Amazon.com Inc v. VB Assets LLC

1. Case Identification

• Patent #: 8,073,681
• Filed: July 28, 2020
• Petitioner(s): 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 Services LLC)
• Patent Owner(s): VB Assets, LLC
• Challenged Claims: 1-42

2. Patent Overview

• Title: System and Method for a Cooperative Conversational Voice User Interface
• Brief Description: The ’681 patent discloses a conversational voice user interface that uses both short-term knowledge (e.g., current conversation history) and long-term knowledge (e.g., user profiles built over time) to determine context, understand a user's intent, and generate an appropriate response.

3. Grounds for Unpatentability

Ground 1: Obviousness over Kennewick - Claims 1-2, 9, 13-14, 21, 25-26, 33, 37, 39, and 41 are obvious over Kennewick.

• Prior Art Relied Upon: Kennewick (Application # 2004/0193420).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Kennewick alone teaches every element of the independent claims. Kennewick describes a speech-based environment that uses context, prior information, and user-specific profile data to understand queries. Specifically, it discloses accumulating short-term knowledge via a "dialog history" for the current conversation and accumulating long-term knowledge through a "user profile" containing a history of past interactions, unconstrained by the current session. Kennewick then uses this combined knowledge to identify context, determine the intended meaning of ambiguous utterances (e.g., "temperature" meaning weather or a measurement), and generate a grammatically adapted natural language response.
  • Key Aspects: Petitioner contended that Kennewick's disclosure is so complete that it effectively anticipates the claims, making an obviousness finding based on this single reference straightforward.

Ground 2: Obviousness over Kennewick and Huang - Claims 5, 8, 17, 20, 29, and 32 are obvious over Kennewick in view of Huang.

• Prior Art Relied Upon: Kennewick (Application # 2004/0193420), Huang ("Spoken Language Processing," 2001).
• Core Argument for this Ground:
  • Prior Art Mapping: Huang teaches a "plan-based" approach to spoken language understanding that classifies a conversation based on goals, participant roles, and information allocation. This plan-based model is used to improve the cooperative nature and inferential abilities of a dialog system. Petitioner argued that Huang's disclosure of identifying a "conversational goal" and classifying the conversation type maps directly to the limitations of claim 5.
  • Motivation to Combine: A POSITA would combine Huang's plan-based methodology with Kennewick's system to enhance its ability to handle complex, natural conversations. Both references share the goal of improving natural responses based on contextual information, and incorporating Huang’s techniques for modeling user intent would further Kennewick's goal of minimizing user interaction while improving response accuracy.
  • Expectation of Success: A POSITA would expect success in adding Huang's plan-based layer to Kennewick's context-rich system, as Kennewick already provides the necessary dialog history and user data to derive the conversational goals and roles taught by Huang.

Ground 3: Obviousness over Kennewick and Seneff - Claims 10, 22, 34, 38, 40, and 42 are obvious over Kennewick in view of Seneff.

• Prior Art Relied Upon: Kennewick (Application # 2004/0193420), Seneff ("Hypothesis Selection and Resolution in the Mercury Flight Reservation System," 2001).

• Core Argument for this Ground:

  • Prior Art Mapping: Seneff discloses a dialog strategy involving a hypothesis selection and verification process. It generates "multiple preliminary interpretations" of an utterance, scores them using confidence levels, and selects the "most plausible solution." If an initial interpretation is found to be incorrect, the system can select an alternative hypothesis from an N-best list. This process directly maps to the claim 10 limitations of generating multiple interpretations, selecting an initial one with the highest confidence, and updating short-term knowledge to select a next-highest confidence interpretation if the first is incorrect.
  • Motivation to Combine: A POSITA would be motivated to integrate Seneff's sophisticated hypothesis scoring and selection process into Kennewick's system to improve the accuracy of determining an utterance's meaning. While Kennewick teaches using context to determine meaning, Seneff provides a concrete, well-known method for scoring and ranking multiple potential meanings, which would be a natural and predictable improvement.
  • Expectation of Success: The combination was argued to be predictable because Kennewick's context-rich environment provides all the necessary inputs (dialog history, user profiles) that Seneff's scoring process uses to evaluate hypotheses.

• Additional Grounds: Petitioner asserted additional obviousness challenges, including combinations of Kennewick with Coffman (Patent 6,839,896) to teach multi-modal inputs/outputs, and with Mitsuyoshi (Application # 2003/0182123) to teach expiring short-term knowledge from a context stack after a psychologically appropriate amount of time.

4. Key Claim Construction Positions

• The Petitioner argued for a specific construction of "implicit hypothesis" (recited in dependent claims 38, 40, and 42). Based on the ’681 patent's specification, this term was construed to mean "a potential intent of a user utterance generated by filling in missing information using contextual data." This construction is central to applying prior art that infers missing information to resolve user queries.

5. Arguments Regarding Discretionary Denial

• Petitioner argued that discretionary denial under §314(a) or §325(d) would be inappropriate. It was argued that while the co-pending district court litigation alleges infringement of claim 41, this petition challenges a much broader set of claims (1-42), making an IPR an efficient alternative for resolving their patentability. Furthermore, regarding §325(d), Petitioner asserted that although the Kennewick reference was cited in an Information Disclosure Statement during prosecution, it was never substantively reviewed by the examiner, as it was one of over 370 references submitted. Therefore, Petitioner argued the Becton factors weigh against discretionary denial.

6. Relief Requested

• Petitioner requests institution of inter partes review and cancellation of claims 1-42 of Patent 8,073,681 as unpatentable.