Kcura LLC v. Blackbird Tech LLC

1. Case Identification

• Case #: IPR2017-00899
• Patent #: 7,809,717
• Filed: February 14, 2017
• Petitioner(s): kCura LLC
• Patent Owner(s): Blackbird Tech LLC d/b/a Blackbird Technologies
• Challenged Claims: 1-30

2. Patent Overview

• Title: Conceptual Search Result Sorting
• Brief Description: The ’717 patent discloses a method and system for sorting document search results based on their similarity to one or more user-selected concepts. The system determines concepts related to a search query, calculates accordance values indicating the similarity of each document to the selected concepts, and displays the results ranked by these values.

3. Grounds for Unpatentability

Ground 1: Obviousness over Lindh and Crow - Claims 1-4, 6, 9-13, 15-19, 23-26, 28, and 30 are obvious over Lindh in view of Crow.

• Prior Art Relied Upon: Lindh (WO 03/060766) and Crow (Application # 2005/0080656).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Lindh taught the core method of the ’717 patent. Lindh described a search system where a user enters a query, receives a set of related concepts, selects one or more of those concepts, and the system re-ranks the document results based on their similarity to the selected concepts. Lindh disclosed using various matrices (term-concept, document-concept) to represent relationships and calculate similarity scores (accordance values). Crow was cited for its teachings on conceptualizing documents (resumes) and search queries into vectors and ranking results based on their conceptual similarity, particularly using a "knowledge-based" system with a known ontology to improve conceptualization.
  • Motivation to Combine: Petitioner asserted a person of ordinary skill in the art (POSITA) would combine Lindh and Crow because they were in analogous fields of concept-based document retrieval and solved similar problems. Lindh itself suggested incorporating known prior-art methods for generating conceptual descriptions. A POSITA would have been motivated to integrate Crow’s more advanced or specific conceptualization techniques (e.g., using a known ontology) into Lindh’s framework to improve the accuracy and relevance of the conceptual matching and search refinement.
  • Expectation of Success: The combination would have yielded predictable results. Integrating Crow’s vector-based conceptualization into Lindh’s matrix-based system was a straightforward application of known data organization and comparison techniques, leading to an enhanced but predictable concept search system.

Ground 2: Obviousness over Lindh, Crow, and Mohan - Claims 5, 20, and 27 are obvious over Lindh and Crow in view of Mohan.

• Prior Art Relied Upon: Lindh (WO 03/060766), Crow (Application # 2005/0080656), and Mohan (Patent 7,890,514).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground built upon the Lindh and Crow combination by adding the teachings of Mohan. The challenged claims (e.g., claim 5) required displaying multiple accordance indicators for a single document, each corresponding to a different concept. While Lindh taught selecting multiple concepts to bias ranking, Petitioner argued it did not explicitly disclose displaying multiple, distinct scores for each concept graphically. Mohan taught a graphical user interface (GUI) for analyzing documents that graphically illustrated the strength of concept scores for selected concepts with respect to specific documents in a table format.
  • Motivation to Combine: A POSITA would have been motivated to incorporate Mohan’s GUI features into the Lindh/Crow system to improve the user interface. Providing a user with a graphical breakdown of a document’s relevance to multiple individual concepts, as taught by Mohan, would be an obvious improvement to Lindh's system, which already calculated these underlying relationships. It would provide the user with more granular information about why a document was ranked highly.
  • Expectation of Success: Integrating Mohan’s graphical display of pre-calculated scores into the Lindh/Crow system was a routine design choice with a high expectation of success, as it involved applying known UI design principles to existing data.

Ground 3: Obviousness over Lindh, Crow, and Melman - Claims 7-8, 14, 21-22, and 29 are obvious over Lindh and Crow in view of Melman.

• Prior Art Relied Upon: Lindh (WO 03/060766), Crow (Application # 2005/0080656), and Melman (Application # 2005/0091204).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground added Melman to the base Lindh/Crow combination to address claims requiring the display of search results in a list with at least one document being displayed in two different portions simultaneously (e.g., a title list and a summary view). The Lindh/Crow combination taught ranking and displaying a list of results. Melman taught a GUI that simultaneously displayed search results in two separate windows or portions: a list of document titles and a corresponding list of document summaries, where the number of items in each list could differ.
  • Motivation to Combine: Petitioner contended that a POSITA would have been motivated to incorporate Melman’s UI design to improve the usability of the Lindh/Crow system. Displaying both a concise title list and a more detailed summary view is a well-known, common-sense approach to presenting search results, allowing users to quickly scan titles while having immediate access to more detailed information. This would have been an obvious and desirable enhancement to the user interface of the primary references.
  • Expectation of Success: Combining Melman's display layout with the ranked search results from Lindh/Crow was a predictable implementation of known UI techniques.

4. Key Claim Construction Positions

• "concept knowledge base" (claims 1, 16): Petitioner proposed the construction "electronic repository containing concepts." This construction was argued to be consistent with the specification's description of the base as a data structure containing concept objects, term objects, and their relationships, and was broad enough to encompass the term-concept matrix taught by Lindh.
• "fuzzy membership score" (claims 3, 18, 25): Petitioner proposed the construction "non-binary value related to similarity." This was based on the specification's equation for an "accordance value" and the patent owner's own litigation positions equating the term with "fuzzy logic," which implies a continuous scale (e.g., 0 to 1) rather than a simple binary value. This construction allowed Petitioner to map the normalized relationship values from Lindh to this limitation.
• "document surrogate" (claims 9, 15, 23, 30): Petitioner proposed the construction "information representing a document." This was based on the common understanding that search engines provide summaries, titles, or snippets rather than full documents, which the specification also described. This construction was important for mapping the teachings of Lindh, which provided a plurality of documents in response to a query.

5. Relief Requested

• Petitioner requested the institution of an inter partes review and the cancellation of claims 1-30 of the ’717 patent as unpatentable under 35 U.S.C. §103.