Create app.py

app.py

@@ -0,0 +1,56 @@
+import gradio as gr
+import gradio.inputs as grinputs
+import gradio.outputs as groutputs
+
+from gensim.models import KeyedVectors
+from gensim.parsing import preprocessing
+
+filters = [
+    preprocessing.remove_stopwords,
+    preprocessing.strip_tags,
+    preprocessing.strip_punctuation,
+    preprocessing.strip_numeric,
+    preprocessing.strip_multiple_whitespaces,
+    preprocessing.strip_non_alphanum,
+    preprocessing.strip_short,
+    preprocessing.remove_stopwords,
+    preprocessing.lower_to_unicode,
+]
+
+def parse_text(text):
+    return text.replace(" ", "").replace(";", ",").split(',')
+
+def clean_words(words):
+    clean_dict = {}
+    for (word, score) in words:
+        prep_word = preprocessing.preprocess_string(word, filters=filters)
+        if len(prep_word) > 0:
+            word = prep_word[0]
+            is_clean = sum( [word.startswith(clean_word) or word.endswith(clean_word) for clean_word in clean_dict.keys()] ) == 0
+            if is_clean:
+                clean_dict[word] = round(score, 2)
+    return clean_dict
+
+path = "cc.fr.300.vec"
+m = KeyedVectors.load_word2vec_format(path)
+
+def on_submit(text, mode):
+    print('{} mode'.format(mode))
+    positive = parse_text(text)
+    if mode=='Close':
+        words = m.most_similar(positive=positive, topn=200)
+    else:
+        words = m.most_similar(positive=positive, topn=10000)[::-1]
+    return str(clean_words(words))[1:-1]
+
+iface = gr.Interface(
+    fn=on_submit, 
+    inputs=[
+        grinputs.Textbox(placeholder='mot1, mot2, mot3, ...', label="Input words (coma separated)"), 
+        grinputs.Radio(['Close', 'Far'], label="Close of Far mode")], 
+    outputs=[
+        groutputs.Textbox(label='Information')],
+    allow_screenshot=False, 
+    allow_flagging=False,
+    )
+iface.launch()