Multilingual and Cross-Lingual Graded Lexical Entailment

Abstract

Grounded in cognitive linguistics, graded lexical entailment (GR-LE) is concerned with finegrained assertions regarding the directional hierarchical relationships between concepts on a continuous scale. In this paper, we present the first work on cross-lingual generalisation of GR-LE relation. Starting from Hyper-Lex, the only available GR-LE dataset in English, we construct new monolingual GR-LE datasets for three other languages, and combine those to create a set of six cross-lingual GR-LE datasets termed CL-HYPERLEX. We next present a novel method dubbed CLEAR (Cross-Lingual Lexical Entailment Attract-Repel) for effectively capturing graded (and binary) LE, both monolingually in different languages as well as across languages (i.e., on CL-HYPERLEX). Coupled with a bilingual dictionary, CLEAR leverages taxonomic LE knowledge in a resource-rich language (e.g., English) and propagates it to other languages. Supported by cross-lingual LE transfer, CLEAR sets competitive baseline performance on three new monolingual GR-LE datasets and six cross-lingual GR-LE datasets. In addition, we show that CLEAR outperforms current state-ofthe-art on binary cross-lingual LE detection by a wide margin for diverse language pairs.

Multi-SimLex: A Large-Scale Evaluation of Multilingual and Crosslingual Lexical Semantic Similarity

“…Table 2 summarizes the inter-translator agreement results for all languages and by part-of-speech subsets. Overall across all languages, the agreement is 84.8%, which is similar to prior work Vulić, Ponzetto, and Glavaš 2019).…”

“…Therefore, the similarity between L 1 -ENG and L 2 -ENG is expected to be higher than between L 1 -L 2 , especially if L 1 and L 2 are typologically dissimilar languages (e.g., HEB-CMN, see Figure 1). This phenomenon is well documented in related prior work (Leviant and Reichart 2015;Mrkšić et al 2017;Vulić, Ponzetto, and Glavaš 2019). Although we acknowledge this as an artifact of the data set design, it would otherwise be impossible to construct a semantically aligned and comprehensive data set across a large number of languages.…”

“…To assess the quality of the resulting crosslingual data sets, we have conducted a verification experiment similar to Vulić, Ponzetto, and Glavaš (2019). We randomly sampled 300 concept pairs in the English-Spanish, English-French, and English-Mandarin crosslingual data sets.…”

“…corresponding monolingual scores (s s , s t ) differ at most by one fifth of the full scale (i.e., | s s − s t |≤ 1.2). This heuristic mitigates the noise due to crosslingual semantic shiftsVulić, Ponzetto, and Glavaš 2019). We refer the reader to the work of Camacho-Collados,…”

Diversicon: Pluggable Lexical Domain Knowledge

“…More recent research, however, highlighted inherent limitations of fully unsupervised approaches in terms of their ability actually to understand meaning in text [20]. Consequently, the computational linguistics and NLP communities are starting to turn towards hybrid approaches that use structured and unstructured lexical knowledge in com-bination [32]. Inversely, the Semantic Web and knowledge graph communities are increasingly incorporating corpusbased approaches into solving traditionally knowledgedriven problems such as entity linking and data integration [34].…”