Temporal Event Detection Using Supervised Machine Learning Based Algorithm

Natural Language Processing is a way for computers to explore, analyze, comprehend, and derive significant sense from any language in a smart and useful way. By using NLP, knowledge can be organized and structured to perform tasks such as automatic summarization, translation, named entity recognition, relationship extraction, sentiment analysis, speech recognition, and topic segmentation. Various NLP applications require the identification of events from text documents. The time and event are closely associated with each other. The time dimension is often used to measure the quality and value of events and it has a strong influence in many domains like topic-detection and tracking, query log analysis. In this work, we present an annotation framework to extract temporal information and to specify the temporal relation between extracted events from news corpus by applying the combination of supervised machine learning technique and rule-based method according to the TimeML task std. Artificial neural network (ANN) is trained by the using the TimeBank and AQUAINT TimeML corpus to recognize the events and temporal expressions and for the temporal normalization, part heuristics rules have been used. The efficiency of the proposed work is measured in sense of precision and recall. The system outperformed to the best systems and it is likely that the technique used could be improved further by considering more aspects of the available information when relating the temporal information with events.