The Distributional Inclusion Hypotheses and Lexical Entailment

By Maayan Zhitomirsky-Geffet, Ido Dagan. 2005

243 citations recorded by the source corpus.

This paper suggests refinements for the Distributional Similarity Hypothesis. Our proposed hypotheses relate the distributional behavior of pairs of words to lexical entailment -- a tighter notion of semantic similarity that is required by many NLP applications. To automatically explore the validity of the defined hypotheses we developed an inclusion testing algorithm for characteristic features of two words, which incorporates corpus and web-based feature sampling to overcome data sparseness. The degree of hypotheses validity was then empirically tested and manually analyzed with respect to the word sense level. In addition, the above testing algorithm was exploited to improve lexical entailment acquisition.

Subfields: Artificial Intelligence

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

Citations in the corpus

Word Representations via Gaussian Embedding (2014, background): “In linguistic semantics, work on the distributional inclusion hypothesis (Geffet & Dagan, 2005), uses traditional count-based vectors to deﬁne regions in vector space (Erk, 2009) such that subordinate concepts are included in these regions.”
Entailment above the word level in distributional semantics (2012, background): “The other strand attempts to reformulate FS’s notion of logical inference in terms that DS can capture (Erk, 2009; Geffet and Dagan, 2005; Kotler-man et al., 2010; Zhitomirsky-Geffet and Dagan, 2010).”
Learning Entailment Rules for Unary Templates (2008, methodology): “With P recision ( l, r ) we obtain a “soft” version of the inclusion hypothesis presented in (Geffet and Dagan, 2005), which expects l to entail r if the “important” features of l appear also in r .”
Learning Semantic Hierarchies via Word Embeddings (2014, background): “Besides Kotlerman et al. (2010) and Lenci and Benotto (2012), other researchers also propose directional distributional similarity methods (Weeds et al., 2004; Geffet and Dagan, 2005; Bhagat et al., 2007; Szpektor et al., 2007; Clarke, 2009).”
Robust Machine Comprehension Models via Adversarial Training (2018, background): “…which are usually calculated based on the distributional hypothesis (Harris, 1954), and do not capture lexical semantic relations such as antonymy (Geffet and Dagan, 2005 (Fellbaum, 1998)) in the context/question, allowing the model to use lexical semantics directly instead of learned statistical…”
Context-theoretic Semantics for Natural Language: an Overview (2009, background): “…is based on the concept of distributional generality (Weeds et al., 2004), a generalisation of the distributional hypothesis of Harris (1985), in which it is assumed that terms with a more general meaning will occur in a wider array of contexts, an idea later developed by Geffet and Dagan (2005).”
Directional distributional similarity for lexical inference (2010, background): “The first one corresponds to valid entailment relations, both substitutable (Geffet and Dagan 2005) and nonsubstitutable (Mirkin et al.”
Context Mover's Distance & Barycenters: Optimal transport of contexts for building representations (2018, methodology): “The intuitive idea for the applicability of our method to this task originates from the Distributional Inclusion Hypothesis (Geffet & Dagan, 2005), which states that a word v entails another word w if “the most characteristic contexts of v are expected to be included in all w ’s contexts (but not ne”
A Context-Theoretic Framework for Compositionality in Distributional Semantics (2011, background): “1 Results were comparable to those obtained by Glickman and Dagan (2005) (see Table 2). In this table, Accuracy is the accuracy on the test set, consisting of 800 entailment pairs, and CWS is the confidence weighted score; see Dagan, Glickman, and Magnini (2005) for the definition.”
Detecting Asymmetric Semantic Relations in Context: A Case-Study on Hypernymy Detection (2017, background): “Finally, the fourth feature captures the distributional inclusion hypothesis (Geffet and Dagan, 2005) – if word v is a hypernym of u , then the set of features of u are included in the set of features of v – by intuitively capturing whether y includes x (Roller et al., 2014).”
Commonsense Knowledge Enhanced Embeddings for Solving Pronoun Disambiguation Problems in Winograd Schema Challenge (2016, background): “The idea to generate such inequality is similar to the physical meaning of lexical entailment [23, 24].”
Combing Context and Commonsense Knowledge Through Neural Networks for Solving Winograd Schema Problems (2016, background): “…pair ( w i , w j ) and ( w k , w g ) and the PMI value of pair ( w i , w j ) is larger than pair ( w k , w g ) , we can make the inequality as: The idea to generate such inequality is similar to the physical meaning of lexical entailment (Geffet and Dagan 2005; Turney and Mohammad 2015).”
Espresso: Leveraging Generic Patterns for Automatically Harvesting Semantic Relations (2006, background): “Recent attention to knowledge-rich problems such as question answering (Pasca and Harabagiu 2001) and textual entailment (Geffet and Dagan 2005) has encouraged natural language processing researchers to develop algorithms for automatically harvesting shallow semantic resources.”