “jumping” does not satisfy
`this limitation: autocontextualization does not teach
`the jumping limitation.
`
`“Jumping” appears in Fratkina 3 times in total, over 2 short
`
`passages related to autocontextualization. As discussed in the previous
`
`section, nodes returned by autocontextualization are not confirmed
`
`and require a series of follow-up questions to avoid misunderstanding.
`
`See § 3.2 above. So instead of teaching autocontextualization to jump
`
`from a breakfast node to a scrambled eggs node, this feature is used
`
`when the user’s input creates “uncertainty as to which branch applies
`
`or when multiple branches are relevant”:
`
`— 14 —
`
`
`
`
`
`
`
`Jumping with autocontextualization is only used to advance
`multiple subgoals when the system is uncertain about which subgoal
`branch applies
`
`“The process of choosing one path over another is sometimes
`difficult, particularly if there is uncertainty as to which branch
`applies or when multiple branches are relevant. It is a good
`practice during taxonomy building to avoid branches that are likely
`to be difficult to choose between, but it is not possible to avoid the
`issue entirely. If such a situation occurs, and the user chooses
`multiple nodes at a branching point, the dialog engine will
`create multiple subgoals for a goal. Subgoals of a goal are a
`mechanism to deal with uncertainty or multiplicity of the
`interaction. Resolution of each subgoal will be pursued in parallel
`by the dialog engine until the conditions are reached that resolve the
`parent goal of the subgoals. In fact, the system uses properties of
`subgoals in the process of goal resolution, so one subgoal is always
`created. Subgoals are identified by the node that represents the most
`detailed information. This node is referred to as focus of a subgoal.
`In the process of goal resolution, the focus is typically advanced
`along the edges of the taxonomy graph, but may jump to a node
`(or set of nodes) more than one edge away from the previous
`focus. Subgoals are resolved when their focus reaches the target set
`of concept nodes.” Ex. 1006, 27:22-43.
`
`
`This passage fails to teach or render obvious the jumping
`
`limitation for two key reasons.
`
`• This passage teaches how to resolve a goal when the user
`input suggests multiple branches in a hierarchy, where a the
`“focus” may advance multiple edges during goal resolution
`as the uncertainty is reduced (i.e. not where a user is
`presented with clear menu options, constructed by a dialog
`designer, and within an pre-identified branch such as
`— 15 —
`
`
`
`
`
`“breakfast”); and
`
`• Nowhere does this passage suggest that user input of
`predetermined menu-like keywords caused a jump—the
`opposite is true: if the user identified keywords associated
`with a non-adjacent node, the system would not be dealing
`with uncertainty over the correct branches or location in the
`hierarchy, i.e. the very problem addressed by
`autocontextualization according to Fratkina.
`
`Bloomreach cites the next jumping-by-autocontextualization
`
`passage without including the passage immediately above that
`
`explains how this autocontextualization actually operates.
`
`Autocontextualization never deals with user-provided keywords
`
`associated with non-adjacent nodes, since it creates nodes based on
`
`user input rather than “jumping” to nodes that it confirms. See also §
`
`3.2 above (distinguishing nodes returned by autocontextualization and
`
`the confirmed nodes used to navigate connected menu nodes.)
`
`Autocontextualized nodes are different from the confirmed nodes
`used to navigate a menu taxonomy
`
`“Since it is assumed the underlying information need of the user
`does not change in the course of a single dialog, the dialog
`engine can always use confirmed nodes to advance goals.
`However sometimes other information cannot be fully trusted.
`In particular, autocontextualization, being an automatic
`process, can make mistakes. Therefore it may not be safe to
`assume that correct concept tags have been extracted from the
`query.” (Ex. 1006, 33:49-56.)
`
`
`
`— 16 —
`
`
`
`“As mentioned above, dialog engine 232 generates followup
`questions to the user in the process of resolving active goals. The
`type of question to be generated can be specified by the dialog
`designer. The answers to these questions advance the state of the
`subgoal to a new location in the taxonomy—e.g. change the focus of
`the subgoal. Changing the focus of a subgoal may be by path
`traversal within the knowledge map (e.g., the focus may change
`from parent node to child node). Autocontextualization can be
`used to jump to a specific place in the taxonomy and the dialog
`designer can explicitly specify a place to jump to.” (Id., 34:32-
`42.)
`
`
`Yet again, Bloomreach fails to explain how nodes created by
`
`autocontextualization are used to navigate an existing hierarchy of
`
`nodes, such as meal options on a menu. That’s because Fratkina’s
`
`invention teaches no such thing.
`
`4.
`
`Conclusion
`
`Guada therefore requests that the Board deny institution on all
`
`grounds raised by BloomReach’s Petition.
`
`
`
`— 17 —
`
`
`
`Dated: October 25, 2019
`
`Respectfully submitted,
`
`/Isaac Rabicoff/
`Isaac Rabicoff
`Reg. No. 74,147
`RABICOFF LAW LLC
`73 W Monroe St
`Chicago, IL 60603
`773-669-4590
`isaac@rabilaw.com
`
`Counsel for Patent Owner
`Guada Technologies LLC
`
`
`
`
`
`— 18 —
`
`
`
`CERTIFICATION UNDER 37 C.F.R. § 42.24(D)
`
`This Petition complies with the requirements of 37 C.F.R. §
`
`42.24. As calculated by the word count feature of Microsoft Word for
`
`Office 365, it contains 2,946 words, excluding the words contained in
`
`the following: Table of Contents, Table of Authorities, List of
`
`Exhibits, Certification Under § 42.24(d), and Certificate of Service.
`
`
`
`
`/Isaac Rabicoff/
`Isaac Rabicoff, Reg. No. 74,147
`
`
`
`— 19 —
`
`
`
`
`
`
`
`CERTIFICATE OF SERVICE
`
`It is hereby certified that on October 25, 2019, a copy of the
`
`foregoing document was served via Electronic Mail upon the
`
`following:
`
`Dion M. Bregman
`Michael J. Lyons
`Ahren C. Hsu-Hoffman
`BloomReach-IPR@morganlewis.com
`
`
`
`/Isaac Rabicoff/
`Isaac Rabicoff, Reg. No. 74,147
`
`
`
`— 20 —
`
`