Data collection

[maherr13]

Thanks for the great work,
I have a question about data collection, you mentioned that you collected data regarding the temporal model using NYT sitemap scrapping,
How did you manage to identify the event and its year though?

[guyrosin]

Hi @maherr13,
Thanks for your interest. You can find details about the data collection in section 5.2 in our paper (preprint: https://arxiv.org/pdf/2012.12065.pdf).
Basically, event data was extracted from DBpedia and we used embeddings based on Wikipedia to represent events. As described in section 3.1, event names often aren't directly mentioned in the news and that's why we had to use the separate event embeddings from Wikipedia and apply the proposed projection method.