Abstract—This research aims to determine the chain of causation of problem events, especially drug-addiction, expressed by several simple sentences from web documents. The chain of causation determination benefits for the problem-solving system. The research has three problems; how to determine a sentence having causative/effect event concept, how to determine the causative/effect event-concept vector size, and how to determine several consecutive causality relations (each causality is a relation between a causative-event-concept vector and an effect-event-concept vector) occurring as the chain of causation. Therefore, we apply WordCo to solve the cause/effect event concepts. We also use Support Vector Machine and WordCo features to solve the causative-event/effect-event vector size/boundary. We then propose using Naïve Bayes to determine the consecutive causality relations between causative event-concept vectors and effect event-concept vectors. The research results provide the high precision of the chain of causation determination from the documents.
Index Terms—Chain of causation, effect boundary, elementary discourse unit, WordCo.
Cite: Chaveevan Pechsiri and Renu Sukharomana, "Chain of Causation Determination from Texts," International Journal of Machine Learning and Computing vol. 7, no. 5, pp. 94-99, 2017.