def save_tsv(filename, essay_code, sentences, dist, label_preds):
header = ["essay code", "unit id", "text", "target", "relation", "drop_flag"]
rep = TreeBuilder(dist)
component_labels = rep.auto_component_labels(AC_breakdown=True)
f = open(filename, "w")
f.write("\t".join(header)+"\n")
label_idx = 0
for i in range(len(sentences)):
output_line = []
output_line.append(essay_code)
output_line.append(str(i+1))
output_line.append(sentences[i])
if component_labels[i] == "non-arg comp.":
output_line.append("")
output_line.append("")
output_line.append("TRUE")
else:
target = i+1+dist[i]
if target==i+1: # point to itself, i.e., root
output_line.append("")
output_line.append("")
else: # not root
output_line.append(str(target))
output_line.append(label_preds[label_idx])
label_idx += 1
output_line.append("FALSE")
f.write("\t".join(output_line)+"\n")
assert(label_idx == len(label_preds))
f.close()
def structured_output_quality(links) -> (List, float, float, float):
"""
Infer component labels automatically from the structure
"""
component_labels = []
tree_ratio = 0
avg_depth = 0
avg_leaf_prop = 0
all_depths = []
n_essays = len(links)
for i in range(len(links)):
rep = TreeBuilder(links[i])
component_labels.append(rep.auto_component_labels(AC_breakdown=True))
if rep.is_tree():
tree_ratio += 1
# evaluate this only when the output forms a tree
depth, leaf_prop = rep.tree_depth_and_leaf_proportion()
avg_depth += depth
all_depths.append(depth)
avg_leaf_prop += leaf_prop
return component_labels, float(tree_ratio) / float(n_essays), float(
avg_depth) / float(tree_ratio), float(avg_leaf_prop) / float(
tree_ratio), all_depths
def create_pairwise_data(sentences, dist):
"""
Create pairwise link labelling data
Args:
sentences (list[str])
dist (list[int])
Returns:
list[tuple(str,str)]
"""
rep = TreeBuilder(dist)
component_labels = rep.auto_component_labels(AC_breakdown=True)
output = []
for i in range(len(sentences)):
if component_labels[i] == "non-arg comp.":
pass
else:
if i+dist[i] != i: # the current sentence does not point to itself, i.e., not a root
source = sentences[i]
target = sentences[i+dist[i]]
output.append((source, target))
return output
# assertion to check whether we have included non-arg-units here
assert (len(sentences) == len(rel_distances))
assert (len(sentences) == len(rel_labels))
# determine where to save the file
if args.split:
split_folder = check_train_or_test(split_info, essay.essay_code)
assert (split_folder != None)
split_folder = split_folder + "/"
else:
split_folder = "" # no split information provided
# component labels
rep = TreeBuilder(rel_distances)
component_labels = rep.auto_component_labels(AC_breakdown=True)
# save to file
save_content_to_file(
args.out_dir + "linking/" + split_folder.lower() +
essay.essay_code + ".sentences", sentences)
save_content_to_file(
args.out_dir + "linking/" + split_folder.lower() +
essay.essay_code + ".vectors", vectors.tolist())
save_content_to_file(
args.out_dir + "linking/" + split_folder.lower() +
essay.essay_code + ".rel_distances", rel_distances)
# save_content_to_file(args.out_dir + "linking/" + split_folder.lower() + essay.essay_code + ".rel_labels", rel_labels)
save_content_to_file(
args.out_dir + "linking/" + split_folder.lower() +
essay.essay_code + ".component_labels", component_labels)