What Substitutes Tell Us - Analysis of an “All-Words” Lexical Substitution Corpus

By Gerhard Kremer, K. Erk, Sebastian Padó, Stefan Thater. 2014

98 citations recorded by the source corpus.

We present the first large-scale English “allwords lexical substitution” corpus. The size of the corpus provides a rich resource for investigations into word meaning. We investigate the nature of lexical substitute sets, comparing them to WordNet synsets. We find them to be consistent with, but more fine-grained than, synsets. We also identify significant differences to results for paraphrase ranking in context reported for the SEMEVAL lexical substitution data. This highlights the influence of corpus construction approaches on evaluation results.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

context2vec: Learning Generic Context Embedding with Bidirectional LSTM (2016, background): “The second is a more recent ‘all-words ’ dataset (Kremer et al., 2014), denoted LST-14, with over 15K target word instances.”
SenticNet 6: Ensemble Application of Symbolic and Subsymbolic AI for Sentiment Analysis (2020, methodology): “As for lexical substitution evaluation datasets, we used the LST-07 dataset from the lexical substitution task of the 2007 Semantic Evaluation (SemEval) challenge [34] and the 15,000 target word all-words LST-14 dataset from SemEval-2014 [30].”
SenticNet 5: Discovering Conceptual Primitives for Sentiment Analysis by Means of Context Embeddings (2018, methodology): “As lexical substitution datasets, we used the LST-07 dataset from the SemEval 2007 lexical substitution task (McCarthy and Navigli 2007) and the 15-thousand target word all-words LST-14 dataset from SemEval 2014 (Kremer et al. 2014).”
A Simple Word Embedding Model for Lexical Substitution (2015, background): “Lexical substitution tasks have become very popular for evaluating context-sensitive lexical inference models since the introduction of the original task in SemEval-2007 (McCarthy and Navigli, 2007) and additional later variants (Biemann, 2013; Kremer et al., 2014).”
Modeling Word Meaning in Context with Substitute Vectors (2015, methodology): “This behavior is consistent with similar ﬁndings in Kremer et al. (2014).”
Annotating Relation Inference in Context via Question Answering (2016, background): “il/ ̃nlp/resources/ downloads/relation_inference_via_qa It is worth noting the lexical substitution datasets (McCarthy and Navigli, 2007; Biemann, 2013; Kremer et al., 2014) also capture instances of relation inference.”