PTAB

IPR2024-00752

Google LLC v. Dialect LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Mobile Systems and Methods of Supporting Natural Language Human-Machine Interactions
  • Brief Description: The ’652 patent relates to a system for processing natural language commands using a speech interface. The system is designed to determine user commands when recognized words alone are insufficient by using context information from prior utterances, which can be synchronized across multiple devices.

3. Grounds for Unpatentability

Ground 1: Obviousness over Coffman - Claims 1-9, 14-18, and 20-26 are obvious over Coffman.

  • Prior Art Relied Upon: Coffman (International Publication No. WO 00/20962).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Coffman, filed nearly six years before the ’652 patent, discloses a remarkably similar conversational computing system that renders the challenged claims obvious. Petitioner asserted that Coffman teaches all limitations of independent claim 1. Specifically, Coffman’s system processes multi-modal inputs where speech alone is insufficient, requiring additional context to disambiguate commands. It discloses generating a "first context stack" on a "first device" that accumulates context information corresponding to prior utterances. Coffman further teaches a distributed system with a master-slave topology where the context stack of a master device is synchronized with the context stack of a slave device, with the master’s context being updated based on information from the slave's stack. Finally, Coffman’s system determines a user’s command based on both the recognized words of an utterance and this updated, synchronized context information. Petitioner mapped limitations of dependent claims to Coffman’s teachings on prompting for user confirmation, using rank scores to identify matching context entries, and associating commands with specific domain agents.
    • Motivation to Combine (for §103 grounds): This ground is based on a single reference.
    • Expectation of Success (for §103 grounds): This ground is based on a single reference.

Ground 2: Obviousness over Coffman and Kennewick - Claims 10-13, 19, and 27 are obvious over Coffman in view of Kennewick.

  • Prior Art Relied Upon: Coffman (International Publication No. WO 00/20962) and Kennewick (Application # 2004/0044516).

  • Core Argument for this Ground:

    • Prior Art Mapping: Petitioner argued that Coffman’s conversational system provides the base for the invention, and Kennewick supplies the specific features recited in these additional claims. For claim 10, which requires a response comprising an "aggregation of one or more responses," Kennewick teaches domain agents that generate their own queries to local or remote sources and then gather and combine the results into a "single best result." For claim 11, requiring the system to obtain and use license agreement information, Kennewick discloses an "agent manager" that verifies software use is within allowed license terms and conditions before allowing access. For claims 12 and 13, which recite using a "cognitive model," Petitioner argued both references teach this concept. Coffman collects user information like preferences and query phrasing, while Kennewick discloses a more detailed "personal profile" containing a user's interaction history, which is explicitly used to improve speech recognition. Kennewick further teaches using profile information from a plurality of users (a "second cognitive model") to enhance recognition accuracy.
    • Motivation to Combine (for §103 grounds): Petitioner asserted a person of ordinary skill in the art (POSITA) would combine these analogous references to achieve predictable benefits. A POSITA would incorporate Kennewick’s response aggregation to provide more complete and useful answers in Coffman's system. Market forces would motivate adding Kennewick’s license management to Coffman’s application platform to ensure proper use and increase platform value. Finally, a POSITA would be expressly motivated by Kennewick’s teachings to use detailed user profiles (cognitive models) to improve the speech recognition accuracy of Coffman’s system.
    • Expectation of Success (for §103 grounds): A POSITA would have an expectation of success because the combination involved the straightforward application of known techniques (response aggregation, license management, user profiling) to improve a similar, agent-based conversational system without undue experimentation.

  • Additional Grounds: Petitioner asserted an additional obviousness challenge against claims 13 and 19 based on the combination of Coffman, Kennewick, and Lee (Application # 2002/0087315). Lee was cited to provide further confirmation for using data compiled from multiple users' histories (a "second cognitive model") to improve the accuracy of word recognition.

4. Arguments Regarding Discretionary Denial

  • §314(a) (Fintiv Factors): Petitioner argued against discretionary denial, stating that the parallel district court litigation is in its very early stages. No trial date has been set, no claim construction has occurred, and discovery has not commenced. Petitioner stated its intent to file a motion to stay the litigation pending the outcome of the inter partes review (IPR).
  • §325(d) (Same or Substantially the Same Art): Petitioner argued that denial would be inappropriate because the examiner materially erred during prosecution of the ’652 patent. Although Coffman and Kennewick were cited in a large Information Disclosure Statement (IDS), the examiner never substantively discussed their teachings. The examiner’s reasons for allowance focused on the "synchronization of context stacks/sets of distinct devices," a feature Petitioner contended is explicitly taught in Coffman and was therefore overlooked.

5. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-27 of Patent 8,849,652 as unpatentable.