DCT
1:23-cv-12200
Kanneh v. Databox Inc
Key Events
Complaint
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Francis Kanneh (California)
- Defendant: Databox, Inc. (Delaware)
- Plaintiff’s Counsel: Pro se
- Case Identification: 1:23-cv-12200, D. Mass., 09/26/2023
- Venue Allegations: Venue is based on Defendant's principal place of business being located in Boston, Massachusetts.
- Core Dispute: Plaintiff alleges that Defendant’s business intelligence dashboard templates, which include forecasting features, infringe a patent related to a system and method for creating forecast charts using machine learning.
- Technical Context: The technology operates in the field of automated data analytics and business intelligence, where platforms ingest data from multiple sources and use algorithms to generate predictive insights and visualizations.
- Key Procedural History: The complaint references pre-suit correspondence, including a June 2023 notice letter sent by the Plaintiff to the Defendant and an August 2023 response from Defendant's counsel denying that the accused products use machine learning. This exchange establishes a basis for the Plaintiff's allegations of pre-suit knowledge and willful infringement.
Case Timeline
| Date | Event |
|---|---|
| 2021-08-02 | ’837 Patent Priority Date |
| 2023-04-18 | ’837 Patent Issue Date |
| 2023-06-01 | Approximate Accused Product Launch Date |
| 2023-06-05 | Plaintiff sends infringement notice letter to Defendant |
| 2023-08-18 | Defendant responds to Plaintiff, denying use of ML |
| 2023-09-26 | Complaint Filing Date |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 11,630,837 - "Computer-Implemented System and Method for Creating Forecast Charts"
- Patent Identification: U.S. Patent No. 11,630,837, "Computer-Implemented System and Method for Creating Forecast Charts," issued April 18, 2023 (the "’837 Patent").
The Invention Explained
- Problem Addressed: The patent's background section identifies a need for a system that can automatically create forecast charts using machine learning (ML) algorithms on large datasets, observing that existing implementations often require manual adjustment by users (’837 Patent, col. 1:49-52).
- The Patented Solution: The invention describes an automated system that connects to multiple data sources, retrieves potentially unstructured data, and converts it into a structured format for analysis (’837 Patent, Abstract). The core of the solution involves using this structured data to train and deploy ML models on a server, which then generate predictive data. This prediction data is returned via an Application Programming Interface (API) to a web application, where it is displayed to the user as a forecast chart within a dashboard (’837 Patent, col. 2:25-40; FIG. 2).
- Technical Importance: The patented approach seeks to automate complex data science workflows for business forecasting, thereby making predictive analytics more accessible to users without specialized expertise (’837 Patent, col. 1:21-31).
Key Claims at a Glance
- The complaint asserts independent system claim 1 and independent method claim 10 (Compl. ¶12).
- Claim 1 (System) Essential Elements:
- A computing device with a processor and memory.
- The processor is configured to: connect a website application to a plurality of data sources.
- Retrieve unstructured data from the sources.
- Format and convert the unstructured data into structured data stored in a database.
- Transform the structured data into forecast charts using dashboard templates.
- Train a plurality of Machine Learning (ML) models on the structured data.
- Deploy the ML models on a server connected to an API.
- Utilize the ML models for "recursive multi-step forecasting" to provide prediction data.
- Create a dashboard that includes the forecast charts based on the prediction data from the API.
- Claim 10 (Method) Essential Elements:
- Connecting a website application to a plurality of data sources.
- Retrieving unstructured data.
- Formatting and converting the unstructured data into structured data.
- Transforming the structured data into forecast charts.
- Training a plurality of ML models on the structured data.
- Deploying the ML models on a server.
- Using the ML models to provide prediction data.
- Creating a dashboard that displays the forecast charts based on the prediction data.
III. The Accused Instrumentality
Product Identification
- The "Accused Products" are identified as "various dashboard templates available on Databox's website that each show future date ranges," specifically implicating the platform's forecasting feature (Compl. ¶10, ¶19).
Functionality and Market Context
- The complaint describes the accused functionality as part of a business intelligence platform that allows users to connect to various data sources (e.g., Facebook Ads, Google Ads) via APIs and SDKs (Compl. ¶21). The platform allegedly ingests this data, processes it, and displays it in customizable dashboards (Compl. ¶22). The central accused feature is a forecasting capability that Plaintiff alleges is "completely automated using machine learning algorithms" to predict future trends based on historical data (Compl. ¶18). The complaint provides a screenshot of a Databox dashboard showing metrics from Facebook Ads and Google Ads, with trend lines and date range options that include future periods like "Tomorrow" and "Next Month" (Compl. FIG. 3, p. 8).
