How Large Are Lions? Inducing Distributions over Quantitative Attributes

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3973–3983 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 3973 How Large Are Lions? Inducing Distributions over Quantitative Attributes Yanai Elazar∗ Bar Ilan Univesity yanaiela@gmail.com Abhijit Mahabal† Pinterest amahabal@gmail.com Deepak Ramachandran Google Research ramachandrand@google.com Tania Bedrax-Weiss Google Research tbedrax@google.com Dan Roth University of Pennsylvania danroth@seas.upenn.edu Abstract Most current NLP systems have little knowl- edge about quantitative attributes of objects and events. We propose an unsupervised method for collecting quantitative information from large amounts of web data, and use it to create a new, very large resource consisting of distributions over physical quantities associ- ated with objects, adjectives, and verbs which we call Distribution over Quantities (DOQ)1. This contrasts with recent work in this area which has focused on making only relative comparisons such as “Is a lion bigger than a wolf?”. Our evaluation shows that DOQ com- pares favorably with state of the art results on existing datasets for relative comparisons of nouns and adjectives, and on a new dataset we introduce. 1 Introduction How much does a lion weigh? How tall can they be? When do people typically eat break- fast? And, how long are concerts? Most peo- ple would know at least an approximate answer to these questions, many of which fall under the (somewhat ill-defined) notion of commonsense knowledge, and some (but certainly not all) of which exist in resources such as Wikipedia, or in knowledge graphs like Freebase (Bollacker et al., 2008). Natural Language Understanding systems should also know (at least approximately) the an- swers to these questions, to better support Ques- tion Answering and Textual Entailment (Dagan ∗Work carried out during an internship at Google. †Work carried out during employment at Google. 1The resource is available at https:// github.com/google-research-datasets/ distribution-over-quantities wolf jaguar tiger lion 0 50 100 150 200 250 Mass (in kg.) (a) Mass distributions for multiple animals. breakfast brunch lunch dinner 8 10 12 14 16 18 Hour of day (b) Typical hours for different meals of the day. Figure 1: Examples of different objects from DOQ in the MASS and TIME dimensions. et al., 2013) and, more generally, in order to sup- port reasoning about events described in natural language and converse with people naturally. Acquiring commonsense knowledge from natu- ral language text has been the subject of a lot of recent work. These approaches focus on facili- tating comparisons between quantitative attributes of nouns (Bagherinezhad et al., 2016; Forbes and 3974 Choi, 2017; Yang et al., 2018), intensity of ad- jectives (De Melo and Bansal, 2013; Cocos et al., 2018) and coarse classification of events duration and relative order (Gusev et al., 2011; Ning et al., 2018). However, they do not have complete cover- age even for comparable objects, as a result of how they are acquired, and lack the ability to assign a numerical value to objects and events (“How hot is it in New York?”), which is often useful for rea- …

Content truncated to 500 words