Invention Title:

Interactive Network for Selecting, Ranking, Summarizing, and Exploring Data Insights

Publication number:

US20250336116

Publication date:
Section:

Physics

Class:

G06T11/206

Inventors:

Assignee:

Applicant:

Smart overview of the Invention

The patent application describes techniques for generating insight summaries and prompts from digital content data. These techniques involve creating a network representation with nodes and connections based on insights extracted from the data. Users can select a subset of these nodes to form prompts, which are then used to generate summaries using generative AI and machine-learning models. These summaries are presented in a user interface, aiming to enhance data interpretation and overcome limitations of conventional analytics methods.

Background

With the continuous growth of datasets and diverse data types, traditional data analytics face challenges in data interpretation. Issues such as excessive information access, lack of specialized knowledge, and inefficiencies in conventional machine learning approaches exacerbate these challenges. The described techniques aim to address these issues by improving the accuracy and efficiency of data summarization through advanced AI methods.

Detailed Description

The invention focuses on automated summary generation to improve dataset interpretation. It addresses challenges like dataset size, expression across datasets, and specialized knowledge requirements. By utilizing a digital insight system, the invention generates summaries from digital content displayed in interfaces such as dashboards. This approach aims to overcome conventional technique limitations that often miss valuable information or fail to accurately represent data insights.

Network Representation and Prompt Generation

The data insight system constructs a network representation by forming connections between extracted insights from source datasets. These connections include layout-based, type-based, topic-based, and other types. The system ranks nodes and connections to form prompts used as inputs for machine-learning models. This process results in insight summaries that balance natural language with factual completeness, offering improved accuracy over traditional methods.

User Interface and Exploration

The system supports an interactive user interface for exploring insights. Features include a network visualization panel for real-time node selection and a story exploration panel for reviewing insights with visualizations. A summary browser panel allows exploration of generated insights sets. These tools enable users to drill down into datasets for additional information, addressing conventional system challenges by providing accurate, automated insight summaries without user intervention.