Anagram Generation Apparatus Using Language Model Pruning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional anagram generation technologies struggle to produce natural and understandable text strings, often generating outputs that humans cannot comprehend when read.

Innovation Solution

A generating device that includes input means for inputting text data and generating means for generating anagrams that satisfy predetermined constraint conditions, such as plausibility of the sequence of letters, using a depth-first search algorithm and a language model to evaluate naturalness.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If conventional anagram generation technology is used to rearrange letters, then all possible text strings can be generated, but the generated text strings are often unnatural and incomprehensible to humans

Engineering Contradiction:

Improveability to generate various anagram combinationsVSAvoidnaturalness and understandability of generated anagrams

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent employs a language model that provides feedback during the depth-first search process to evaluate whether the current letter sequence forms a natural word or phrase. This feedback mechanism allows the system to prune unrealistic branches and guide the search toward generating anagrams that are both valid and comprehensible, resolving the contradiction between generating diverse combinations and ensuring naturalness.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent changes the evaluation parameters from simply checking letter composition to incorporating linguistic plausibility assessment through a language model. By introducing naturalness as a constraint parameter during the generation process, the system filters out incomprehensible anagrams while maintaining the ability to generate valid letter rearrangements.

Inventive Principle:

Principle #35Parameter changes

2Productivity

If a depth-first search algorithm is used to generate anagrams, then the search can be pruned for efficiency, but additional computational constraints are introduced

Engineering Contradiction:

Improvegeneration speed through search pruningVSAvoidcomputational constraints and language model integration

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent performs preliminary actions by integrating the language model constraints into the search process before completing the full anagram generation. The depth-first search algorithm uses the language model to evaluate partial sequences early in the search, allowing pruning of unrealistic branches before they are fully explored, thus improving efficiency while managing complexity through early constraint application.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12210844B2Generation apparatus, generation method and program

Publication Date: 2025.01.28 NIPPON TELEGRAPH & TELEPHONE CORP

AI summary

Included are input means for inputting first data that is data relating to a plurality of letters included in a text string that is a generation target, and generating means for generating second data that is data relating to the text string that satisfies predetermined constraint conditions including at least a condition relating to plausibility of the sequence of letters, on the basis of the first data.