arnocaj

They explicitely mention Question Answering. Could it be that they use something like BERT trained with Squad dataset, and fine tuned on additional content? If so, Bert is very intense in terms of required GPU hardware...

Who would be the right people to initiate a paradigm shift? Celebrities? Influencers?

I think the governments have to enforce an open standard which would allow intercompatibility between different social networks.

I would not consider word embeddings to be state of the art anymore. Word Embeddings are like TF-IDF when word embeddings came out. Have a look at BERT model that just recently got published and is outperforming all kind if NLP tasks with one main Architecture. I would consider BERT language model two levels higher than word embeddings, as it considers full context sensitive embeddings, dependent on the text left and right of the word in parallel.

The hairball problem sometimes has solutions. this is a filtering soltuion which works with all general layout methods: http://visone.info/wiki/index.php/Backbone_Layout

It also scales for large graphs