eighth commit

app.py

@@ -1,29 +1,97 @@
-import gradio as gr
 from transformers import pipeline
+import gradio as gr
+
+# ==============================
+# 1️⃣ Load semua model
+# ==============================
+
+# Sentiment Analysis (IndoBERT)
+sentiment_model = pipeline(
+    "sentiment-analysis",
+    model="w11wo/indonesian-roberta-base-sentiment-classifier"
+)
+
+# Named Entity Recognition (NER)
+ner_model = pipeline(
+    "ner",
+    model="cahya/bert-base-indonesian-NER",
+    aggregation_strategy="simple"
+)
 
-# Load model IndoBERT + NER + Topic
-sentiment_model = pipeline("sentiment-analysis", model="indobenchmark/indobert-base-p1")
-ner_model = pipeline("ner", model="cahya/bert-base-indonesian-NER", aggregation_strategy="simple")
-topic_model = pipeline("text-classification", model="w11wo/indonesian-roberta-topic-classification")
+# Zero-shot Topic Classification (multilingual)
+topic_model = pipeline(
+    "zero-shot-classification",
+    model="joeddav/xlm-roberta-large-xnli"
+)
 
-def analyze(text):
+# Daftar label topik yang ingin dikenali
+TOPIC_LABELS = [
+    "politik",
+    "ekonomi",
+    "olahraga",
+    "teknologi",
+    "pendidikan",
+    "kesehatan",
+    "hiburan",
+    "sosial"
+]
+
+# ==============================
+# 2️⃣ Fungsi analisis teks
+# ==============================
+
+def analyze_text(text):
+    if not text or not text.strip():
+        return {
+            "error": "Teks kosong. Silakan masukkan kalimat Bahasa Indonesia."
+        }
+
+    # Analisis sentimen
     sentiment = sentiment_model(text)[0]
-    ner = ner_model(text)
-    topic = topic_model(text)[0]
 
+    # Deteksi entitas
+    entities = ner_model(text)
+    entity_result = [
+        {
+            "entity": e["entity_group"],
+            "word": e["word"],
+            "score": round(e["score"], 4)
+        }
+        for e in entities
+    ]
+
+    # Klasifikasi topik
+    topic = topic_model(text, TOPIC_LABELS)
+    topic_result = {
+        "label": topic["labels"][0],
+        "score": round(topic["scores"][0], 4)
+    }
+
+    # Gabungkan hasil
     return {
-        "sentiment": sentiment,
-        "entities": [{"entity": e["entity_group"], "word": e["word"]} for e in ner],
-        "topic": topic,
+        "sentiment": {
+            "label": sentiment["label"],
+            "score": round(sentiment["score"], 4)
+        },
+        "entities": entity_result,
+        "topic": topic_result
     }
 
+# ==============================
+# 3️⃣ UI Gradio
+# ==============================
+
 demo = gr.Interface(
-    fn=analyze,
-    inputs=gr.Textbox(label="Masukkan teks bahasa Indonesia", lines=3),
-    outputs="json",
-    title="Analisis Teks Bahasa Indonesia",
-    description="Deteksi Sentimen, Entitas, dan Topik menggunakan model IndoBERT & RoBERTa."
+    fn=analyze_text,
+    inputs=gr.Textbox(
+        lines=3,
+        placeholder="Masukkan kalimat Bahasa Indonesia..."
+    ),
+    outputs=gr.JSON(label="Hasil Analisis"),
+    title="Analisis Sentimen, Entitas, & Topik Bahasa Indonesia",
+    description="Gunakan AI berbasis IndoBERT & XLM-R untuk analisis sentimen, pengenalan entitas, dan deteksi topik otomatis."
 )
 
+# Jalankan app
 if __name__ == "__main__":
     demo.launch()