Constructing a Lexicon of Relational Nouns

Constructing a Lexicon of Relational Nouns Edward Newell, Jackie C.K. Cheung edward.newell@mail.mcgill.ca jcheung@cs.mcgill.ca McGill University Abstract Relational nouns refer to an entity by virtue of how it relates to another entity. Their identification in text is a prerequisite for the correct semantic interpretation of a sentence, and could be used to improve information extraction. Although various systems for extracting relations expressed using nouns have been developed, there are no dedicated lexical resources for relational nouns. We contribute a lexicon of 6,224 labeled nouns which includes 1,446 relational nouns. We describe the bootstrapped annotation of relational nouns, and develop a classifier that achieves 70.4% F1 when tested on held out nouns that are among the most common 2,500 word types in Gigaword. We make the lexicon and classifier available to the scientific community. Keywords: relational noun, relation extraction, possessive, lexicon 1. Introduction The extraction of relations is a fundamental aspect of natu- ral language understanding, playing a central role in knowl- edge base construction, question answering, and recogniz- ing textual entailment. Semantic relations can be expressed via a syntactically diverse set of constructions. Consider the following approximate paraphrases: (1a) Jack befriended Jill 5 years ago. (1b) Jack and Jill’s friendship is now 5 years old. (1c) Jack has been Jill’s friend for 5 years. Whereas (1a) establishes the relation by means of a verb, (1b) and (1c) do so by means of nouns. In example (1c), friend is part of a semantically motivated class of nouns called relational nouns, which we focus on in this work. Relational nouns refer to an entity by virtue of how it re- lates to something else (Barker, 2011). In the above exam- ple, friend establishes a relation between its referent, Jack, and the external entity, Jill, which could be depicted using a two-place predicate such as friend(Jack, Jill). This differs from how a sortal (i.e. non-relational) noun like person is interpreted. The sentence Jill is a person could be depicted as a unary predicate, person(Jill). In other words, there is an absolute set of people, but there is no absolute set of friends, without first specifying the person with whom they are friends. Despite this semantic difference, relational nouns behave syntactically like other nouns. Semantic parsers such as Boxer that are trained on CCGBank do not currently distin- guish between relational and non-relational nouns, leading to errors in sentences that contain them (Bos, 2008). Re- lation extraction systems such as RENOUN (Yahya et al., 2014) and RELNOUN (Pal and Mausam, 2016) rely on au- tomatically extracted patterns, and also do not make a dis- tinction between relational and non-relational nouns. Var- ious information extraction systems could be improved by the identification of relational nouns. In this work, we cre- ate a high-quality lexicon of relational nouns using boot- strapped manual annotation. Relational nouns pose a conceptually difficult annotation task, in part because the meaning of most nouns involves some relation at least indirectly. Part of our contribution is a series of decisions about what should …

Content truncated to 500 words