Script Model Construction via Shared Argument Feature Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional script model construction methods require large amounts of high-reliability learning data, leading to high costs and low accuracy due to insufficient data collection, particularly failing to provide adequate clues for predicting the coherence of event slots, resulting in low accuracy contextual analysis.
Innovation Solution
A data processing device and method that extracts event slot sequences with shared arguments from documents, generates case example data using feature vectors that include event slot and shared argument histories, and constructs a script model using a discriminative model for machine learning, integrating frequency statistics of both sequences to predict the coherence of event slots.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a huge amount of learning data is collected to construct a script model with high accuracy, then the accuracy of the script model is improved, but the cost and time required for data collection increase significantly
Solution Approach 1:
The patent segments the learning data into two distinct components: event slot sequences and shared argument sequences. By separating these components, the system can collect and process data more efficiently, reducing the overall time and cost while maintaining model accuracy through targeted data collection for each component
Solution Approach 2:
The patent introduces a new dimension to the data structure by explicitly separating event slot sequences from shared argument sequences. This dimensional separation allows for more efficient data processing and model training, reducing the data collection burden while improving model performance
2Ease of manufacture
If conventional script model construction methods are used, then the process is simple, but the accuracy is low due to insufficient data reliability and lack of coherence prediction clues
Solution Approach 1:
The patent segments the data processing into distinct modules: event slot sequence extraction, shared argument sequence extraction, and model construction. This segmentation maintains process simplicity while improving accuracy by ensuring each component is processed and analyzed separately with appropriate attention to coherence relationships
Solution Approach 2:
The patent introduces shared argument sequences as an intermediary element that bridges event slot sequences and provides additional coherence clues. This intermediary component enables better prediction of event slot coherence while maintaining a relatively simple overall construction process
Data Source
AI summary
According to an embodiment, a data processing device includes an extractor, a generator, and a constructor. The extractor is configured to extract, from a document having been subjected to predicate argument structure analysis and anaphora resolution, an element sequence including elements each being a combination of predicate having a shared argument and case type information of the shared argument, together with the shared argument. The generator is configured to produce case example data expressed by a feature vector for each attention element which is one of the elements. The feature vector includes feature value(s) about a sub-sequence having the attention element and feature value(s) about a sequence of the shared argument corresponding to the sub-sequence. The constructor is configured to construct a script model for estimating the elements each following antecedent context by performing machine learning based on a discriminative model using the case example data.