app.py

import subprocess
import sys

def install(package):
    subprocess.check_call([sys.executable, "-m", "pip", "install", package], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)

install("keybert")
install("keyphrase_vectorizers")
import warnings
from keybert import KeyBERT
from keyphrase_vectorizers import KeyphraseCountVectorizer
import gradio as gr

with warnings.catch_warnings():
    warnings.simplefilter("ignore", category=UserWarning)
    embedding = 'all-mpnet-base-v2'
    key_model = KeyBERT(model=embedding)
    vectorizer_params = KeyphraseCountVectorizer(spacy_pipeline='en_core_web_sm', pos_pattern='<J.*>*<N.*>+', stop_words='english', lowercase=True)

def get_keywords(course_name, course_desc):
    keywords_list = []
    course_name, course_desc = course_name.strip().lower(), course_desc.strip().lower()
    data = course_name+". "+course_desc
    keywords = key_model.extract_keywords(data, vectorizer=vectorizer_params, stop_words='english', top_n=7, use_mmr=True)
    keywords_list = list(dict(keywords).keys())
    return ", ".join(keywords_list)
    
iface = gr.Interface(fn=get_keywords, inputs=[gr.Textbox(label="Course Name"), gr.Textbox(label="Course Description")], outputs=gr.Textbox(label="Relevant Tags"),
                     title="College Course Tags Generator", description="Generating tags/keywords based on Keyphrase-BERT Extraction'")
iface.launch()