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

Kremer, Gerhard; Erk, Katrin; Padó, Sebastian; Thater, Stefan

doi:10.3115/v1/e14-1057

Cited by 86 publications

(107 citation statements)

References 19 publications

Supporting

Mentioning

103

Contrasting

Order By: Relevance

“…9 All (max, no co-clustering) 0.250* 0.695* 0.690* Choose-K: # WN Synsets (max, co-clustering) 0.241** 0.690** 0.683** Choose-K: Optimize NMI (avg) 0.282 0.668 0.662 Choose-K: Optimize NMI (max, no co-clustering) 0.331* 0.719 * 0.714 *** Choose K: Optimize NMI (max, co-clustering) 0.314** 0.718 **** 0.710 ** baseline and cluster sense inventories are capable of improving these GAP scores when we use the best sense as a filter. Syntactic models generally give very good results with small paraphrase sets (Kremer et al, 2014) but their performance seems to degrade when they need to deal with larger and noisier substitute sets (Apidianaki, 2016). Our results suggest that finding the most appropriate sense of a target word in context can improve their lexical substitution results.…”

Section: Resultsmentioning

confidence: 62%

See 1 more Smart Citation

Word Sense Filtering Improves Embedding-Based Lexical Substitution

Cocos¹,

Apidianaki²,

Callison-Burch³

2017

Proceedings of the 1st Workshop on Sense, Concept and Entity Representations and Their Applications

View full text Add to dashboard Cite

The role of word sense disambiguation in lexical substitution has been questioned due to the high performance of vector space models which propose good substitutes without explicitly accounting for sense. We show that a filtering mechanism based on a sense inventory optimized for substitutability can improve the results of these models. Our sense inventory is constructed using a clustering method which generates paraphrase clusters that are congruent with lexical substitution annotations in a development set. The results show that lexical substitution can still benefit from senses which can improve the output of vector space paraphrase ranking models.

show abstract

Section: Resultsmentioning

confidence: 62%

“…The first is the "Concepts in Context" (CoInCo) corpus (Kremer et al, 2014), containing over 15K sentences corresponding to nearly 4K unique target words. We divide the CoInCo dataset into development and test sets by first finding all target words that have at least 10 sentences.…”

Section: Datasetsmentioning

confidence: 99%

Word Sense Filtering Improves Embedding-Based Lexical Substitution

Cocos¹,

Apidianaki²,

Callison-Burch³

2017

Proceedings of the 1st Workshop on Sense, Concept and Entity Representations and Their Applications

View full text Add to dashboard Cite

show abstract

“…However, as also reported in Kremer et al (2014), the performance gain achieved by taking the given context into consideration is smaller than in LS07. Again, this seems to be due to the nature of LS14, which is not biased to ambiguous target words.…”

Section: Resultsmentioning

confidence: 73%

“…A more recent dataset (Kremer et al, 2014), denoted LS14, provides the same kind of data as LS07, but instead of target words that were specifically selected to be ambiguous as in LS07, the target words here are simply all the content words in text documents extracted from news and fiction corpora. LS14 is also much larger than LS07 with over 15K target word instances.…”

Section: Lexical Substitution Datasetsmentioning

confidence: 99%

Modeling Word Meaning in Context with Substitute Vectors

Melamud

Dagan

Goldberger

2015

Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

View full text Add to dashboard Cite

Context representations are a key element in distributional models of word meaning. In contrast to typical representations based on neighboring words, a recently proposed approach suggests to represent a context of a target word by a substitute vector, comprising the potential fillers for the target word slot in that context. In this work we first propose a variant of substitute vectors, which we find particularly suitable for measuring context similarity. Then, we propose a novel model for representing word meaning in context based on this context representation. Our model outperforms state-of-the-art results on lexical substitution tasks in an unsupervised setting.

show abstract

“…Lexical substitution systems perform substitute ranking in context using vector-space models (Thater et al, 2011;Kremer et al, 2014;Melamud et al, 2015). Recently, Apidianaki (2016) showed that a syntax-based substitution model can successfully filter the paraphrases available in the Paraphrase Database (PPDB) (Ganitkevitch et al, 2013) to select the ones that are adequate in specific contexts.…”

Section: Related Workmentioning

confidence: 99%

Simplification Using Paraphrases and Context-Based Lexical Substitution

Kriz¹,

Miltsakaki²,

Apidianaki³

et al. 2018

Proceedings of the 2018 Conference of the North American Chapter Of the Association for Computational Linguistics: Hu

View full text Add to dashboard Cite

Lexical simplification involves identifying complex words or phrases that need to be simplified, and recommending simpler meaningpreserving substitutes that can be more easily understood. We propose a complex word identification (CWI) model that exploits both lexical and contextual features, and a simplification mechanism which relies on a wordembedding lexical substitution model to replace the detected complex words with simpler paraphrases. We compare our CWI and lexical simplification models to several baselines, and evaluate the performance of our simplification system against human judgments. The results show that our models are able to detect complex words with higher accuracy than other commonly used methods, and propose good simplification substitutes in context. They also highlight the limited contribution of context features for CWI, which nonetheless improve simplification compared to contextunaware models.

show abstract

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

Cited by 86 publications

References 19 publications

Word Sense Filtering Improves Embedding-Based Lexical Substitution

Word Sense Filtering Improves Embedding-Based Lexical Substitution

Modeling Word Meaning in Context with Substitute Vectors

Simplification Using Paraphrases and Context-Based Lexical Substitution

Contact Info

Product

Resources

About