Hello Everyone!
I’m looking for a way that I can generate a few paragraphs of plausible text that bridges a give start sentence(s) and end sentence(s).
Is this something that can be done with the deeppavlov.ai and maybe BERT do you think?
If it is would you be able to give any pointers as I’m a newbie at this stuff! 
Cheers
Fred
Hi Fred!
I’m not aware of the models to solve the “bridging” task you mention. The closest analog is TextGraphs-13 Shared Task on Multi-Hop Inference Explanation Regeneration where a question and an answer should be connected by explanatory statements (from predefined pool).
In DeepPavlov we tried straightforward approach to extractive summarization by feeding a title of the wikipedia article to the BERT, and ranking all the sentences from this article to predict the next sentence in summary. But here we have only “starting” sentence without any “end” sentence.
Hi and thanks so much for your reply!
I will take a look at the system you mention and see if I can make it do what I need.
I actually also just found this paper+code that seems to actually be able to do what I need I will try that out too. https://github.com/umanlp/tg2019task
Mine seems to be a task that is not very common to need to solve? I suppose I’m more interested in ‘story generation’ and maybe that’s not yet commercially viable and therefore not much effort goes towards that kind of thing.
Cheers, Fred