Open Supermaxman opened 4 years ago
Hello Max,
Yes I could do that. But it may take a little bit time to clean as I am a little bit busy these days. Do you need it urgently?
Thanks,
I appreciate it! No rush at all, would just be nice to have for future work building on this.
Is there any chance you could release the code which produces the candidate answer chunks from the POS tags & recognized named entities? It would be very helpful for using this system in practice on new tweets, and would make additional research making use of this architecture easier.
The process is described in your paper as "We provide the annotators with a list of candidate answers, which include all noun chunks and name entities extracted by a Twitter tagging tool (Ritter et al., 2011). We also combine noun chunks if they are adjacent or separated by a preposition phrase."
It may be a very simple process, but I would like to be able to reproduce your provided candidate chunks 100%, and implementing the above steps could leave some room for different linguistical choices.