File size: 8,792 Bytes
7d1de06
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
"""
IT Ticket Classifier - HuggingFace Spaces App
Gradio interface for classifying IT support tickets
"""

import gradio as gr
import torch
import torch.nn as nn
from transformers import DistilBertModel, AutoTokenizer
from huggingface_hub import hf_hub_download
import re
import os
import numpy as np

# Configuration
HF_REPO_ID = "TuShar2309/ticket-classifier"
MODEL_FILENAME = "ticket_classifier.pt"

CLASS_NAMES = [
    "Access Management", "Backup", "Database", "Email", 
    "General Inquiry", "Hardware", "Network", "Other",
    "Printing", "Security", "Software", "Storage"
]

# Category descriptions for display
CATEGORY_INFO = {
    "Access Management": "πŸ” Login, permissions, MFA, account issues",
    "Backup": "πŸ’Ύ Backup and restore operations",
    "Database": "πŸ—„οΈ SQL, database connectivity, queries",
    "Email": "πŸ“§ Outlook, calendar, mailbox issues",
    "General Inquiry": "❓ How-to questions, policies",
    "Hardware": "πŸ’» Laptop, monitor, keyboard, mouse",
    "Network": "🌐 WiFi, VPN, internet connectivity",
    "Other": "πŸ“‹ Miscellaneous requests",
    "Printing": "πŸ–¨οΈ Printers, scanning, print queue",
    "Security": "πŸ”’ Threats, malware, security incidents",
    "Software": "πŸ“¦ Application issues, installations",
    "Storage": "πŸ“ OneDrive, SharePoint, file storage"
}


class TicketPreprocessor:
    def __init__(self):
        self._email = re.compile(r'\b[\w.-]+@[\w.-]+\.\w+\b')
    
    def clean(self, text):
        return ' '.join(self._email.sub('[EMAIL]', str(text or '')).lower().split())
    
    def combine(self, subject, description):
        return f"[SUBJECT] {self.clean(subject)} [SEP] [DESCRIPTION] {self.clean(description)}"


class TicketClassifier(nn.Module):
    def __init__(self, num_classes, model_name="distilbert-base-uncased", dropout=0.3):
        super().__init__()
        self.bert = DistilBertModel.from_pretrained(model_name)
        self.classifier = nn.Sequential(
            nn.Dropout(dropout),
            nn.Linear(768, 256),
            nn.GELU(),
            nn.Dropout(dropout),
            nn.Linear(256, num_classes)
        )
    
    def forward(self, input_ids, attention_mask):
        outputs = self.bert(input_ids=input_ids, attention_mask=attention_mask)
        return self.classifier(outputs.last_hidden_state[:, 0, :])
    
    def predict_proba(self, input_ids, attention_mask):
        logits = self.forward(input_ids, attention_mask)
        return torch.softmax(logits, dim=-1)


# Load model
print("Loading model...")
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Device: {device}")

try:
    model_path = hf_hub_download(repo_id=HF_REPO_ID, filename=MODEL_FILENAME)
    print(f"Model downloaded: {model_path}")
    
    tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
    model = TicketClassifier(num_classes=len(CLASS_NAMES))
    
    checkpoint = torch.load(model_path, map_location=device)
    if 'model_state_dict' in checkpoint:
        model.load_state_dict(checkpoint['model_state_dict'])
    else:
        model.load_state_dict(checkpoint)
    
    model.to(device)
    model.eval()
    MODEL_LOADED = True
    print("Model loaded successfully!")
except Exception as e:
    print(f"Error loading model: {e}")
    MODEL_LOADED = False

preprocessor = TicketPreprocessor()


def classify_ticket(subject, description):
    """Classify a ticket and return results."""
    if not subject and not description:
        return "⚠️ Please enter a subject or description", "", ""
    
    if not MODEL_LOADED:
        return "❌ Model not loaded", "", ""
    
    try:
        # Preprocess and tokenize
        combined = preprocessor.combine(subject, description)
        inputs = tokenizer(
            combined,
            return_tensors="pt",
            truncation=True,
            max_length=256,
            padding='max_length'
        ).to(device)
        
        # Predict
        with torch.no_grad():
            probs = model.predict_proba(inputs['input_ids'], inputs['attention_mask'])[0]
        
        probs_np = probs.cpu().numpy()
        top_indices = probs_np.argsort()[::-1]
        
