PTAB

IPR2015-00343

Google Inc. v. Network-1 Technologies, Inc.

Log In

1. Case Identification

2. Patent Overview

  • Title: Algorithms for Automatic Content Recognition
  • Brief Description: The ’179 patent describes systems for identifying an unknown media work by generating a "compact electronic representation" of it and using a "non-exhaustive neighbor search" to find a match in a database. Based on the identified work, a corresponding action is determined and associated with the work.

3. Grounds for Unpatentability

Ground 1: Claims 1-3, 6-14, 18-19, 21-26, and 30-37 are obvious over Levy in view of Arya under 35 U.S.C. §103.

  • Prior Art Relied Upon: Levy (Patent 6,505,160) and Arya (a 1998 publication titled "An Optimal Algorithm for Approximate Nearest Neighbor Searching in Fixed Dimensions").
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Levy discloses a base system for identifying media works and performing related actions. Levy teaches extracting a "fingerprint" from a media signal, sending it to a server that maps the identifier to an action in a database, and executing that action (e.g., a transaction or displaying information). However, Levy's system implies an exact match search. Arya discloses a non-exhaustive, approximate nearest neighbor search algorithm designed for efficient similarity searching in multimedia databases, which addresses the limitations of an exact-match system.
    • Motivation to Combine: A Person of Ordinary Skill in the Art (POSITA) would combine Levy with Arya to improve Levy's system. Petitioner asserted that a POSITA would understand that fingerprints generated from different sources of the same media can have slight discrepancies. An approximate neighbor search, as taught by Arya, is necessary to identify near matches, making the system more robust. Arya explicitly states its algorithm is useful for pattern recognition in multimedia databases, providing a clear motivation.
    • Expectation of Success: A POSITA would have a reasonable expectation of success in combining these references, as Arya's algorithm was specifically designed to solve the problem of efficiently finding approximate matches in the exact type of data sets used by Levy.

Ground 2: Claims 1-3, 6, 8-14, 19, 21-26, 30-31, and 34-37 are anticipated by Conwell under 35 U.S.C. §102.

  • Prior Art Relied Upon: Conwell (Patent 6,970,886).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner contended that Conwell anticipates every element of the challenged claims. Conwell teaches a system for directing users to websites related to digital content. It discloses generating an identifier from content via a hashing algorithm (the "compact electronic representation"), looking up the identifier in a database to find a corresponding URL (the "action"), and returning the URL to the user. Petitioner argued Conwell's search is a "neighbor search" because its hashing algorithm maps similar but non-identical inputs to the same hash output, thereby finding "neighbor" results. The search is "non-exhaustive" because it involves a direct lookup in a sorted table using an index, not a brute-force comparison of all possible entries.

Ground 3: Claims 1-3, 8-14, 19, 21-22, 25, 27, 29-31, and 34-37 are anticipated by Wood under 35 U.S.C. §102.

  • Prior Art Relied Upon: Wood (Patent 7,743,092).

  • Core Argument for this Ground:

    • Prior Art Mapping: Petitioner argued Wood discloses all claim elements in a system for identifying a song hummed by a user. Wood extracts musical features ("compact electronic representation") from a vocalized specimen, compares the resulting pattern to a library, and identifies a match. This comparison constitutes a "neighbor search" as it calculates a distance score between the specimen and library entries to find a candidate. The search is "non-exhaustive" because Wood teaches it can be sped up by comparing only memorable portions of a song (e.g., the chorus) or ignoring periods of silence. After identification, Wood discloses performing actions such as transmitting song information for display or offering a download for purchase.

  • Additional Grounds: Petitioner asserted an additional obviousness challenge (Ground 4) against claims 1-3, 8, 10-14, 18-19, 21-27, 29, 31, and 34-37 over Ghias (Patent 5,874,686) in view of Philyaw (Patent 6,098,106). The argument was that a POSITA would have been motivated to combine Ghias's method for identifying content based on intrinsic features (hummed melodies) with Philyaw's system for linking identified content to actions (like advertisements) to create a more efficient system that does not require embedding identifiers into content.

4. Key Claim Construction Positions

  • "compact electronic representation" / "extracted features": Petitioner proposed these terms, which appear in the independent claims, should be construed synonymously to mean "a quantitative representation of the features of a media work, or a subset of the features of a media work." This broad construction was argued to be supported by the specification and the understanding of a POSITA, encompassing various feature extraction methods.
  • "non-exhaustive...search": Petitioner proposed construing this term to mean "a search that locates a match without conducting a brute force comparison of all possible matches, and all data within all possible matches." This construction was central to mapping prior art that used efficient lookups or partial comparisons (e.g., hash tables, searching only a song's chorus) instead of exhaustive, brute-force methods.
  • "neighbor search": Petitioner proposed this term be construed as "identifying a close, but not necessarily exact, match." This construction was key to applying prior art that accounted for minor variations in input signals, such as by using distance scores or approximate matching algorithms.

5. Relief Requested

  • Petitioner requests the institution of an inter partes review (IPR) and cancellation of claims 1-3, 6-14, 18-19, 21-27, and 29-37 of the ’179 patent as unpatentable.