Update src/webApp.py

src/webApp.py

@@ -433,24 +433,26 @@ if st.button("Analyse Review"):
             f"The model is **{s_pct}%** confident that this review sentiment is "
             f"**likely {s_word}** as a result of the following words:"
         )
-
-        # Word importance 
+        
+        # Word importance
         with st.spinner("Computing word importance..."):
             word_importance = get_word_importance(review, tokenizer, bert_model, s_label)
-            if word_importance:
-                word_fig = word_importance_chart(word_importance, s_label)
-                if word_fig:
-                    st.plotly_chart(word_fig, use_container_width=True)
-                    
-                    st.markdown("#### What the words mean for this prediction")
-                    pos_words = [(w, v) for w, v in word_importance if v > 0][:5]
-                    neg_words = [(w, v) for w, v in word_importance if v < 0][:5]
+
+        if word_importance:
+            word_fig  = word_importance_chart(word_importance, s_label)
+            pos_words = [(w, v) for w, v in word_importance if v > 0][:5]
+            neg_words = [(w, v) for w, v in word_importance if v < 0][:5]
+
+            if word_fig:
+                st.plotly_chart(word_fig, use_container_width=True)
+
+            st.markdown("#### What the words mean for this prediction")
 
             if s_word == "Positive":
                 if pos_words:
-                    word_list = ", ".join([f"{w}" for w, _ in pos_words])
+                    word_list = ", ".join([f"**{w}**" for w, _ in pos_words])
                     st.markdown(
-                        f"Words that most likely pushed the model toward **Positive**: {word_list}. "
+                        f"Words that most pushed the model toward **Positive**: {word_list}. "
                         "These words carry positive connotations that the model strongly associates "
                         "with genuine, satisfied reviews."
                     )
@@ -464,7 +466,7 @@ if st.button("Analyse Review"):
                 if neg_words:
                     word_list = ", ".join([f"**{w}**" for w, _ in neg_words])
                     st.markdown(
-                        f"Words that most likely pushed the model toward **Negative**: {word_list}. "
+                        f"Words that most pushed the model toward **Negative**: {word_list}. "
                         "These words carry negative connotations that the model strongly associates "
                         "with dissatisfied or critical reviews."
                     )
@@ -476,6 +478,7 @@ if st.button("Analyse Review"):
                     )
 
 
+
         # suspicious review detection
         st.subheader("Suspicious Review Detection")
 
@@ -499,10 +502,12 @@ if st.button("Analyse Review"):
 
         # Interpretation sentence
         st.markdown(
-            f"The model is **{s_pct}%** confident that this review sentiment is "
-            f"**likely {s_word}** as a result of the following words:"
+            f"The model is **{prob_pct}%** confident that this review is "
+            f"**{likely} suspicious** as a result of the following features:"
         )
 
+        # Detailed bullet point explanations per feature
+
         # Detailed bullet point explanations per feature
         bullets = build_suspicious_bullets(features, rating)
         for bullet in bullets: