TRIPS and TRIOS System for TempEval-2: Extracting Temporal Information from Text

Proceedings of the 5th International Workshop on Semantic Evaluation, ACL 2010, pages 276–283, Uppsala, Sweden, 15-16 July 2010. c⃝2010 Association for Computational Linguistics TRIPS and TRIOS System for TempEval-2: Extracting Temporal In- formation from Text Naushad UzZaman University of Rochester Rochester, NY, USA naushad@cs.rochester.edu James F. Allen University of Rochester Rochester, NY, USA james@cs.rochester.edu Abstract Extracting temporal information from raw text is fundamental for deep language understand- ing, and key to many applications like ques- tion answering, information extraction, and document summarization. In this paper, we describe two systems we submitted to the TempEval 2 challenge, for extracting temporal information from raw text. The systems use a combination of deep semantic parsing, Markov Logic Networks and Conditional Random Field classifiers. Our two submitted systems, TRIPS and TRIOS, approached all tasks and outperformed all teams in two tasks. Furthermore, TRIOS mostly had second-best performances in other tasks. TRIOS also out- performed the other teams that attempted all the tasks. Our system is notable in that for tasks C – F, they operated on raw text while all other systems used tagged events and tem- poral expressions in the corpus as input. 1 Introduction The recent emergence of language processing applications like question answering, information extraction, and document summarization has drawn attention to the need for systems that are temporally aware. For example, for a QA system in newswire domain, if we want to know who was the President of Bangladesh in the January of 1983, and we only had documents that tell us about the President from 1980 to 1985 then a temporally aware system will help the QA sys- tem to infer who was president in the January of 1983 as well. In medical domain for patient’s history record, doctors write all the information about patients’ medical record, usually not in chronological order. Extracting a temporal struc- ture of the medical record will help practitioner understand the patient’s medical history easily. For people who have trouble reading and under- standing, be it dyslexic people, or people who are not native English speakers, a temporal structure of document could help them to follow the story better. Extracting temporal information will benefit in almost any application processing natural language text. In this paper, we present the first step towards our goal of building such temporal structures. We participated in all tasks in TempEval 2, which includes work on extracting events, event features, temporal expressions, and various tem- poral relations. We first describe our systems. Next, we show the performance of our system and compare with best performing systems on TempEval-2. Fi- nally, we describe our future directions. 2 Our System Modules Our approach for all the tasks is best described as hybrid between linguistically motivated solutions and machine learning classifiers. We do deep semantic parsing and use hand-coded rules to extract events, features and temporal expressions from the logical forms produced by the parser. In parallel, we filter events, extract event features, temporal expressions, classify temporal relations using machine-learning classifiers. We describe …

Content truncated to 500 words