Convolutional Neural Network Lexical Analysis via Multi-Length Word Windows
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting specific words related to event information from text, while considering surrounding context, face a trade-off where increasing the window length improves ambiguity resolution but decreases accuracy in word position detection.
Innovation Solution
An information processing apparatus and method that generate training word groups using windows of varying lengths, associate words with semantic vectors, and update a convolutional neural network model to accurately identify specific words by optimizing probabilities based on these groups.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the window length is increased to extract more surrounding context words, then the ambiguity of meaning is resolved better, but the accuracy of word position detection decreases
Solution Approach 1:
The patent divides the single window extraction process into multiple segmentation levels by creating word groups with different window lengths (e.g., window length 1, 2, 3). Each segmentation captures context at different granularities, allowing the system to resolve ambiguity through multi-scale analysis while maintaining position accuracy through hierarchical processing.
Solution Approach 2:
The patent adds a dimensional aspect to window length by processing the same text with multiple window sizes simultaneously. This transforms the single-dimension problem (one window length) into a multi-dimensional solution space where different window lengths provide complementary information for both ambiguity resolution and position detection.
2Quantity of substance
If multiple windows of different lengths are used to extract word groups, then the proportion of positive examples increases, but the processing complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing the text to create multiple word groups with different window lengths before the main classification task. This preliminary segmentation organizes the data in a way that increases positive example proportion while structuring the complexity in a manageable, hierarchical manner that reduces overall processing burden.
Data Source
AI summary
An information processing apparatus includes a lexical analysis unit that generates a training word string, a group generation unit that generates a plurality of training word groups, a matrix generation unit that generates, for each training word group, a training matrix in which a plurality of words and respective semantic vectors of the words are associated, a classification unit that calculates, for a word of each position of the training word string, a probability of the word corresponding to a specific word, using the training matrices generated by the matrix generation unit and a determination model that uses a convolutional neural network, and an optimization processing unit that updates parameters of the determination model, such that the probability of the word labeled as corresponding to the specific word is high, among the probabilities of the words of the respective positions of the training word string calculated by the classification unit.