        # Primary prediction
        primary_idx = top_indices[0]
        primary_cat = CLASS_NAMES[primary_idx]
        primary_conf = probs_np[primary_idx] * 100
        
        # Status
        if primary_conf >= 80:
            status = "βœ… **High Confidence** - Auto-route recommended"
        elif primary_conf >= 60:
            status = "⚠️ **Medium Confidence** - Review suggested"
        else:
            status = "πŸ” **Low Confidence** - Human review required"
        
        # Format primary result
        primary_result = f"""
## {CATEGORY_INFO.get(primary_cat, primary_cat)}

### Predicted Category: **{primary_cat}**
### Confidence: **{primary_conf:.1f}%**

{status}
"""
        
        # Format alternatives
        alternatives = "### Other Possibilities:\n\n"
        for i in range(1, min(4, len(top_indices))):
            idx = top_indices[i]
            cat = CLASS_NAMES[idx]
            conf = probs_np[idx] * 100
            alternatives += f"- **{cat}**: {conf:.1f}%\n"
        
        # Confidence bar
        conf_display = f"{'β–ˆ' * int(primary_conf / 5)}{'β–‘' * (20 - int(primary_conf / 5))} {primary_conf:.1f}%"
        
        return primary_result, alternatives, conf_display
        
    except Exception as e:
        return f"❌ Error: {str(e)}", "", ""


# Example tickets
examples = [
    ["VPN not connecting", "Cannot connect to corporate VPN from home, getting timeout error"],
    ["Suspicious email received", "Got an email asking for my password, looks like phishing"],
    ["Need SharePoint access", "Just joined the marketing team, need access to the team SharePoint"],
    ["Laptop screen flickering", "My laptop screen has been flickering intermittently since yesterday"],
    ["Outlook not receiving emails", "Haven't received any emails in Outlook for the past 3 hours"],
    ["How to reset password", "What is the process to reset my Active Directory password?"],
    ["Printer not working", "Print jobs stuck in queue and won't print"],
    ["SQL query slow", "Database query that used to take 2 seconds now takes 10 minutes"],
]


# Create Gradio interface
with gr.Blocks(
    title="IT Ticket Classifier",
    theme=gr.themes.Soft(primary_hue="green", secondary_hue="blue"),
    css="""
    .gradio-container { max-width: 900px !important; }
    .primary-result { font-size: 1.2em; }
    """
) as demo:
    gr.Markdown("""
    # 🎫 IT Service Desk Ticket Classifier
    
    **Powered by DistilBERT** | Classifies tickets into 12 IT support categories
    
    Enter a ticket subject and description below to get the predicted category.
    """)
    
    with gr.Row():
        with gr.Column(scale=1):
            subject_input = gr.Textbox(
                label="πŸ“‹ Ticket Subject",
                placeholder="e.g., VPN not connecting",
                lines=1
            )
            description_input = gr.Textbox(
                label="πŸ“ Ticket Description",
                placeholder="e.g., Cannot connect to corporate VPN from home, getting timeout error after 30 seconds...",
                lines=4
            )
            classify_btn = gr.Button("πŸ” Classify Ticket", variant="primary", size="lg")
        
        with gr.Column(scale=1):
            primary_output = gr.Markdown(label="Primary Prediction")
            confidence_output = gr.Textbox(label="Confidence", interactive=False)
            alternatives_output = gr.Markdown(label="Alternatives")
    
    classify_btn.click(
        fn=classify_ticket,
        inputs=[subject_input, description_input],
        outputs=[primary_output, alternatives_output, confidence_output]
    )
    
    gr.Examples(
        examples=examples,
        inputs=[subject_input, description_input],
        outputs=[primary_output, alternatives_output, confidence_output],
        fn=classify_ticket,
        cache_examples=False
    )
    
    gr.Markdown("""
    ---
    ### πŸ“Š Supported Categories
    
    | Category | Description |
    |----------|-------------|
    | Access Management | Login, permissions, MFA |
    | Backup | Backup and restore |
    | Database | SQL, queries, DB issues |
    | Email | Outlook, calendar |
    | General Inquiry | How-to questions |
    | Hardware | Devices, laptops |
    | Network | WiFi, VPN, internet |
    | Other | Miscellaneous |
    | Printing | Printers, scanning |
    | Security | Threats, incidents |
    | Software | Applications |
    | Storage | OneDrive, SharePoint |
    
    ---
    **Model**: DistilBERT fine-tuned on 5,760 IT support tickets
    """)


if __name__ == "__main__":
    demo.launch()