Sentence Conversion via Deep Learning Rewrite Model

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Rule-based sentence rewrite methods in computer technologies suffer from poor generalization performance and stiffness, resulting in non-fluent generated sentences.

Innovation Solution

A sentence conversion method and apparatus utilizing a pre-trained deep learning-based rewrite model that converts input sentences into semantically equivalent but stylistically different outputs without partitioning, allowing for improved generalization and fluency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If rule-based sentence rewrite methods are used, then the implementation is simple and straightforward, but the generalization performance is poor and the generated sentences are stiff and non-fluent

Engineering Contradiction:

Improveease of implementationVSAvoidgeneralization performance

Core Design Contradiction:

Ease of manufactureVSReliability

Solution Approach 1:

The patent replaces the mechanical rule-based system with a deep learning neural network model. The rewrite model is trained on parallel sentence corpora to learn transformation patterns, substituting the rigid rule-matching mechanism with a flexible neural network that can generalize to unseen sentences while maintaining fluency in the rewritten output.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Device complexity

If rule-based sentence rewrite methods are used, then the system complexity is low, but the fluency and naturalness of generated sentences deteriorate

Engineering Contradiction:

Improvesystem complexityVSAvoidsentence fluency

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Ease of operation

If sentence partitioning is performed before rewrite, then the rewrite process becomes more manageable, but the semantic coherence and overall fluency of the generated sentence deteriorates

Engineering Contradiction:

Improvemanageability of rewrite processVSAvoidsemantic coherence

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent applies partial action by processing the entire sentence as a single unit rather than partitioning it. The neural network model processes the complete sentence input and generates the rewritten output in one pass, avoiding the semantic fragmentation that would result from dividing the sentence into smaller portions for separate processing.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20230196001A1Sentence conversion techniques

Publication Date: 2023.06.22 BEIJING SOGOU TECHNOLOGY DEVELOPMENT CO LTD

AI summary

Some aspects of the disclosure provide a method for sentence conversion. The method includes receiving a first sentence that is inputted by a user, inputting the first sentence into a first sentence based rewrite model to obtain a second sentence having a same semantic as the first sentence but a different style from the first sentence. The first sentence based rewrite model converts the first sentence into the second sentence without partitioning the first sentence into smaller portions. The method also includes displaying the second sentence. Apparatus and non-transitory computer-readable storage medium counterpart embodiments are also contemplated