Idest: Learning a Distributed Representation for Event Patterns

Human Language Technologies: The 2015 Annual Conference of the North American Chapter of the ACL, pages 1140–1149, Denver, Colorado, May 31 – June 5, 2015. c⃝2015 Association for Computational Linguistics IDEST: Learning a Distributed Representation for Event Patterns Sebastian Krause∗ LT Lab, DFKI Alt-Moabit 91c 10559 Berlin, Germany skrause@dfki.de Enrique Alfonseca Katja Filippova Daniele Pighin Google Research Brandschenkestrasse 110 8810 Zurich, Switzerland {ealfonseca,katjaf,biondo}@google.com Abstract This paper describes IDEST, a new method for learning paraphrases of event patterns. It is based on a new neural network architecture that only relies on the weak supervision sig- nal that comes from the news published on the same day and mention the same real-world en- tities. It can generalize across extractions from different dates to produce a robust paraphrase model for event patterns that can also cap- ture meaningful representations for rare pat- terns. We compare it with two state-of-the-art systems and show that it can attain compara- ble quality when trained on a small dataset. Its generalization capabilities also allow it to leverage much more data, leading to substan- tial quality improvements. 1 Introduction Most Open Information Extraction (Open-IE) sys- tems (Banko et al., 2007) extract textual relational patterns between entities automatically (Fader et al., 2011; Mausam et al., 2012) and optionally organize them into paraphrase clusters. These pattern clusters have been found to be useful for Question Answer- ing (Lin & Pantel, 2001; Fader et al., 2013) and re- lation extraction (Moro & Navigli, 2012; Grycner & Weikum, 2014), among other tasks. A related Open-IE problem is that of automati- cally extracting and paraphrasing event patterns: those that describe changes in the state or attribute values of one or several entities. An existing ap- proach lo learn paraphrases of event patterns is to build on the following weak supervision signal: ∗Work performed during an internship at Google news articles that were published on the same day and mention the same entities should contain good paraphrase candidates. Two state-of-the-art event paraphrasing systems that are based on this assump- tion are NEWSSPIKE (Zhang & Weld, 2013) and HEADY (Alfonseca et al., 2013; Pighin et al., 2014). These two systems have a lot in common, yet they have never been compared with each other. They have specific weak and strong points, and there are many ways in which they are substantially different: • Scope of generalization. In NEWSSPIKE the paraphrase clusters are learned separately for each publication day and entity set, and the sys- tem cannot generalize across events of the same type involving different entities occurring on the same or on different days. For example, if the event verbs has married and wed appear in news about two entities A and B marrying, and has married and tied the knot with appear in news involving two different entities C and D, NEWSSPIKE is not able to infer that wed and tied the knot with are also paraphrases, unless a post-processing is done. HEADY overcomes this limitation thanks to a global model that learns event …

Content truncated to 500 words