Textual Entailment Knowledge Base for NLP Question Answering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional computing systems are limited in understanding and processing unstructured data, and existing natural language processing systems struggle to identify correct answers when the language used in questions does not match the language used in answers, even if the answer exists in the corpus, due to the lack of effective textual entailment capabilities.
Innovation Solution
A natural language processing system generates textual entailment pairs by querying a passage database, selecting anchor pairs, and extracting entailment pairs based on these anchors, using components like passage retrieval, scoring, and entailment pair extraction to identify directional relationships between text fragments, even if they use different words.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional computing systems process unstructured data using preprogrammed functionality, then processing capability is limited, but system complexity remains low
Solution Approach 1:
The patent introduces textual entailment relationships as an intermediary mechanism between questions and answers. This mediator enables the system to bridge linguistic differences by identifying directional relationships between text fragments, allowing answers to be matched even when wording differs from the question
Solution Approach 2:
The system precomputes and stores textual entailment pairs in a database before actual question answering occurs. This preliminary action creates a reusable knowledge base of directional relationships that can be quickly queried during runtime, improving processing capability without adding complexity to the real-time system
2Measurement precision
If natural language processing systems use exact matching for questions and answers, then processing speed is fast, but accuracy decreases when language does not match
Solution Approach 1:
The patent creates copies of textual relationships in the form of precomputed entailment pairs stored in a database. Instead of performing complex semantic analysis during runtime, the system queries these pre-generated copies of relationships, maintaining fast processing while improving accuracy through comprehensive textual relationship matching
Data Source
AI summary
Generating textual entailment pair by a natural language processing (NLP) system. The NLP system receives two input texts, such as a question and a candidate answer. The NLP system queries a database and retrieves passages likely to include text that support the candidate answer. The NLP system generates parse trees and performs term matching on the passages and scores them according to the matching. The NLP system detects anchor pairs in the question and in the passage and aligns subgraphs (within the parse trees) of one to the other based on matching. The NLP system identifies aligned terms in the question and the passage that are not in the aligned subgraphs. The NLP system identifies text fragments, for the question and the passage, within the non-aligned segments of their respective parse trees, that connect the aligned term to the aligned portion of the subgraph.