Annotation Assisting Apparatus for Zero Anaphora Resolution

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The existing methods for Japanese anaphora resolution, including zero anaphora resolution, face challenges in achieving high precision and recall due to the high cost and labor-intensive process of creating training data, which requires extensive human judgment and annotation.

Innovation Solution

An annotation assisting apparatus that utilizes morphological analysis, dependency parsing, and language knowledge to automatically detect zero anaphors and their antecedents, providing candidates for human selection, thereby reducing the need for extensive manual annotation and improving efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual annotation is used to create training data for (zero-)anaphora resolution, then high precision and recall can be achieved, but the cost and time required become prohibitively high

Engineering Contradiction:
Improveprecision of anaphora resolutionVSAvoidtime for creating training data
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary automated annotation using morphological analysis, dependency parsing, and language knowledge to generate candidate annotations before human review. This preliminary action reduces the time and effort required for manual annotation while maintaining high precision, as annotators only need to review and correct candidates rather than create annotations from scratch

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary automated annotation layer that generates candidate annotations using morphological analysis, dependency parsing, and language knowledge. This intermediary provides high-quality candidates that require minimal human correction, thus achieving high precision while significantly reducing the time and cost of training data creation

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If manual annotation is used to create training data for (zero-)anaphora resolution, then high precision and recall can be achieved, but the cost becomes prohibitively high

Engineering Contradiction:
Improveprecision of anaphora resolutionVSAvoidcost of creating training data
Core Design Contradiction:
Measurement precisionVSEase of manufacture

Solution Approach 1:

The system performs preliminary automated annotation using morphological analysis, dependency parsing, and language knowledge to generate candidate annotations before human review. This preliminary action reduces the time and effort required for manual annotation while maintaining high precision, as annotators only need to review and correct candidates rather than create annotations from scratch

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary automated annotation layer that generates candidate annotations using morphological analysis, dependency parsing, and language knowledge. This intermediary provides high-quality candidates that require minimal human correction, thus achieving high precision while significantly reducing the time and cost of training data creation

Inventive Principle:
Principle #24Intermediary (Mediator)

3Loss of time

If automated methods are used for (zero-)anaphora resolution, then the cost and time of training data creation are reduced, but the precision and recall remain insufficient

Engineering Contradiction:
Improvetime for creating training dataVSAvoidprecision of anaphora resolution
Core Design Contradiction:
Loss of timeVSMeasurement precision

Solution Approach 1:

The system implements feedback through interactive review where annotators correct automated annotations and the system learns from these corrections. The feedback loop allows the automated system to improve its precision over time while maintaining efficient training data creation, combining the speed of automation with the precision of human judgment

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system introduces an intermediary automated annotation layer that generates candidate annotations using morphological analysis, dependency parsing, and language knowledge. This intermediary provides high-quality candidates that require minimal human correction, thus achieving high precision while significantly reducing the time and cost of training data creation

Inventive Principle:
Principle #24Intermediary (Mediator)

4Reliability

If more training data is collected to improve (zero-)anaphora resolution performance, then the performance can be enhanced, but the cost and time for data preparation increase

Engineering Contradiction:
Improveperformance of anaphora resolutionVSAvoidtime for preparing training data
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary automated annotation using morphological analysis, dependency parsing, and language knowledge to generate candidate annotations before human review. This preliminary action reduces the time and effort required for manual annotation while maintaining high precision, as annotators only need to review and correct candidates rather than create annotations from scratch

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary automated annotation layer that generates candidate annotations using morphological analysis, dependency parsing, and language knowledge. This intermediary provides high-quality candidates that require minimal human correction, thus achieving high precision while significantly reducing the time and cost of training data creation

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS10157171B2Annotation assisting apparatus and computer program therefor
Publication Date: 2018.12.18 NAT INST OF INFORMATION & COMM TECH
  • US10157171B2 patent drawing
  • US10157171B2 patent drawing
  • US10157171B2 patent drawing

AI summary

An annotation data generation assisting system includes: an input/output device receiving an input through an interactive process; morphological analysis system 380 and dependency parsing system performing morphological and dependency parsing on text data in text archive; first to fourth candidate generating units detecting a zero anaphor or a referring expression in the dependency relation of a predicate in a sequence of morphemes, identifying a position as an object of annotation and estimating candidates of expressions to be inserted by using language knowledge; a candidate DB storing estimated candidates; and an interactive annotation device reading candidates of annotation from candidate DB and annotate a candidate selected by an interactive process by input/output device.