Tagging Model Generation for Phrase Dependency Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for attaching tags to text, such as named entity extraction using CRF, fail to recognize phrases based on word dependencies, leading to incomplete recognition of locations like "TOUKYOU NI ARU TAWA" as a single location in systems like automated response systems.
Innovation Solution
A tagging model generation apparatus and method that considers word dependencies by generating a tagging model using learning data with word-related information, including part-of-speech information, and joint probability-related information to accurately attach tags to phrases in text.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional CRF-based named entity extraction is used, then individual words can be tagged with location names, but phrases based on word dependencies cannot be recognized collectively as locations
Solution Approach 1:
The patent segments the tagging problem into two parts: individual word tagging using CRF and phrase-level dependency recognition using a separate module. The text is processed to identify both word-level tags and phrase-level relationships, allowing collective recognition of location phrases while maintaining individual word tagging accuracy.
Solution Approach 2:
The patent introduces an intermediary component that bridges CRF-based word tagging and phrase-level recognition. This intermediary module analyzes word dependencies and relationships, enabling the system to recognize phrases like 'TOUKYOU NI ARU TAWA' as collective locations while preserving the underlying word-level tagging information.
2Productivity
If tag attachment is performed based on individual word features, then processing speed is maintained, but phrase-based location recognition accuracy deteriorates
Solution Approach 1:
The patent performs preliminary word-level tagging using CRF before conducting phrase-level dependency analysis. This preliminary action enables fast initial processing while setting the stage for subsequent phrase recognition, maintaining overall processing efficiency without sacrificing phrase-based location accuracy.
Solution Approach 2:
The patent implements a dynamic two-stage processing system where the first stage (CRF) operates at high speed for individual words, and the second stage (dependency analysis) dynamically processes phrase relationships. This dynamic approach balances processing speed with phrase recognition accuracy by applying different processing intensities at different stages.
Data Source
AI summary
Provided is a technique for generating a tagging model for attaching a tag in consideration of a phrase based on dependency between words. A tagging model generation apparatus includes a learning section 2 which generates, by using inputted learning data, a tagging model including probability-related information serving as information related to the probability that each tag is associated with each word-related information, and joint probability-related information serving as information related to a joint probability which serves as the probability of appearance of each tag in which appearance frequencies of a plurality of consecutive tags associated with pieces of word-related information of a plurality of consecutive words in each text are taken into consideration, and a storage section 3 which stores the generated tagging model.