Corpus Pattern Paraphrasing via Syntactic Slot Alignment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional techniques for paraphrasing sentences are limited as they rely on stored data in a corpus and cannot generate new paraphrases unless the related sentence is stored, restricting prediction capabilities to database limitations.

Innovation Solution

A corpus pattern paraphrasing system that analyzes sentences to determine regular structures, applies deep learning, and aligns word slots with substitute and representative words to generate paraphrases with the same semantic meaning, independent of stored data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If conventional techniques rely on stored data in a corpus of sentences, then paraphrasing can be performed using available data, but the system cannot generate new paraphrases beyond what is stored in the database

Engineering Contradiction:

Improveparaphrase generation capabilityVSAvoidprediction capability

Core Design Contradiction:

Adaptability or versatilityVSLoss of information

Solution Approach 1:

The system performs preliminary analysis of the corpus to extract regular structures and patterns before actual paraphrasing tasks. By pre-processing the corpus to identify syntactical patterns, semantic relationships, and word substitutions, the system builds a knowledge framework that enables generation of new paraphrases not explicitly stored in the database

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary layer of pattern extraction and semantic analysis between the stored corpus and the paraphrasing output. This intermediary processing layer analyzes syntactical structures, identifies regular patterns, and generates paraphrases based on learned patterns rather than direct retrieval, enabling creation of novel paraphrases while maintaining semantic fidelity

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If the system analyzes regular structures and applies deep learning to generate paraphrases, then new paraphrases can be created beyond stored data, but the system complexity increases

Engineering Contradiction:

Improveparaphrase generation capabilityVSAvoidsystem complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system segments the complex paraphrasing task into distinct functional modules: corpus analysis module for extracting regular structures, pattern recognition module for identifying syntactical patterns, semantic analysis module for understanding meaning relationships, and paraphrase generation module for creating new sentences. This segmentation reduces overall system complexity by making each component specialized and manageable

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system creates universal pattern templates that can be applied across multiple paraphrasing scenarios. By extracting general syntactical patterns and semantic relationships from the corpus, the system develops multi-functional knowledge structures that serve various paraphrasing needs, reducing the need for scenario-specific complex processing

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS11334721B2System, method, and recording medium for corpus pattern paraphrasing

Publication Date: 2022.05.17 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A corpus pattern paraphrasing method, system, and non-transitory computer readable medium, include aligning slots of patterns for verbal phrases based on syntactical and lexical features along with calculated synonyms to predict paraphrases that are not previously stored in a corpus of sentences in a database.