DCT

2:10-cv-00132

First American CoreLogic Inc v. Fiserv Inc

Log In

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 2:10-cv-00132, E.D. Tex., 06/09/2010
  • Venue Allegations: Venue is alleged to be proper based on defendants conducting systematic and continuous business in the district, having minimum contacts, and committing alleged acts of infringement within the Eastern District of Texas.
  • Core Dispute: Plaintiff alleges that eleven different defendants' Automated Valuation Model (AVM) products and services infringe a patent related to computer-implemented real estate appraisal using predictive modeling.
  • Technical Context: The lawsuit concerns Automated Valuation Models (AVMs), which are software tools that use statistical modeling to provide real estate value estimations, a technology central to the mortgage and real estate industries.
  • Key Procedural History: This First Amended Complaint was filed in 2010. Subsequent to the filing, the patent-in-suit, U.S. Patent No. 5,361,201, was the subject of post-grant proceedings. A disclaimer for claim 5 was filed in 2012. Following Covered Business Method (CBM) reviews, the USPTO issued a Post-Grant Review Certificate in 2016 cancelling claims 1, 6, 9, and 10, and noting that claims 2-4, 7, 8, and 11-20 were disclaimed. Cumulatively, these post-filing events have resulted in the cancellation or disclaimer of all original claims of the patent.

Case Timeline

Date Event
1992-10-19 '201 Patent Priority Date
1994-11-01 '201 Patent Issue Date
2010-06-09 Complaint Filing Date
2012-12-26 Disclaimer of Claim 5 Filed
2016-01-12 Post-Grant Review Certificate Issued

II. Technology and Patent(s)-in-Suit Analysis

U.S. Patent No. 5,361,201 - "Real Estate Appraisal Using Predictive Modeling"

  • Patent Identification: U.S. Patent No. 5,361,201, “Real Estate Appraisal Using Predictive Modeling,” issued November 1, 1994 (the “’201 Patent”).

The Invention Explained

  • Problem Addressed: The patent seeks to remedy problems with traditional real estate appraisal, which it characterizes as being subject to the appraiser's individual bias, difficult to apply consistently, and insufficient to capture complex, non-linear interactions among property variables ('201 Patent, col. 2:1-9). This uncertainty creates risk for parties relying on the appraisal, such as lenders and investors ('201 Patent, col. 2:1-5).
  • The Patented Solution: The invention proposes an automated system that uses one or more "predictive models," such as neural networks, to generate a property value estimate ('201 Patent, col. 2:30-35). The system first develops these models by training them on a large set of historical sales data that includes individual property characteristics (e.g., square footage) and broader area characteristics (e.g., neighborhood trends) ('201 Patent, Abstract; Fig. 9). Once a model is created and stored, it can be applied to new properties to generate an appraised value, an estimated error range, and "reason codes" explaining the factors that contributed to the final valuation ('201 Patent, col. 2:51-57).
  • Technical Importance: The patent describes an early, data-driven approach to automate and objectify real estate valuation, moving the process from a human-centric appraisal to a computer-implemented, statistical one. ('201 Patent, col. 2:10-19).

Key Claims at a Glance

  • The complaint alleges infringement of "one or more of the claims" of the ’201 Patent without specifying which ones (Compl. ¶20). Independent claim 1 is representative of the asserted method.
  • Independent Claim 1:
    • A computer-implemented process for appraising a real estate property, comprising the steps of:
    • collecting training data;
    • developing a predictive model from the training data;
    • storing the predictive model;
    • obtaining individual property data for the real estate property; and
    • generating a signal indicative of an appraised value for the real estate property responsive to application of the obtained individual property data to the stored predictive model.

III. The Accused Instrumentality

Product Identification

The complaint accuses numerous Automated Valuation Model (AVM) products and services, including Fiserv’s “CASA® AVM and ValuServ” (Compl. ¶20), IntelliReal’s “iAVM™” (Compl. ¶24), Interthinx’s “Clear VALUE® AVM” (Compl. ¶28), Lender Processing Services' "LPS AVM Connect™" (Compl. ¶32), and Zillow's "Zillow.com" online platform (Compl. ¶48).

Functionality and Market Context

The complaint broadly characterizes the accused instrumentalities as "AVMs for real estate appraisal" (Compl. ¶¶ 2-12). It alleges that these systems are used to generate automated property valuations, thereby performing a function that falls within the scope of the ’201 Patent's claims (Compl. ¶20). The large number of defendants suggests a competitive market for AVM services at the time the suit was filed.

