ConceptNet at SemEval-2017 Task 2: Extending Word Embeddings with Multilingual Relational Knowledge

Abstract

This paper describes Luminoso's participation in SemEval 2017 Task 2, "Multilingual and Cross-lingual Semantic Word Similarity", with a system based on ConceptNet. ConceptNet is an open, multilingual knowledge graph that focuses on general knowledge that relates the meanings of words and phrases. Our submission to SemEval was an update of previous work that builds high-quality, multilingual word embeddings from a combination of ConceptNet and distributional semantics. Our system took first place in both subtasks. It ranked first in 4 out of 5 of the separate languages, and also ranked first in all 10 of the cross-lingual language pairs.

TechNet: Technology semantic network based on patent data

“…The ontology-based databases retrieve the unstructured data and relate them using various techniques. However, none of the publicly available ontology-based databases, such as the popular WordNet (Fellbaum, 2012;Miller et al, 1990), ConceptNet (Speer et al, 2016;Speer & Havasi, 2012;Speer & Lowry-Duda, 2017) or BabelNet (Navigli & Ponzetto, 2012), derive the relations between entities from an engineering design viewpoint.…”

“…These terms closely related to "wireless charger" represent technical concepts regarding functions, components, configurations or working mechanisms. By contrast, neither WordNet (Fellbaum, 2012;Miller et al, 1990), ConceptNet (Speer et al, 2016;Speer & Havasi, 2012;Speer & Lowry-Duda, 2017) nor the semantic network of Shi et al (2017) contain the "wireless charger" term. In particular, we checked Google Knowledge Graph's term recommendations for "wireless charger", and the results are more related to consumer brands and products that have wireless charging capabilities ( Table 6).…”

“…A large semantic network is normally composed of a library of semantic entities (e.g., words or phrases) and their semantic relations, which are often statistically or linguistically "learned" based on collaboratively edited and accumulated knowledge databases, such as Wikipedia. Over the past decade, semantic networks have been enabled by the development of large-scale knowledge basis and ontology databases, such as WordNet (Fellbaum, 2012;Miller et al, 1990), ConceptNet (Speer et al, 2016;Speer & Havasi, 2012;Speer & Lowry-Duda, 2017), never-ending language learning (NeLL) (Mitchell et al, 2015), Freebase (Bollacker et al, 2008(Bollacker et al, , 2007 and Yago (Rebele et al, 2016), for various general applications in text data mining, natural language processing (NLP), knowledge discovery, information retrieval and artificial intelligence. Likewise, the proprietary Google Knowledge Graph 1 provides the backbone behind Google's semantic search and answer features for web searches, Gmail, and Google Assistant, for example.…”

Incorporating Structured Commonsense Knowledge in Story Completion

“…We leverage the implicit knowledge by using a numberbatch word embedding (Speer, Chin, and Havasi 2017), which is trained on data from ConceptNet, word2vec, GloVe, and OpenSubtitles. The numberbatch achieves good performance on tasks related to commonsense knowledge (Speer and Lowry-Duda 2017). For instance, the cosine similarity between "diet" and "overweight" in numberbatch is 0.453, but it is 0.326 in GloVe.…”

ELiRF-UPV at SemEval-2018 Task 11: Machine Comprehension using Commonsense Knowledge

“…It is designed to represent the general knowledge involved in understanding language. ConceptNet could be used in combination with sources of distributional semantics, particularly the word2vec Google News skip-gram embeddings (Mikolov et al, 2013)) and GloVe 1.2 (Pennington et al, 2014), to produce new embeddings, NumberBatch embeddings, with state-of-the-art performance across many wordrelatedness evaluations (Speer and Lowry-Duda, 2017).…”