Datasets:
Matthew Franglen
commited on
Commit
•
22c9e6d
1
Parent(s):
cdea156
Add make target to generate all the laptop 2014 data
Browse files- Makefile +42 -0
- src/alignment.py +1 -1
- src/main.py +2 -0
Makefile
CHANGED
@@ -14,6 +14,48 @@ hub/login : .make/requirements
|
|
14 |
hub/create : .make/requirements
|
15 |
poetry run huggingface-cli repo create aste-v2 --type dataset
|
16 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
## Install Python Dependencies
|
19 |
requirements : .make/requirements
|
|
|
14 |
hub/create : .make/requirements
|
15 |
poetry run huggingface-cli repo create aste-v2 --type dataset
|
16 |
|
17 |
+
data : \
|
18 |
+
data/2014/laptop/aste/test.gz.parquet \
|
19 |
+
data/2014/laptop/aste/valid.gz.parquet \
|
20 |
+
data/2014/laptop/sem-eval/train.gz.parquet \
|
21 |
+
data/2014/laptop/aste/train.gz.parquet \
|
22 |
+
data/2014/laptop/sem-eval/test.gz.parquet \
|
23 |
+
data/2014/laptop/sem-eval/valid.gz.parquet
|
24 |
+
|
25 |
+
|
26 |
+
data/2014/laptop/aste/train.gz.parquet : .make/requirements
|
27 |
+
poetry run python -m src.main aste \
|
28 |
+
--aste-file data/raw/aspect-sentiment-triplet-extraction/laptop-2014/train_triplets.txt \
|
29 |
+
--output-file $@
|
30 |
+
|
31 |
+
data/2014/laptop/aste/valid.gz.parquet : .make/requirements
|
32 |
+
poetry run python -m src.main aste \
|
33 |
+
--aste-file data/raw/aspect-sentiment-triplet-extraction/laptop-2014/dev_triplets.txt \
|
34 |
+
--output-file $@
|
35 |
+
|
36 |
+
data/2014/laptop/aste/test.gz.parquet : .make/requirements
|
37 |
+
poetry run python -m src.main aste \
|
38 |
+
--aste-file data/raw/aspect-sentiment-triplet-extraction/laptop-2014/test_triplets.txt \
|
39 |
+
--output-file $@
|
40 |
+
|
41 |
+
data/2014/laptop/sem-eval/train.gz.parquet : .make/requirements
|
42 |
+
poetry run python -m src.main sem-eval \
|
43 |
+
--aste-file data/raw/aspect-sentiment-triplet-extraction/laptop-2014/train_triplets.txt \
|
44 |
+
--sem-eval-file data/raw/sem-eval/2014/Laptop_Train_v2.xml \
|
45 |
+
--output-file $@
|
46 |
+
|
47 |
+
data/2014/laptop/sem-eval/valid.gz.parquet : .make/requirements
|
48 |
+
poetry run python -m src.main sem-eval \
|
49 |
+
--aste-file data/raw/aspect-sentiment-triplet-extraction/laptop-2014/dev_triplets.txt \
|
50 |
+
--sem-eval-file data/raw/sem-eval/2014/Laptop_Train_v2.xml \
|
51 |
+
--output-file $@
|
52 |
+
|
53 |
+
data/2014/laptop/sem-eval/test.gz.parquet : .make/requirements
|
54 |
+
poetry run python -m src.main sem-eval \
|
55 |
+
--aste-file data/raw/aspect-sentiment-triplet-extraction/laptop-2014/test_triplets.txt \
|
56 |
+
--sem-eval-file data/raw/sem-eval/2014/Laptops_Test_Data_PhaseA.xml \
|
57 |
+
--output-file $@
|
58 |
+
|
59 |
|
60 |
## Install Python Dependencies
|
61 |
requirements : .make/requirements
|
src/alignment.py
CHANGED
@@ -18,7 +18,7 @@ def find_closest_text(
|
|
18 |
result = original_text.map(no_space_replacements)
|
19 |
non_perfect_matches = result.isna().sum()
|
20 |
|
21 |
-
assert non_perfect_matches / len(original) <= 0.
|
22 |
"Poor alignment with replacement text. "
|
23 |
f"{non_perfect_matches:,} of {len(original),} rows did not match well"
|
24 |
)
|
|
|
18 |
result = original_text.map(no_space_replacements)
|
19 |
non_perfect_matches = result.isna().sum()
|
20 |
|
21 |
+
assert non_perfect_matches / len(original) <= 0.20, (
|
22 |
"Poor alignment with replacement text. "
|
23 |
f"{non_perfect_matches:,} of {len(original),} rows did not match well"
|
24 |
)
|
src/main.py
CHANGED
@@ -32,6 +32,7 @@ def aste(
|
|
32 |
|
33 |
print(df.sample(3))
|
34 |
|
|
|
35 |
df.to_parquet(output_file, compression="gzip")
|
36 |
|
37 |
|
@@ -58,6 +59,7 @@ def sem_eval(
|
|
58 |
|
59 |
print(df.sample(3))
|
60 |
|
|
|
61 |
df.to_parquet(output_file, compression="gzip")
|
62 |
|
63 |
|
|
|
32 |
|
33 |
print(df.sample(3))
|
34 |
|
35 |
+
output_file.parent.mkdir(exist_ok=True, parents=True)
|
36 |
df.to_parquet(output_file, compression="gzip")
|
37 |
|
38 |
|
|
|
59 |
|
60 |
print(df.sample(3))
|
61 |
|
62 |
+
output_file.parent.mkdir(exist_ok=True, parents=True)
|
63 |
df.to_parquet(output_file, compression="gzip")
|
64 |
|
65 |
|