Learning WordNet-Based Classification Rules

TL;DR: A new classification p aradigm, which automatically acquires WordNet-Based rules from a corpus, is presented, which could adapt t o a user’s emotional state and adjust it s interaction etiquette.

Abstract: A new classification p aradigm, which automatically acquires WordNet-Based rules from a corpus, is presented. The a pproach is applied to d eveloping an autonomous s oftware a gent t hat can recognize e motions which are e xpressed in n atural l anguage during an interactive human-computer environment. Such an agent could adapt t o a user’s emotional state a nd d ynamically adjust it s interaction etiquette. Hierarchical concepts of WordNet’s noun and verb h ypernymy are the basic building blocks of the c lassification rules. A greedy learning algorithm automatically determines which hierarchical concepts are best suited for each rule. A corpus of 5000 emotional sentences has been compiled from 502 test subjects and serves as input to the system.