Causal web determination from texts

Abstract

The research aim is to determine a causal web from downloaded guru web-board documents. The causal web which benefits a diagnosis service assistant of a problem-solving system consists of several cause-effect pair sequences where each cause-effect pair has a cause-effect relation and the last cause-effect pair of each cause-effect pair sequence has the same effect concept. Each causative/effect concept is expressed by an elementary discourse unit or a simple sentence. The research has three problems; how to determine the cause-effect pair with an overlap problem between a causative-verb concept set and an effect-verb concept set, how to determine cause-effect pair sequences including causative/effect boundary determination, and how to determine the causal web on the extracted cause-effect pair sequences without redundant sequences. We use a word co-occurrence to represent a sentence’s event/state with a causative/effect concept. We then propose using a self-Cartesian product on a collected word co-occurrence set and Naïve Bayes including categorized verb groups to extract each cause-effect pair sequence including the boundary determination without the verb-concept-overlap influence. And we use a dynamic template matching technique to determine the causal web without the redundancy. The research result has a high percentage correctness of the causal web determination. © 2020 J. Adv. Inf. Technol.