PTAB

IPR2025-00866

Samsung Electronics Co Ltd v. VB Assets LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Cooperative Conversational Voice User Interface
  • Brief Description: The ’681 patent discloses a voice user interface that interprets user utterances containing ambiguous words. The system uses a conversational speech engine that leverages both short-term shared knowledge (from the current conversation) and long-term shared knowledge (from past conversations) to identify a context, determine the user's intended meaning, and generate an adapted response.

3. Grounds for Unpatentability

Ground 1: Obviousness over SmartKom and Kobsa - Claims 1-9, 11-21, 23-33, 35-36 are obvious over SmartKom in view of Kobsa.

  • Prior Art Relied Upon: SmartKom ("SmartKom: Foundations of Multimodal Dialogue Systems," 2006) and Kobsa ("User Models in Dialog Systems," 1989).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that SmartKom, a comprehensive multimodal dialogue system, discloses nearly all limitations of the challenged claims. SmartKom’s system processes user utterances, including those with ambiguous words (a concept known as "sense ambiguity"), using various context models to determine user intent. It explicitly discloses using "shared knowledge services" that correspond to the claimed short-term knowledge (e.g., a "dialogue model" for the current conversation) and long-term knowledge (e.g., a "domain model"). Petitioner asserted that Kobsa, a foundational survey on user modeling, provides the necessary details for implementing the "user model" component mentioned but not fully detailed in SmartKom. Kobsa teaches incrementally constructing, storing, and updating persistent, long-term user models based on interactions over time, directly corresponding to the ’681 patent's "long-term shared knowledge" derived from past conversations.
    • Motivation to Combine (for §103 grounds): A POSITA would combine these references because SmartKom explicitly suggests the use of a "user model" for personalization but provides limited implementation details. A POSITA would be motivated to consult a well-known, coherent survey like Kobsa to implement this feature, thereby improving the system's ability to provide user-specific and tailored responses. The fact that an editor of SmartKom was also a co-editor of Kobsa would have further directed a POSITA to this combination.
    • Expectation of Success (for §103 grounds): A POSITA would have had a reasonable expectation of success because SmartKom utilizes a standard software architecture and Kobsa’s user models are described as software and storage constructs that could be predictably integrated.

Ground 2: Obviousness over Barbara and Ross - Claims 1, 5-8, 10-13, 17-20, 22-25, 29-32, 34-36 are obvious over Barbara in view of Ross.

  • Prior Art Relied Upon: Barbara (Application # 2004/0101198) and Ross (Application # 2002/0173960).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner contended that Barbara discloses a voice interface system that interprets user intent by evaluating utterances against a knowledge base and an information database. Barbara explicitly teaches resolving sense ambiguity (e.g., determining if "Toto" refers to a pet or a restaurant) by using context captured from current interactions (short-term knowledge) and user information built up over time (long-term knowledge). However, Barbara lacks detail on the server architecture for processing these utterances. Petitioner argued that Ross supplies this missing detail, disclosing a robust "conversation manager" architecture that functions as the claimed "conversational speech engine." Ross teaches processing spoken utterances, using a "conversational record" to track dialogue history, and a reasoning facility to generate responses.
    • Motivation to Combine (for §103 grounds): A POSITA reviewing Barbara’s system would recognize its lack of architectural detail for the server-side processing and response generation. This would motivate a search for a reference like Ross, which provides a detailed conversation manager architecture. Combining Ross would provide Barbara's system with flexible response generation capabilities and an efficient method for managing dialogue history, as Ross teaches purging conversational records when no longer relevant to active goals.
    • Expectation of Success (for §103 grounds): Success would be predictable, as the combination involves implementing Ross's known software architecture and data management techniques into Barbara's known voice interface system.
  • Additional Grounds: Petitioner asserted additional obviousness challenges based on the Barbara and Ross combination, further modified by O'Neill (Application # 2005/0278180) to add teachings on confirming user intent, and by Franco (Application # 2010/0125458) to add teachings on automated error correction for incorrect interpretations.

4. Key Claim Construction Positions

  • Petitioner noted that in related district court litigation, the parties agreed that the term "speech recognition engine" means "software or hardware that recognizes the words or phrases in the natural language utterance."
  • Petitioner applied this construction and stated that no other claim terms require construction to resolve the patentability issues presented in the petition.

5. Arguments Regarding Discretionary Denial

  • To address potential discretionary denial under Fintiv, Petitioner stipulated that if the IPR is instituted, it will not pursue in the related district court proceeding any invalidity ground that it raised or reasonably could have raised in the IPR. This is known as a Sotera stipulation, which weighs strongly against discretionary denial.

6. Relief Requested

  • Petitioner requests the institution of an inter partes review and the cancellation of claims 1-36 of the ’681 patent as unpatentable under 35 U.S.C. §103.