Dell Inc v. SElene Communication Technologies LLC

1. Case Identification

• Case #: IPR2014-01411
• Patent #: 6,363,377
• Filed: August 28, 2014
• Petitioner(s): Dell, Inc., EMC Corporation, Hewlett Packard, Inc., Monster Worldwide, Inc., Oracle Corporation, and Thomson Reuters Corporation
• Patent Owner(s): Selene Communication Technologies, LLC
• Challenged Claims: 1-14

2. Patent Overview

• Title: Search Data Processor
• Brief Description: The ’377 patent describes methods and systems for information retrieval, focusing on two main areas: refining and improving search queries, and organizing search query results. The query refinement claims involve automatically generating improved queries by substituting items from a hierarchical structure, while the result organization claims involve clustering records based on shared patterns.

3. Grounds for Unpatentability

Ground 1: Claims 1, 3-7, 11, and 13 are obvious over Gauch and Salton

• Prior Art Relied Upon: Gauch (a 1991 technical report on automatic query reformulation) and Salton (a 1982 paper on automatic Boolean query processing).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Gauch taught an expert system that automatically improves a user's Boolean search query using a hierarchical thesaurus. This system reformulates queries based on a user-specified target number of matches by adding related terms (synonyms, parents, children) from the thesaurus to broaden or narrow the search. This process inherently discloses identifying a level in a hierarchy tree (claim 1(c)) and substituting items to formulate new queries (claim 1(d)). Salton was cited for its disclosure of converting a natural language query into a standard Boolean query, a feature Gauch suggested as future work. The combination, therefore, taught receiving a natural language query, converting it to Boolean, and then using a hierarchical thesaurus to generate additional queries. Petitioner asserted that Gauch’s expanded query, which includes all original and additional terms, is sent to the search engine at once, satisfying the "in parallel" limitation.
  • Motivation to Combine: A POSITA would combine Salton with Gauch to improve Gauch's system by allowing users to input queries in natural language rather than requiring them to formulate a Boolean query, thereby making the system simpler to use.
  • Expectation of Success: A POSITA would have a reasonable expectation of success in combining the references because Salton's well-known technique for converting a natural language to a Boolean query could be straightforwardly integrated as a front-end to Gauch's existing Boolean query reformulation system.

Ground 2: Claims 1, 3-7, 11, and 13 are obvious over Fox and Frakes

• Prior Art Relied Upon: Fox (a 1988 book chapter on improving retrieval using a relational thesaurus) and Frakes (a 1992 information retrieval textbook).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner contended that Fox disclosed improving extended Boolean search queries by adding terms from a relational thesaurus. This process inherently involved looking up search terms in the thesaurus, which constitutes identifying a level in a hierarchy (claim 1(c)), and adding related terms (e.g., synonyms), which constitutes formulating additional queries (claim 1(d)). Frakes was cited for its teaching of using stemming—a well-known concept—to expand search queries. The combination of Fox's thesaurus-based expansion with Frakes's stemming-based expansion disclosed a multi-level hierarchical approach to query refinement. Petitioner argued the combined query, with original and expanded terms joined by Boolean operators, would be sent to the search engine as a single unit, satisfying the "in parallel" limitation.
  • Motivation to Combine: A POSITA would combine Frakes's stemming techniques with Fox's relational thesaurus technique as a simple and well-known way to further improve the performance of the query expansion process.
  • Expectation of Success: Success would be expected because applying stemming (from Frakes) is a standard, predictable technique in information retrieval that was known to be compatible with other query expansion methods like the thesaurus-based approach taught by Fox.

Ground 3: Claims 9, 10, 12, and 14 are obvious over Krellenstein and Zamir

• Prior Art Relied Upon: Krellenstein (Patent 5,924,090) and Zamir (a 1997 paper on clustering web documents).

• Core Argument for this Ground:

  • Prior Art Mapping: Petitioner argued Krellenstein taught a system for organizing search results by grouping them into categories (clusters) based on common characteristics (shared patterns). This disclosed the core steps of claims 9 and 12, including receiving records, analyzing fields to find patterns, and partitioning records into clusters. Zamir was cited for its disclosure of a specific method for scoring (weighting) document clusters. Zamir's score was based on the number of documents in a cluster and the length of the shared phrase (pattern), directly teaching the "analyzing a best cluster for display by weighing" limitation of claims 9 and 14.
  • Motivation to Combine: A POSITA seeking to improve the clustering method of Krellenstein would have been motivated to incorporate the specific and effective weighting/scoring technique from Zamir, which was in the same field of endeavor, to better identify the most relevant clusters for presentation to the user.
  • Expectation of Success: A POSITA would expect success in applying Zamir's scoring model to Krellenstein's clustering system because weighting clusters to determine relevance was a known technique, and Zamir provided a concrete, functional method for doing so.

• Additional Grounds: Petitioner asserted numerous other obviousness challenges, including combinations of Gauch/Salton with Smeaton (for application to a global information network), Thornton (for an alternative "in parallel" construction), and Stanfill I/II (for a parallel processing search engine architecture). Additional grounds were also asserted against the clustering claims based on Krellenstein in view of Frakes.

4. Key Claim Construction Positions

• "general Boolean language" (claims 1, 11, 13): Petitioner proposed this term be construed as a “generalized Boolean language which includes AND, OR and NOT operators and other operators.” This broad construction was argued to be consistent with the specification, which discusses various operators, and would encompass extended Boolean systems like those described in the prior art.
• "forwarding the initial search query and the additional search queries in parallel to the search engine" (claims 1, 11, 13): Petitioner argued this should be construed as “sending the initial search query and the additional search queries to the search engine at the same time.” This construction, based on the Patent Owner's litigation positions, would cover sending a single, concatenated query containing both original and new terms (e.g., linked by "OR" operators). Petitioner also presented alternative arguments for constructions requiring separate communication paths.

5. Relief Requested

• Petitioner requested institution of an inter partes review and cancellation of claims 1-14 of the ’377 patent as unpatentable.