AI-Based Computer Program Deduplication for Lower Maintenance Overhead

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In organizations, particularly larger ones, multiple teams develop similar programs that consume excessive computing resources for storage and maintenance, and updates to one program may lead to sub-optimal operation of others with similar functionality.

Innovation Solution

A deduplication system (DS) using artificial intelligence (AI) to identify and deduplicate similar computing programs by comparing dimensions such as address, table, input, and output features, recommending or performing deduplication, merging, or refactoring to reduce redundancy and resource overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple teams develop similar programs independently, then each team can customize programs for their specific purposes, but computing resources are consumed excessively for storage and maintenance

Engineering Contradiction:
Improveprogram customizationVSAvoidcomputing resource consumption
Core Design Contradiction:
Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The patent implements a program registry that serves multiple functions: storing program information, enabling similarity detection, facilitating deduplication, and supporting reuse. This centralized registry allows the system to handle diverse program types and purposes while managing them through a universal interface, reducing the need for separate management systems for each program

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges multiple similar programs into a single standardized version by identifying duplicates through similarity comparison. When similar programs are detected, the system consolidates them into one canonical version that can serve multiple purposes, thereby reducing storage requirements and maintenance overhead while preserving the functionality needed by different teams

Inventive Principle:
Principle #5Merging (Combining)

2Productivity

If multiple similar programs are maintained separately, then each program can be updated independently, but maintenance burden increases significantly

Engineering Contradiction:
Improveupdate speedVSAvoidmaintenance complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system merges maintenance operations by identifying similar programs and consolidating them into a single maintained version. Updates to the canonical program automatically benefit all programs that use or reference it, reducing the number of separate maintenance operations required while maintaining the ability to update programs independently when necessary

Inventive Principle:
Principle #5Merging (Combining)

3Reliability

If similar programs are not identified, then program diversity is maintained, but optimization opportunities are lost when one program is updated

Engineering Contradiction:
Improveprogram optimizationVSAvoidprogram identification complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces manual program comparison and identification processes with an automated AI-based similarity detection system. The AI model analyzes program code, structure, and functionality to automatically identify similar programs, eliminating the need for manual analysis while providing accurate optimization opportunities across the program portfolio

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12455734B2Deduplication of computer programs using similarity measures and artificial intelligence
Publication Date: 2025.10.28 SAP SE
  • US12455734B2 patent drawing
  • US12455734B2 patent drawing
  • US12455734B2 patent drawing

AI summary

Various embodiments for deduplication of APIs using similarity measures and artificial intelligence are described herein. An embodiment operates by receiving a request to compare a first computing program to a second computing program, wherein each computing program includes an address, one or more tables accessed by a respective computing program, one or more input parameters, and one or more output parameters. Similarity measures are calculated between the addresses, tables, input parameters, and output parameters of the two computing programs. The similarity measures are provided to a trained artificial intelligence (AI) model, which generates a similarity determination. Performed an action based on the similarity determination.