|
import gradio as gr |
|
from gradio_rich_textbox import RichTextbox |
|
|
|
from helper.text_preprocess import space_punc |
|
from helper.alignment_mappers import select_model |
|
from helper.pos_taggers import select_pos_tagger |
|
from helper.translators import select_translator |
|
|
|
|
|
def bn_postagger(src, translator, model_name, tagger): |
|
""" |
|
Bangla PoS Tagger |
|
""" |
|
|
|
src = space_punc(src) |
|
|
|
tgt_base, tgt = select_translator(src, translator) |
|
|
|
model_name = select_model(model_name) |
|
|
|
result, pos_accuracy = select_pos_tagger(src, tgt, model_name, tagger) |
|
|
|
return tgt_base, result, pos_accuracy |
|
|
|
|
|
with gr.Blocks(css="styles.css") as demo: |
|
gr.HTML("<h1>Bangla PoS Taggers</h1>") |
|
gr.HTML("<p>Parts of Speech (PoS) Tagging of Bangla Sentence using Bangla-English <strong>Word Alignment</strong></p>") |
|
|
|
with gr.Row(): |
|
with gr.Column(): |
|
inputs = [ |
|
gr.Textbox( |
|
label="Enter Bangla Sentence", |
|
placeholder="বাংলা বাক্য লিখুন" |
|
), |
|
gr.Dropdown( |
|
choices=["Google", "BanglaNMT", "MyMemory"], |
|
label="Select a Translator" |
|
), |
|
gr.Dropdown( |
|
choices=["Google-mBERT (Base-Multilingual)", "Neulab-AwesomeAlign (Bn-En-0.5M)", "BUET-BanglaBERT (Large)", "SagorSarker-BanglaBERT (Base)", "SentenceTransformers-LaBSE (Multilingual)"], |
|
label="Select a Model" |
|
), |
|
gr.Dropdown( |
|
choices=["spaCy", "NLTK", "Flair", "TextBlob"], |
|
label="Select a PoS Tagger" |
|
) |
|
] |
|
|
|
btn = gr.Button(value="Submit", elem_classes="mybtn") |
|
gr.ClearButton(inputs) |
|
|
|
with gr.Column(): |
|
outputs = [ |
|
gr.Textbox(label="English Translation"), |
|
RichTextbox(label="PoS Tags"), |
|
gr.Textbox(label="PoS Tagging Accuracy (Based on Unknown(UNK) Tags)") |
|
] |
|
|
|
btn.click(bn_postagger, inputs, outputs) |
|
|
|
gr.Examples([ |
|
[ |
|
"বাংলাদেশ দক্ষিণ এশিয়ার একটি সার্বভৌম রাষ্ট্র।", |
|
"Google", |
|
"Neulab-AwesomeAlign (Bn-En-0.5M)", |
|
"NLTK" |
|
], |
|
[ |
|
"বাংলাদেশের সংবিধানিক নাম কি?", |
|
"Google", |
|
"Google-mBERT (Base-Multilingual)", |
|
"spaCy" |
|
], |
|
[ |
|
"বাংলাদেশের সাংবিধানিক নাম গণপ্রজাতন্ত্রী বাংলাদেশ।", |
|
"Google", |
|
"Google-mBERT (Base-Multilingual)", |
|
"TextBlob" |
|
], |
|
[ |
|
"তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।", |
|
"Google", |
|
"Neulab-AwesomeAlign (Bn-En-0.5M)", |
|
"spaCy" |
|
], |
|
[ |
|
"তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।", |
|
"BanglaNMT", |
|
"Google-mBERT (Base-Multilingual)", |
|
"spaCy" |
|
], |
|
[ |
|
"তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।", |
|
"MyMemory", |
|
"Google-mBERT (Base-Multilingual)", |
|
"spaCy" |
|
], |
|
[ |
|
"বিশ্বের আরও একটি সেরা ক্লাব।", |
|
"Google", |
|
"Neulab-AwesomeAlign (Bn-En-0.5M)", |
|
"Flair" |
|
] |
|
|
|
], inputs) |
|
|
|
|
|
|
|
|
|
if __name__ == "__main__": |
|
demo.launch() |
