Causality Recognition Apparatus Using Multicolumn Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for recognizing causality in natural language texts face challenges due to the lack of explicit clue terms, leading to low precision and limited coverage, especially in sentences without explicit indicators of causality.
Innovation Solution
A causality recognizing apparatus that uses a combination of word vectors and background knowledge to identify causality between phrases, including a multicolumn neural network that processes candidate phrases, context vectors, binary patterns, answers from a why-type question-answering system, and related passages to determine causality without relying on explicit clue terms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods use clue terms for causality recognition, then precision is improved for sentences with explicit indicators, but coverage deteriorates for sentences without such terms
Solution Approach 1:
The patent introduces multiple intermediary components including background knowledge bases, question-answering systems, and pattern matching mechanisms that mediate between the input text and causality recognition. These intermediaries enable the system to recognize causality without relying solely on explicit clue terms, thereby expanding coverage while maintaining precision through multiple verification pathways.
Solution Approach 2:
The causality recognition system combines multiple diverse information sources and processing methods into a composite approach. It integrates background knowledge, contextual analysis, pattern recognition, and question-answering capabilities to create a robust system that handles both explicit and implicit causality expressions effectively.
2Measurement precision
If the system uses multiple information sources and processing methods, then causality recognition precision is improved, but system complexity increases
Solution Approach 1:
The patent divides the causality recognition system into multiple independent functional modules including background knowledge processing, contextual analysis, pattern matching, and question-answering components. Each module handles a specific aspect of causality recognition, making the overall complex system manageable through clear segmentation and modular design.
Solution Approach 2:
The system employs universal components that serve multiple functions. For example, the background knowledge base supports various types of queries and reasoning tasks, while the question-answering system can handle different types of causal questions. This multi-functionality reduces the need for separate specialized components, managing complexity while maintaining high precision.
Data Source
AI summary
A causality recognizing apparatus includes a candidate vector generating unit configured to receive a causality candidate for generating a candidate vector representing a word sequence forming the candidate; a context vector generating unit generating a context vector representing a context in which noun-phrases of cause and effect parts of the causality candidate appear; a binary pattern vector generating unit, an answer vector generating unit and a related passage vector generating unit, generating a word vector representing background knowledge for determining whether or not there is causality between the noun-phrase included in the cause part and the noun-phrase included in the effect part; and a multicolumn convolutional neural network learned in advance to receive these word vectors and to determine whether or not the causality candidate has causality.