IV. Analysis of Infringement Allegations

The complaint does not provide a claim chart or detailed, element-by-element mapping of the accused products to the patent claims. The infringement theory is articulated in general terms, alleging on "information and belief" that each defendant's AVM systems "falling within the scope of one or more of the claims of the '201 Patent" are manufactured, used, sold, or offered for sale (Compl. ¶20). The core allegation is that the defendants' AVMs necessarily perform the patented process of developing and/or applying a predictive model using property data to generate a valuation.

No probative visual evidence provided in complaint.

  • Identified Points of Contention:
    • Technical Questions: A primary factual question is whether each of the defendants' distinct AVM products actually performs all the steps of any single asserted claim. For instance, the complaint does not specify what evidence demonstrates that each defendant engages in "developing a predictive model from the training data" (Claim 1) as opposed to merely using a pre-existing or third-party model.
    • Scope Questions: The complaint's undifferentiated allegations against a wide variety of AVM products raises the question of whether the technical operation of each product is the same. The dispute may turn on whether the specific algorithms and data processing methods used by each defendant meet the limitations of the claims as construed by the court.

V. Key Claim Terms for Construction

  • The Term: "predictive model"

  • Context and Importance: This term is the technological heart of the patent. Its construction is critical because it will determine whether the various algorithms used by the defendants—which may range from simple regression to more complex machine learning systems—fall within the scope of the claims.

  • Intrinsic Evidence for Interpretation:

    • Evidence for a Broader Interpretation: The specification suggests the term is not limited to a single technology, stating that while neural networks are a preferred embodiment, "any type of predictive modeling technique may be used, such as regression modeling" (’201 Patent, col. 6:13-16).
    • Evidence for a Narrower Interpretation: The patent's abstract, figures, and detailed description heavily emphasize "neural network models" as the solution to the problems of prior art methods ('201 Patent, Abstract; col. 2:32-34; col. 6:7-13). A defendant could argue that the scope of "predictive model" should be understood in the context of this specific, repeatedly-disclosed embodiment.

  • The Term: "developing a predictive model from the training data"

  • Context and Importance: This term is central to distinguishing the creation of a valuation tool from its mere use. Infringement of method claims like Claim 1 depends on the performance of this "developing" step. Practitioners may focus on this term to determine if defendants who use, but do not create, their own models can be liable for direct infringement.

  • Intrinsic Evidence for Interpretation:

    • Evidence for a Broader Interpretation: This phrase could be interpreted to cover any automated or semi-automated process where a set of data is used to define the parameters of a valuation algorithm.
    • Evidence for a Narrower Interpretation: The specification describes a multi-stage process for model development, including aggregating data by geographic region, training a neural network, and then training a separate error model ('201 Patent, col. 5:65-col. 6:23; Fig. 8). A party could argue that "developing" requires these specific, disclosed steps.

VI. Other Allegations

  • Indirect Infringement: The complaint alleges that defendants contribute to and induce infringement by "offering to sell and selling its systems... to customers, buyers, sellers, users and others that directly infringe the '201 Patent" (Compl. ¶¶ 20, 24, 28). The basis is the provision of AVM tools that allegedly cause end-users to perform the patented method.
  • Willful Infringement: For each defendant, the complaint alleges on information and belief that the defendant "is aware of the existence of the '201 Patent and, despite such knowledge, continues to willfully, wantonly and deliberately engage in acts of infringement" (Compl. ¶¶ 21, 25, 29).

VII. Analyst’s Conclusion: Key Questions for the Case

  • A Threshold Question of Viability: The most significant issue in this case is procedural. Given that every claim of the ’201 Patent was either cancelled or disclaimed in post-grant proceedings that occurred after the complaint was filed, the fundamental basis for the lawsuit has been eliminated. The central question is therefore not one of infringement, but of the case's ability to proceed at all without any valid and enforceable patent claims.
  • A Question of Claim Scope: Setting aside the patent's invalidity, a core issue would have been one of definitional scope: whether the term "predictive model", as described in a 1992-filed patent focused on neural networks, could be broadly construed to cover the potentially diverse range of statistical and algorithmic AVM technologies employed by eleven different defendants nearly two decades later.
  • An Evidentiary Question of Conduct: The complaint's "information and belief" allegations against a multitude of defendants would have presented a significant evidentiary challenge. A key question would have been whether the plaintiff could prove that each defendant independently performed every claimed step, particularly the act of "developing a predictive model", a fact-intensive inquiry that would likely vary for each accused product.