Knowledgebase

• Word2Vec and FastText Word Embedding with Gensim
• Gensim Tut
• uMap - Visualization of high-dimension-vectors
• fast approximate nearest neighbor search
  • Hnswlib
  • SPTAG

Interesting approaches worth to investigate further

• Word Embeddings for Entity annotated texts
• Vector Embedding of Wikipedia Concepts and Entities
• Probing for Semantic Classes: Diagnosing the Meaning Content of Word Embeddings (blog post, paper)

Survey Paper

• A Survey of Word Embeddings Evaluation Methods

Tools

Entity Linking

• OpenTapioca (Demo, Source Code, Paper)
• Wikilinks NEL
• DBpedia spotlight
• ambiverse-nlu

Benchmarking Paper

• Evaluation methods for unsupervised word embeddings
• WordRep: A Benchmark for Research on Learning Word Representations
• Word Embeddings Evaluation and Combination
• Automatic generation of tunable analogy benchmarks for word representations

BigGraph Facebook

Trained Embeddings on Knowledge-Graph from WikiData as Knowledgebase

PyTorch-BigGraph (PBG) is a distributed system for learning graph embeddings for large graphs, particularly big web interaction graphs with up to billions of entities and trillions of edges. PBG was introduced in the PyTorch-BigGraph: A Large-scale Graph Embedding Framework paper, presented at the SysML conference in 2019.

• PyTorch-BigGraph: A Large-Scale Graph Embedding System
• Source Code
• Training Data Source
• Pre-Trained Embedding (36GiB, gzip-compressed)