A Corpus of Preposition Supersenses

Abstract

We present the first corpus annotated with preposition supersenses, unlexicalized categories for semantic functions that can be marked by English prepositions . That scheme improves upon its predecessors to better facilitate comprehensive manual annotation. Moreover, unlike the previous schemes, the preposition supersenses are organized hierarchically. Our data will be publicly released on the web upon publication.

Proceedings of the 11th Linguistic Annotation Workshop

“…Verb resources such as VerbNet and PropBank include verbs of causation. Likewise, preposition schemes (e.g., Schneider et al, 2015Schneider et al, , 2016 include some purpose-and explanationrelated senses. None of these, however, unifies all linguistic realizations of causation into one framework; they are concerned with specific classes of words, rather than the semantics of causality.…”

Proceedings of the 6th Joint Conference on Lexical and Computational Semantics (*SEM 2017)

“…The composition of the DiMSUM corpus (Schneider et al, 2016), and the token-level lemma and part-of-speech annotations it provides, influenced our feature extraction. Most of the text in the DiMSUM corpus is social media text.…”

“…We use the DiMSUM dataset (Schneider et al, 2016) for our experiments, which allows for direct comparison with previous results. Table 1 displays the source corpora from which the dataset was constructed; their domain (i.e., reviews, tweets, or TED talks); the number of sentences, words, MWEs, and gappy (i.e., discontiguous) MWES in each source corpus; and the percentage of tokens belonging to an MWE in each source corpus.…”

“…More recently, Constant and Nivre (2016) incorporate MWE identification along with dependency parsing by forming two representations for a sentence: a tree that represents the syntactic dependencies, and a forest of lexical trees that includes the MWEs identified in the sentence. The recent SemEval shared task on Detecting Minimal Semantic Units and their Meanings (DiMSUM) focused on MWE identification along with supersense tagging (Schneider et al, 2016). The best performing system for MWE identification for this shared task was that of Kirilin et al (2016) which took into consideration all of the basic features used by Schneider et al (2014a) and two novel feature sets.…”

Double Trouble: The Problem of Construal in Semantic Annotation of Adpositions

“…Preliminary analysis suggests that this scheme will work well not only for English, but also for the other languages examined. §5 surveys some of the phenomena that our new analysis addresses with examples from multiple languages; §6 suggests that this added flexibility at the token level removes the need for a great deal of complexity in the supersense inventory itself: i.e., we can get away with Figure 1: Preposition supersense hierarchy (from Schneider et al, 2016). Top-level categories are circled and subcategories radiate outward.…”