Spaces:
Sleeping
Sleeping
Commit
·
de79df8
1
Parent(s):
d68b3f4
Update app.py
Browse files
app.py
CHANGED
|
@@ -7,6 +7,20 @@ def isNoneWords(word):
|
|
| 7 |
return True
|
| 8 |
else:
|
| 9 |
return False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
def top_similarity_route(word):
|
| 12 |
if isNoneWords(word):
|
|
@@ -18,10 +32,90 @@ def top_similarity_route(word):
|
|
| 18 |
sim_res += f'{item[0]}: {round(item[1], 4)}\n'
|
| 19 |
return sim_res
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
-
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
title = 'Calculate word similarity based on Tencent AI Lab Embedding'
|
| 26 |
-
|
| 27 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
return True
|
| 8 |
else:
|
| 9 |
return False
|
| 10 |
+
|
| 11 |
+
def word_analogy(word1, word2, word3):
|
| 12 |
+
analogy_words = model.similar_by_vector(model.word_vec(word1) - model.word_vec(word2) + model.word_vec(word3))
|
| 13 |
+
sim_res = ""
|
| 14 |
+
for item in analogy_words:
|
| 15 |
+
sim_res += f'{item[0]}: {round(item[1], 4)}\n'
|
| 16 |
+
return sim_res
|
| 17 |
+
|
| 18 |
+
def similarity_route(word1, word2):
|
| 19 |
+
if isNoneWords(word1) or isNoneWords(word2):
|
| 20 |
+
return "word is null or not in model!"
|
| 21 |
+
else:
|
| 22 |
+
return float(model.similarity(word1, word2))
|
| 23 |
+
|
| 24 |
|
| 25 |
def top_similarity_route(word):
|
| 26 |
if isNoneWords(word):
|
|
|
|
| 32 |
sim_res += f'{item[0]}: {round(item[1], 4)}\n'
|
| 33 |
return sim_res
|
| 34 |
|
| 35 |
+
def top_similar_words_layout():
|
| 36 |
+
with gr.Column():
|
| 37 |
+
with gr.Row():
|
| 38 |
+
with gr.Column():
|
| 39 |
+
word = gr.Textbox(lines=1, label='Input word', placeholder='Input word here')
|
| 40 |
+
with gr.Row():
|
| 41 |
+
clear = gr.ClearButton()
|
| 42 |
+
submit = gr.Button("Submit")
|
| 43 |
+
output = gr.Textbox(lines=20, label='Similar words', placeholder='Output here')
|
| 44 |
|
| 45 |
+
submit.click(fn=top_similarity_route, inputs=[word], outputs=[output])
|
| 46 |
+
|
| 47 |
+
examples=[['兔子', '松鼠']]
|
| 48 |
+
ex = gr.Examples(
|
| 49 |
+
examples=examples,
|
| 50 |
+
fn=top_similarity_route,
|
| 51 |
+
inputs=[word],
|
| 52 |
+
outputs=[output],
|
| 53 |
+
cache_examples=False,
|
| 54 |
+
run_on_click=False
|
| 55 |
+
)
|
| 56 |
+
|
| 57 |
|
| 58 |
+
def similarity_layout():
|
| 59 |
+
with gr.Column():
|
| 60 |
+
with gr.Row():
|
| 61 |
+
with gr.Column():
|
| 62 |
+
with gr.Row():
|
| 63 |
+
word1 = gr.Textbox(lines=1, label='Input word1', placeholder='Input word1 here')
|
| 64 |
+
word2 = gr.Textbox(lines=1, label='Input word2', placeholder='Input word2 here')
|
| 65 |
+
with gr.Row():
|
| 66 |
+
clear = gr.ClearButton()
|
| 67 |
+
submit = gr.Button("Submit")
|
| 68 |
+
output = gr.Textbox(lines=1, label='Similar words', placeholder='Output here')
|
| 69 |
+
|
| 70 |
+
submit.click(fn=similarity_route, inputs=[word1, word2], outputs=[output])
|
| 71 |
+
|
| 72 |
+
examples=[['淘宝', '京东', 0.7887385]]
|
| 73 |
+
ex = gr.Examples(
|
| 74 |
+
examples=examples,
|
| 75 |
+
fn=similarity_route,
|
| 76 |
+
inputs=[word1, word2],
|
| 77 |
+
outputs=[output],
|
| 78 |
+
cache_examples=False,
|
| 79 |
+
run_on_click=False
|
| 80 |
+
)
|
| 81 |
+
|
| 82 |
+
def word_analogy_layout():
|
| 83 |
+
with gr.Column():
|
| 84 |
+
with gr.Row():
|
| 85 |
+
with gr.Column():
|
| 86 |
+
with gr.Row():
|
| 87 |
+
word1 = gr.Textbox(lines=1, label='Input word1', placeholder='Input word1 here')
|
| 88 |
+
word2 = gr.Textbox(lines=1, label='Input word2', placeholder='Input word2 here')
|
| 89 |
+
word3 = gr.Textbox(lines=1, label='Input word3', placeholder='Input word3 here')
|
| 90 |
+
with gr.Row():
|
| 91 |
+
clear = gr.ClearButton()
|
| 92 |
+
submit = gr.Button("Submit")
|
| 93 |
+
output = gr.Textbox(lines=1, label='Analogy words', placeholder='Output here')
|
| 94 |
+
|
| 95 |
+
submit.click(fn=word_analogy, inputs=[word1, word2, word3], outputs=[output])
|
| 96 |
+
|
| 97 |
+
examples=[['国王', '男人', '女人', '王后']]
|
| 98 |
+
ex = gr.Examples(
|
| 99 |
+
examples=examples,
|
| 100 |
+
fn=word_analogy,
|
| 101 |
+
inputs=[word1, word2, word3],
|
| 102 |
+
outputs=[output],
|
| 103 |
+
cache_examples=False,
|
| 104 |
+
run_on_click=False
|
| 105 |
+
)
|
| 106 |
+
|
| 107 |
+
if __name__ == '__main__':
|
| 108 |
+
model = KeyedVectors.load_word2vec_format('../word_sim_demo/tencent-ailab-embedding-zh-d100-v0.2.0-s/tencent-ailab-embedding-zh-d100-v0.2.0-s.txt', binary=False)
|
| 109 |
title = 'Calculate word similarity based on Tencent AI Lab Embedding'
|
| 110 |
+
|
| 111 |
+
with gr.Blocks() as demo:
|
| 112 |
+
gr.HTML(title)
|
| 113 |
+
with gr.Column(elem_id="col-container"):
|
| 114 |
+
with gr.Tab("Top similar words"):
|
| 115 |
+
top_similar_words_layout()
|
| 116 |
+
with gr.Tab("Similarity of words"):
|
| 117 |
+
similarity_layout()
|
| 118 |
+
with gr.Tab("Word analogy"):
|
| 119 |
+
word_analogy_layout()
|
| 120 |
+
|
| 121 |
+
demo.queue(max_size=64).launch()
|