Language-Model Execution Plans for Collaborative Data Analytics

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

There is a communication gap between data engineers and domain experts due to differing technical and high-level expertise, leading to cumbersome and error-prone data analysis processes, often resulting in incorrect reports.

Innovation Solution

A system using machine learning-based language models generates and modifies execution plans for natural language data analytics requests, allowing collaboration between users with varying expertise levels by iteratively refining these plans based on user feedback and implicit feedback from multiple users.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data engineers with technical expertise process the data, then the technical complexity is handled, but the high-level analytical understanding is lost

Engineering Contradiction:

Improvetechnical processing accuracyVSAvoidhigh-level analytical understanding

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The patent introduces an intermediary system comprising a language model and execution plan generator that translates natural language analytical requests into technical execution plans. This intermediary bridges the gap between domain experts who lack technical expertise and data engineers who lack high-level analytical understanding, allowing both parties to work effectively without direct communication while preserving both technical accuracy and analytical intent.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Loss of information

If domain experts with high-level understanding analyze the data, then the analytical goals are clear, but the technical expertise to interact with various systems is lacking

Engineering Contradiction:

Improvehigh-level analytical understandingVSAvoidtechnical system interaction

Core Design Contradiction:

Loss of informationVSEase of operation

Solution Approach 1:

The system enables domain experts to perform data analysis independently through natural language interfaces without requiring technical expertise. The execution plan generator automatically handles the conversion of analytical requests into technical operations across multiple data systems, allowing domain experts to self-serve their analytical needs without relying on data engineers.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The intermediary translation layer converts domain expert analytical intent into technical system operations, eliminating the need for domain experts to directly interact with complex technical systems while preserving their high-level analytical understanding.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Reliability

If multiple iterations of interactions occur between users with different expertise, then communication is attempted, but the process becomes cumbersome and error-prone

Engineering Contradiction:

Improvecommunication accuracyVSAvoiditeration time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system incorporates feedback mechanisms where execution plans are automatically evaluated and refined based on their outcomes. The language model receives feedback from execution results and iteratively improves the translation between natural language requests and technical operations, reducing errors and improving accuracy over time without requiring human iterations.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The automated intermediary system eliminates the need for multiple human interaction iterations by directly translating and executing analytical requests, significantly reducing both the time required and the potential for communication errors between users with different expertise levels.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Productivity

If automated execution plans are generated using language models, then productivity increases, but the complexity of the system increases

Engineering Contradiction:

Improvedata analytics throughputVSAvoidsystem architecture complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system segments the data analytics process into distinct modular components: natural language input processing, execution plan generation, plan execution, and result delivery. Each component is independently managed and can be developed, tested, and maintained separately, reducing the manageability of system complexity while maintaining high productivity through automated end-to-end processing.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS12579361B1Collaborative data analytics using machine learning based language models

Publication Date: 2026.03.17 ISOTOPES AI INC

AI summary

A system allows multiple users to interact with the system to perform data analysis. The system allows users to specify data analytics requests using high-level natural language requests. The system generates an execution plan based on natural language requests received from a particular user and executes the execution plan. The system detects a failure in executing the execution plan and determines whether to request help from another entity. The system may change the state of execution of the execution plan to a waiting state in which the system blocks execution until it receives an approval to proceed from an entity. The system receives a request to modify the execution plan from the entity and modifies the execution plan based on the request. The system iteratively performs the modification of the execution plan until an acceptable execution plan is generated.