IV. Analysis of Infringement Allegations
’837 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| connect a plurality of data sources to a website application... | The Databox platform uses a REST API and SDKs to connect to various data sources. | ¶21 | col. 6:33-35 |
| retrieve... unstructured data from the plurality of data sources; | The platform can "integrate any data, including unstructured data." | ¶21 | col. 6:39-41 |
| format unstructured data and convert the unstructured data into structured data... stores the structured data in a table of a database; | The platform obtains data, transforms it, and uses it in dashboard templates. | ¶21, ¶22 | col. 6:41-48 |
| transform... the structured data into one or more forecast charts by performing a plurality of computations... wherein the structured data is used on one or more of a plurality of dashboard templates... | The platform creates dashboards from templates that include forecast charts. | ¶22 | col. 6:50-65 |
| train a plurality of Machine Learning (ML) models on the structured data of the dashboard templates; | Plaintiff alleges on information and belief that Databox trains ML models, supported by public statements from its CEO and data scientists about using machine-learning methods. | ¶18, ¶20, ¶23 | col. 7:1-4 |
| deploy the Machine Learning (ML) models on a server... connected to an application programming interface (API)... | Plaintiff alleges on information and belief that Databox deploys ML models on a server connected to an API. | ¶23 | col. 7:4-7 |
| wherein the Machine Learning (ML) models are used to provide a prediction data based on the structured data; | The Databox CEO is quoted as stating the forecasting feature is "completely automated using machine learning algorithms" to "forecast future growth." This statement is from a LinkedIn post provided as a screenshot in the complaint. | ¶18, ¶25; FIG. 1, p. 6 | col. 7:21-23 |
| create a dashboard based on the dashboard template selected by the user, wherein the dashboard includes the one or more forecast charts... displayed based on the prediction data... | The Databox platform allows users to create dashboards that display forecast charts predicting data over a future period. The complaint includes a screenshot of a LinkedIn post from the CEO announcing the rollout of the forecasting feature. | ¶22, ¶24; FIG. 2, p. 6 | col. 7:23-30 |
- Identified Points of Contention:
- Technical/Factual Question: The central dispute appears to be factual: whether the Accused Products actually "train" and "deploy" "Machine Learning (ML) models" to generate forecasts. The complaint presents public statements from Defendant’s CEO as evidence of ML use (Compl. ¶18), but also acknowledges and attaches Defendant’s pre-suit denial stating that "[n]one of the Databox dashboard templates leverage machine learning to forecast results" (Compl. ¶15-16). This creates a direct conflict of fact that will likely be the focus of discovery and expert testimony.
- Scope Question: A potential claim construction issue may arise regarding the term "unstructured data." The complaint alleges that Databox integrates "unstructured data," but the examples provided involve API data from services like Google Ads, which is typically delivered in a semi-structured format (e.g., JSON). The case may turn on whether such data falls within the scope of "unstructured data" as contemplated by the patent.
V. Key Claim Terms for Construction
The Term: "Machine Learning (ML) models"
- Context and Importance: This term is at the heart of the infringement dispute, as Defendant has explicitly denied that its products use "machine learning" (Compl. ¶15). The definition will be critical to determine whether the forecasting algorithms used by Databox, if any, meet this claim limitation. Practitioners may focus on this term because the case may depend on the distinction between a complex ML model and a simpler statistical projection or "trend line" (Compl. ¶18, ¶25).
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification does not provide a formal definition but gives examples of models such as "linear regressor, SGD regressor, lasso, [and] ElasticNet" (’837 Patent, FIG. 3, block 306), which could suggest the term covers a range of predictive statistical models.
- Evidence for a Narrower Interpretation: The claims require the models to utilize "a recursive multi-step forecasting" approach and "adjust a parameter" to fit the data (’837 Patent, col. 12:48-52). A party could argue that these specific functional requirements limit the scope of "ML models" to only those capable of performing these particular steps, potentially excluding simpler algorithms.
The Term: "unstructured data"
- Context and Importance: The claims require the system to retrieve and process "unstructured data." The nature of the data ingested by the accused Databox platform will be compared against the scope of this term.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The patent does not provide a limiting definition. The specification describes a process of fetching "RAW API Data," processing it, and formatting it into a "row and column form" (’837 Patent, col. 10:22-29), which could support an argument that any data requiring such transformation from its raw state is "unstructured" for the purposes of the invention.
- Evidence for a Narrower Interpretation: The primary example of "unstructured data" provided in the patent is a JSON object (’837 Patent, FIG. 6). Because JSON has inherent structure (key-value pairs), a party could argue that the term implies data that is not fully random but requires formatting to fit a relational database schema, potentially excluding other data types or arguing that standard API data is too structured to qualify.
VI. Other Allegations
- Indirect Infringement: The complaint alleges induced infringement, asserting that Defendant’s advertisements and instructions on its website intend for customers to use the Accused Products in a manner that infringes claims 1 and 10 (Compl. ¶35-37). Conclusory allegations of contributory infringement are also made (Compl. ¶54).
- Willful Infringement: The willfulness allegation is based on alleged knowledge of the ’837 Patent since at least June 5, 2023, the date of Plaintiff's notice letter (Compl. ¶11, ¶40). The complaint further alleges that Defendant's continued activities after receiving this notice have been deliberate and willful (Compl. ¶33).
VII. Analyst’s Conclusion: Key Questions for the Case
- A central evidentiary question will be one of technical fact: do the accused Databox forecasting features operate using "Machine Learning (ML) models" that are trained and deployed as required by the patent claims—as Plaintiff alleges based on marketing statements—or do they utilize simpler, non-infringing statistical methods, as Defendant contended in pre-suit communications?
- A key legal question will be one of claim construction: how broadly will the court define "unstructured data"? The viability of the infringement claim may depend on whether semi-structured data retrieved from third-party APIs is found to fall within the scope of this term as used in the patent.