The Java Version of the C & C Parser Version 0 . 95

Acknowledgements The first Java version of the C&C parser was created by Stephen Clark, based heavily on the C++ implementation of C&C. This first attempt had a working version of the DP-based chart parser, but a buggy version of the beam-search chart parser. Darren Foong, for his 2014/15 Part II project in the Computer Laboratory [15], did an outstanding job in taking this buggy version and producing a working beam-search parser, as well as considerably cleaning up the Java code. Luana Bulat spent considerable time in producing a new markedup file covering all the lexical categories in CCGbank, one of the features of the new parser. Wenduan Xu provided invaluable assistance along the way. Thanks to Laura Rimell, a long-suffering user of C&C and collaborator of the first author, for acting as a sounding board while the Java version was being built. Thanks to Yue Zhang, whose original shift-reduce beam search parser was the precursor to the chart-based beam search parser presented here; and Ryan McDonald, whose Cambridge NLIP seminar in 2014 provided the inspiration for appying beam search to the chart. Thanks to Julia Hockenmaier for helping Luana create the new markedup file, and of course for creating CCGbank, without which none of this research would exist. Thanks to Mark Steedman, whose shoulders we have been standing on for some time, and finally James Curran, whose original design for C&C is heavily reflected in this new version, and from whom the first author learnt most of what he knows about practical computing.