Samsung Electronics Co Ltd v. VB Assets LLC

1. Case Identification

• Case #: IPR2025-00867
• Patent #: 7,818,176
• Filed: April 16, 2025
• Petitioner(s): Samsung Electronics Co. Ltd. and Samsung Electronics America, Inc.
• Patent Owner(s): VB Assets, LLC
• Challenged Claims: 1-52

2. Patent Overview

• Title: Selecting and Presenting Advertisements Based on Natural Language Processing
• Brief Description: The ’176 patent discloses a method and system for presenting targeted advertisements in response to voice-based user input. The system uses an Automatic Speech Recognition (ASR) engine to generate a preliminary interpretation of a user's utterance, which is then passed to a conversational language processor that establishes context to select and present a relevant advertisement.

3. Grounds for Unpatentability

Ground 1: Obviousness over Taschereau-Hansen Combination - Claims 1-2, 9-11, 27-28, and 35-37 are obvious over Taschereau in view of Hansen.

• Prior Art Relied Upon: Taschereau (WO 2004/055781) and Hansen (Patent 5,640,490).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Taschereau taught a telephone directory assistance system that provided targeted advertisements based on user voice requests (e.g., for a specific business), thus disclosing the overall framework of the independent claims. However, Taschereau was allegedly silent on the specific implementation of its ASR system. Petitioner asserted that Hansen supplied these missing details by teaching a well-known speech recognition process that identifies phonemes in an audio signal and uses a phonetic dictionary with linguistic rules (an "acoustic grammar") to translate the phonemes into syllables and words, thereby generating a preliminary interpretation of the utterance.
  • Motivation to Combine: A POSITA would combine Hansen's detailed ASR implementation with Taschereau's advertising system to provide the necessary technical details for Taschereau's ASR. Taschereau itself expressed a desire to improve its ASR system, providing an explicit reason to look to known speech recognition techniques like those in Hansen.
  • Expectation of Success: A POSITA would have a reasonable expectation of success because the combination merely integrated a known, well-understood component (Hansen’s ASR) into a similar, known system (Taschereau's voice-driven ad platform) operating in the same technical field.

Ground 2: Obviousness over Taschereau-Hansen-Franco Combination - Claims 12-18, 25-26, 38-44, and 51-52 are obvious over Taschereau and Hansen in view of Franco.

• Prior Art Relied Upon: Taschereau (WO 2004/055781), Hansen (Patent 5,640,490), and Franco (Application # 2010/0125458).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground built upon the Taschereau-Hansen combination by adding Franco to teach the claimed feature of detecting a misrecognition and reinterpreting the utterance. Petitioner contended that Franco described methods for correcting speech recognition errors, such as when a user repeats an utterance after an initial incorrect recognition. Franco taught combining the acoustic evidence from the initial and subsequent utterances to produce a more accurate result, thereby disclosing an "adaptive misrecognition engine" that reinterprets words in response to a "predetermined event" (e.g., a repeated utterance).
  • Motivation to Combine: Taschereau acknowledged that ASR accuracy decreases as grammars become larger, providing a clear motivation to incorporate an error-correction solution. Franco addressed this exact problem, stating a "need in the art" for such correction methods and suggesting its teachings were broadly applicable to systems like the one disclosed by Taschereau and Hansen.
  • Expectation of Success: The combination was argued to be predictable, as Franco's error correction module was described as a subsystem that could be coupled to a processor via standard communication channels, which are inherent in the telephone-based system of Taschereau.

Ground 4: Obviousness over Ramer-Cross Combination - Claims 1-52 are obvious over Ramer in view of Cross.

• Prior Art Relied Upon: Ramer (Patent 7,577,665) and Cross (Patent 8,332,218).

• Core Argument for this Ground:

  • Prior Art Mapping: As an alternative to the Taschereau-based grounds, Petitioner argued that Ramer taught a mobile internet search platform that used speech recognition to deliver targeted advertisements based on user queries and context (e.g., location, user history, shopping habits). Like Taschereau, Ramer allegedly lacked specific details of its ASR implementation. Petitioner asserted that Cross supplied these details by teaching a context-based ASR engine that uses a language-specific acoustic model to infer phonemes from speech, map them to words in a lexicon, and use contextual analysis (e.g., recent browsing history) to refine recognition.
  • Motivation to Combine: A POSITA implementing Ramer's system would have been motivated to look to known ASR technologies like Cross for the implementation details. Ramer itself suggested the use of "known voice recognition techniques." Cross provided a suitable, well-known, context-based ASR system that was commercially available and could be readily modified.
  • Expectation of Success: The result of the combination would have been predictable, as it involved the simple substitution of a known element (Ramer's generic ASR) with another well-known and more detailed element (Cross's context-based ASR).

• Additional Grounds: Petitioner asserted an additional obviousness challenge (Ground 3) based on the combination of Taschereau, Hansen, and Jones (Application # 2007/0174258) to teach the claims related to tracking user interaction with advertisements and updating user models.

4. Key Claim Construction Positions

• Petitioner argued that the Board should apply claim constructions from a related district court case (VB Assets, LLC v. Amazon.com Servs. LLC) and a prior IPR proceeding (Amazon.com, Inc. v. VB Assets, LLC, IPR2020-01390), asserting the challenged claims are unpatentable under any of these constructions. Key proposed constructions include:
  • "model": "an approximation, representation, or idealization of selected aspects of the structure, behavior, operation, or other characteristics of a real-world process, concept, or system."
  • "acoustic grammar": "grammar of phonotactic rules of the English language that maps phonemes to syllables."
  • "speech recognition engine": "software or hardware that recognizes the words or phrases in the natural language utterance."

5. Arguments Regarding Discretionary Denial

• §314(a) (Fintiv): Petitioner stipulated that, if the IPR is instituted, it will not pursue in the related district court proceeding any ground that it raised or reasonably could have raised in the IPR. This stipulation was offered to simplify the Board's Fintiv analysis and weigh against discretionary denial.
• §325(d): Petitioner argued that discretionary denial under §325(d) is inappropriate because the prior art and arguments presented in the petition are not the same or substantially the same as art previously considered by the USPTO during prosecution or in the prior IPR2020-01390. Petitioner contended that the petition presents a compelling case of unpatentability that warrants consideration.

6. Relief Requested

• Petitioner requests institution of an inter partes review and cancellation of claims 1-52 of the ’176 patent as unpatentable.