PROJECT TITLE :

A Natural Language Process-Based Framework for Automatic Association Word Extraction

ABSTRACT:

In psychology, word association has been extensively explored for exposing mental representations and relationships of humans. However, due to the old manually collected methods, the number of possible associative cue-response words is severely limited. Meanwhile, due to the significant progress in Natural Language Processing (NLP) jobs, a vast number of plain texts may be collected quickly. This shows that instead of manually collecting association terms, it may be possible to find them automatically from a text corpus. This research makes a little step toward presenting a deep learning-based framework for automatic association word extraction as an original approach. The framework is divided into two stages: detection of connection words and creation of machine association networks. The use of an attention mechanism-based Reading Comprehension (RC) algorithm to automatically find useful association terms is investigated. The correlation coefficient between semantic similarities of machine and human association words is introduced as an appropriate assessment for analyzing association consistency to validate the value of the retrieved association words. The studies are carried out on two text datasets, from which over 20k association words are derived, which is more than the largest human association word dataset currently available. The experiment also shows that machine association words are often compatible with human association words in terms of semantic similarity, highlighting the potential for machine association words to be used in future psychology and NLP research.