Run clean_text first

Author: caduribas

training/app.py

@@ -1,13 +1,31 @@
 import joblib
 import streamlit as st
+import re
+
+
+def clean_text(text: str) -> str:
+    """
+    Change abstract by removing HTML tags, URLs, content inside brackets,
+    and extra whitespace, and converts it to lowercase.
+
+    :param text: abstract text
+    :return: cleaned text
+    """
+    text = text.lower()
+    text = re.sub(r'<[^>]*>', " ", text)
+    text = re.sub(r"\[.*?\]", "", text)
+    text = re.sub(r"https?://\S+|www\.\S+", "", text)
+    text = re.sub(r"\s+", " ", text)
+    return text.strip()
 
 
 model = joblib.load("training/svc_pipeline.pkl")
 st.title("Text Classification - RNA Related or Not")
 user_input = st.text_area("Enter text for classification:", "")
 
 if st.button("Classify"):
-    if user_input.strip():
+    if user_input:
+        user_input = clean_text(user_input)
         prediction = model.predict([user_input])[0]
         if prediction:
             st.success(f"Prediction: RNA-related")