A Human-Machine Language Dictionary

• Main Authors: Fei Liu, Shirin Akther Khanam, Yi-Ping Phoebe Chen
• Format: Article
• Language: English
• Published: Atlantis Press 2020-06-01
• Series: International Journal of Computational Intelligence Systems
• Subjects: Text mining; Natural language processing; Knowledge representation
• Online Access: https://www.atlantis-press.com/article/125941276/view

In this paper, we propose a framework for building a human-machine language dictionary. Given a concept/word, an application can extract the definition of the concept from the dictionary, and consequently “understand” its meaning. In the dictionary, a concept is defined through its relations with other concepts. Relations are specified in the machine language. To a certain degree, the proposed dictionary has a resemblance to WordNet, which consists of a set of concepts/words with synonyms being linked to form the net. WordNet plays an important role in text mining, such as sentiment analysis, document classification, text summarization and question answering systems, etc. However, merely providing synonyms is not sufficient. The proposed dictionary provides a definition for each concept. Based on the definition, the application can accurately estimate the distance and similarity between concepts. As a monotonic mapping, the algorithm for estimating distances and similarities is proved to be always convergent. We envisage that the dictionary will become an important tool in all Text Mining disciplines.