Microsoft Corp v. Improved Search LLC

1. Case Identification

• Case #: IPR2017-01614
• Patent #: 6,604,101
• Filed: June 23, 2017
• Petitioner(s): Microsoft Corporation
• Patent Owner(s): Improved Search, LLC
• Challenged Claims: 1-28

2. Patent Overview

• Title: Method and System for Translingual Translation of Query and Search and Retrieval of Multilingual Information on a Computer Network
• Brief Description: The ’101 patent discloses a system for improving multilingual search results by first performing "dialectal standardization" on a user's query. This process involves replacing a query term with a more commonly known word in the source language before translating the standardized term into a target language for searching a computer network, such as the World Wide Web.

3. Grounds for Unpatentability

Ground 1: Claims 1-6, 12-17, 22, and 27-28 are obvious over Williamowski in view of Sullivan.

• Prior Art Relied Upon: Williamowski (Patent 6,381,598) and Sullivan (Patent 5,956,711).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Williamowski taught the foundational elements of the challenged claims, including a cross-lingual information retrieval system that receives a query in a first language, extracts keywords, translates them into a second language, and performs a search on the World Wide Web. However, Williamowski's "linguistic preprocessing" (e.g., stemming) did not explicitly teach dialectal standardization. Sullivan was argued to supply this missing element by disclosing a system that standardizes query terms by mapping keywords from different styles and dialects (e.g., "color" vs. "colour," "couch" vs. "settee") to a single, standard term to improve search accuracy. The combination of Williamowski's search framework with Sullivan's dialectal standardization renders the independent claims obvious.
  • Motivation to Combine: A person of ordinary skill in the art (POSITA) would combine these references to improve the functionality of Williamowski's system. Both patents address the same field of computerized information retrieval. A POSITA would recognize the problem of dialectal variations in a global search system like Williamowski's and would have looked to known solutions like Sullivan's to replace ambiguous or regional terms with standard ones, thereby improving search relevance and reliability. Williamowski itself suggested that its invention could be practiced with "other configurations of information retrieval systems."
  • Expectation of Success: A POSITA would have a reasonable expectation of success in this combination because it involved integrating a known query normalization technique (from Sullivan) into an existing query preprocessing pipeline (in Williamowski). This was presented as a predictable combination of known elements to achieve a predictable result.

Ground 2: Claims 1-7, 12-17, and 22-28 are obvious over Williamowski in view of Sullivan, and in further view of Poznanski.

• Prior Art Relied Upon: Williamowski (Patent 6,381,598), Sullivan (Patent 5,956,711), and Poznanski (Patent 6,360,196).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground builds on the first, with Poznanski added to explicitly teach limitations related to translating search results back into the source language, as recited in dependent claims 6, 7, and 17. While Williamowski taught displaying results, Petitioner argued that Poznanski more clearly disclosed a "multilingual glosser" that translates search results obtained in the target language back to the source language for the user. Poznanski also taught that this translation could be performed on all or part of the search results, satisfying the "selectively choose to translate" limitation.
  • Motivation to Combine: A POSITA would be motivated to add Poznanski's functionality to the Williamowski/Sullivan system to create a more complete and user-friendly product. Providing the ability to translate foreign-language search results back into the user's native tongue is a logical and necessary feature for a cross-lingual search tool to be effective. This would increase the system's utility and efficiency by allowing a user to understand and parse results without requiring external translation tools.
  • Expectation of Success: The combination was argued to be predictable. Language translation was a well-known technique, and integrating Poznanski's result-translation feature into the proposed system was a straightforward extension of functionality.

Ground 3: Claims 8-11 and 18-21 are obvious over Williamowski, Sullivan, and Poznanski, in further view of Redpath.

• Prior Art Relied Upon: Williamowski (Patent 6,381,598), Sullivan (Patent 5,956,711), Poznanski (Patent 6,360,196), and Redpath (Patent 6,347,316).
• Core Argument for this Ground:
  • Prior Art Mapping: This four-way combination addresses claims requiring the user to be able to select between different quality levels of translation (e.g., "machine translations" vs. "well-translated pages") and requiring the server to maintain a collection of these pages. Petitioner asserted that Redpath taught these specific features. Redpath disclosed a system that provided both raw machine translations and human-perfected "perfect files," allowing a user to select between them based on their need for speed versus accuracy. Redpath also disclosed maintaining these different translated versions on a server.
  • Motivation to Combine: A POSITA would be motivated to incorporate Redpath's translation quality options to address the known accuracy deficiencies of machine translation. By providing users a choice, the combined system would be more flexible and powerful, allowing a user to select a quick but potentially inaccurate translation for time-sensitive information or a more accurate, human-reviewed translation for critical documents. Redpath's express goal was to provide "accurately translated" documents, a motivation directly applicable to improving the Williamowski/Sullivan/Poznanski system.
  • Expectation of Success: Integrating Redpath's features was presented as an obvious improvement. The combination would yield a more robust system by addressing a well-known problem (machine translation inaccuracy) with a known solution (providing multiple translation quality tiers).

4. Key Claim Construction Positions

• "Dialectal Standardization": Petitioner argued for adopting the construction proposed in related district court litigation: "to map keywords from different styles and dialects into standard and less ambiguous keywords," or the court's interpretation of "replacing the at least one content word in the first language with a commonly-known word in the first language so that the second-language search engine will recognize it."
• "Contextual Search": Petitioner contended that the Board's construction from a prior proceeding should apply: "identification of/identifying relevant documents from the domain-unlimited set of documents available on the World Wide Web, based on words contained in the documents."
• "Means for Selecting": For this means-plus-function term in claim 19, Petitioner asserted the function is "selecting among the search results found" and the corresponding structure disclosed in the specification is "a mouse or other pointing device."

5. Relief Requested

• Petitioner requested the institution of an inter partes review and the cancellation of claims 1-28 of the ’101 patent as unpatentable.