Hybrid Library for Data Science Code Translation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Data scientists face challenges in deploying statistical models due to the lack of suitable tools for translating development language code into production environments, leading to visibility issues and potential errors, which introduces latency and requires manual configuration of production computing environments.

Innovation Solution

A development system that uses a hybrid library to translate source code from a development programming language into a production programming language, preserving relationships, functions, and configurations by defining low-level transformations that can be combined into macro-transformations, enabling direct deployment from development to production environments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If third party tools are used to translate source code from development language to production language, then code translation is achieved, but visibility and assurance of correct translation is lost

Engineering Contradiction:
Improvecode translation capabilityVSAvoidtranslation visibility
Core Design Contradiction:
Adaptability or versatilityVSLoss of information

Solution Approach 1:

The patent introduces an intermediary translation service that operates between development and production environments. This service provides structured translation with visibility through translation logs, metadata tracking, and approval workflows, allowing organizations to maintain control and visibility while enabling cross-language code translation

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The translation process is segmented into discrete, trackable units with individual translation records for each code element. This segmentation enables granular visibility into translation status, quality metrics, and approval states, allowing organizations to monitor and control the translation process at fine-grained levels

Inventive Principle:
Principle #1Segmentation

2Productivity

If source code is directly accepted in production environment, then deployment speed is improved, but reliability and security are compromised

Engineering Contradiction:
Improvedeployment speedVSAvoidproduction reliability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent implements preliminary validation, translation, and approval actions in the development environment before code reaches production. Translation services pre-process code with quality checks, and approval workflows ensure readiness before deployment, enabling faster production deployment while maintaining reliability through advance preparation

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system incorporates feedback loops where translation quality metrics, validation results, and production performance data feed back into the translation service. This feedback enables continuous improvement of translation accuracy and reliability while maintaining deployment speed through automated adjustments

Inventive Principle:
Principle #23Feedback

3Adaptability or versatility

If manual configuration of production environment is performed, then resource support for models is achieved, but latency and complexity are introduced

Engineering Contradiction:
Improveresource configurationVSAvoidconfiguration latency
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent enables self-service configuration where the translation service automatically detects production environment requirements and configures resources without manual intervention. The system self-adapts to production constraints and automatically provisions necessary resources, eliminating manual configuration latency while maintaining adaptability

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system dynamically adjusts configuration parameters based on translation results and production environment characteristics. By automatically modifying resource allocation parameters, performance settings, and deployment configurations, the system achieves adaptive resource support without manual configuration delays

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3449361B1Method and system for developing and deploying data science transformations from a development computing environment into a production computing environment
Publication Date: 2022.11.23 INTUIT INC
  • EP3449361B1 patent drawingFigure 1
  • EP3449361B1 patent drawingFigure 2
  • EP3449361B1 patent drawingFigure 3

AI summary

A method and system facilitates the development of data science transformations in one programming language and the deployment of the data science transformations in another programming language, according to one embodiment. The method and system preserves relationships, functions, configurations, and characteristics between combinations of data transformations, according to one embodiment. The preservation of the relationships, functions, configurations, and characteristics is enabled by developing and providing a set of low-level (e.g., atomic) transformations that enable users to build their own models, libraries, and configurations into macro-transformations (e.g., conglomerate transformations), according to one embodiment. Deploying the data science transformations into production computing environments is useful for providing or supporting a number of types of software services, such as, predicting user behavior, customizing user experiences, supporting marketing initiatives, providing empirically-backed recommendations, predicting user preferences for software interactions, and/or otherwise exposing patterns that are identified from historical and transactional data, according to one embodiment.