Yes. ConceptNet [1] and distributional word embeddings go really well together, and can compare word meanings better than either one alone. Here's the preprint

Yes. ConceptNet [1] and distributional word embeddings go really well together, and can compare word meanings better than either one alone. Here's the preprint of the AAAI 2017 paper [2]. [1] http://www.conceptnet.io/ http://www.conceptnet.io/ [2] https://arxiv.org/pdf/1612.03975v1.pdf https://arxiv.org/pdf/1612.03975v1.pdf