The two enclosed CSV files contain the results of our measurements.
“Grammar1_Alto.csv" contains the results using the larger Grammar 1 run
on Alto (shown in Fig. 5), while “Grammar2_Alto_Bolinas.csv" contains the results using
the smaller Grammar 2 compared on Bolinas and Alto 8shown in Fig. 6).

One line corresponds to one parsed graph. The first column,
"GraphIndexInCorpus", gives the graph’s index in the "Little Prince"
AMR bank, v1.4 (see isi.amr.edu).

The columns “n”, “d” and “D_3” give the graph’s node count, maximum degree and 3-seperability respectively (the latter two are not featured in “Grammar2_Alto_Bolinas.csv”).

“Language.Size” shows the number of parses we found for each graph.

Finally, the columns starting with “Time” provide the measured runtimes of our experiments. “Time.Bolinas” denotes the Bolinas parser, “Time.TopDown” and “Time.BottomUp” denote the two modes of the Alto parser. In our evaluation, we then took the geometric means grouped by node count.

For details on Alto, see https://bitbucket.org/tclup/alto.

For details on Bolinas, see http://www.isi.edu/publications/licensed-sw/bolinas.
