Commit
·
6b83453
1
Parent(s):
37dce2a
Add get_alignments_table function to generate a Markdown table of Spacy PoS tags
Browse files- helper/alignment_mappers.py +60 -0
helper/alignment_mappers.py
CHANGED
|
@@ -122,4 +122,64 @@ def get_word_index_mapping(source="", target="", model_name=""):
|
|
| 122 |
return result
|
| 123 |
|
| 124 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 125 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 122 |
return result
|
| 123 |
|
| 124 |
|
| 125 |
+
def get_alignments_table(
|
| 126 |
+
source="",
|
| 127 |
+
target="",
|
| 128 |
+
model_name=""):
|
| 129 |
+
"""Get Spacy PoS Tags and return a Markdown table"""
|
| 130 |
|
| 131 |
+
sent_src, sent_tgt, align_words = get_alignment_mapping(
|
| 132 |
+
source=source, target=target, model_name=model_name
|
| 133 |
+
)
|
| 134 |
+
|
| 135 |
+
mapped_sent_src = []
|
| 136 |
+
|
| 137 |
+
html_table = '''
|
| 138 |
+
<table>
|
| 139 |
+
<thead>
|
| 140 |
+
<th>Source</th>
|
| 141 |
+
<th>Target</th>
|
| 142 |
+
</thead>
|
| 143 |
+
'''
|
| 144 |
+
|
| 145 |
+
for i, j in sorted(align_words):
|
| 146 |
+
punc = r"""!()-[]{}।;:'"\,<>./?@#$%^&*_~"""
|
| 147 |
+
if sent_src[i] in punc or sent_tgt[j] in punc:
|
| 148 |
+
mapped_sent_src.append(sent_src[i])
|
| 149 |
+
|
| 150 |
+
html_table += f'''
|
| 151 |
+
<tbody>
|
| 152 |
+
<tr>
|
| 153 |
+
<td> {sent_src[i]} </td>
|
| 154 |
+
<td> {sent_tgt[j]} </td>
|
| 155 |
+
</tr>
|
| 156 |
+
'''
|
| 157 |
+
else:
|
| 158 |
+
mapped_sent_src.append(sent_src[i])
|
| 159 |
+
|
| 160 |
+
html_table += f'''
|
| 161 |
+
<tr>
|
| 162 |
+
<td> {sent_src[i]} </td>
|
| 163 |
+
<td> {sent_tgt[j]} </td>
|
| 164 |
+
</tr>
|
| 165 |
+
'''
|
| 166 |
+
|
| 167 |
+
unks = list(set(sent_src).difference(set(mapped_sent_src)))
|
| 168 |
+
for word in unks:
|
| 169 |
+
|
| 170 |
+
html_table += f'''
|
| 171 |
+
<tr>
|
| 172 |
+
<td> {word} </td>
|
| 173 |
+
<td> N/A </td>
|
| 174 |
+
</tr>
|
| 175 |
+
'''
|
| 176 |
+
|
| 177 |
+
html_table += '''
|
| 178 |
+
</tbody>
|
| 179 |
+
</table>
|
| 180 |
+
'''
|
| 181 |
+
|
| 182 |
+
pos_accuracy = ((len(sent_src) - len(unks)) / len(sent_src))
|
| 183 |
+
pos_accuracy = f"{pos_accuracy:0.2%}"
|
| 184 |
+
|
| 185 |
+
return html_table, pos_accuracy
|