import dash from dash import html, dcc import dash_bootstrap_components as dbc dash.register_page(__name__, path='/', name="Control Room") layout = html.Div([ # --- Paper Header Section --- html.Div([ html.H1("Deep Learning for Financial Statement Fraud Detection", className="text-white fw-bold", style={"fontFamily": "serif", "letterSpacing": "1px", "fontSize": "3rem"}), html.P("A Comparative Analysis of Transformers-based Architectures in SEC Filing Audits", className="text-info text-uppercase fw-light mb-4", style={"letterSpacing": "3px", "fontSize": "0.9rem"}), # Abstract-style Introduction html.Div([ html.P([ html.B("Abstract: "), "This intelligence system leverages Bidirectional Encoder Representations from Transformers (BERT) " "to identify anomalous patterns in corporate filings. By integrating FinBERT and DeBERTa-v3 architectures " "with Parameter-Efficient Fine-Tuning (PEFT), the engine achieves high-precision risk scoring, " "transforming qualitative financial narratives into quantitative fraud probability metrics." ], className="text-muted mx-auto", style={"maxWidth": "800px", "textAlign": "justify", "lineHeight": "1.6"}) ], className="mb-5") ], className="text-center mt-5 mb-5"), # --- Quantitative Performance Metrics (The "Results" Row) --- dbc.Row([ dbc.Col( html.Div([ html.Small("DATABASE SCOPE", className="text-muted fw-bold"), html.H2("1,284", className="text-white fw-bold mt-2"), html.Div(style={"height": "2px", "width": "40px", "background": "#00d2ff", "margin": "10px 0"}), html.P("Processed Document Samples", className="text-muted small") ], className="modern-card p-4", style={"borderLeft": "1px solid #00d2ff"}), width=4 ), dbc.Col( html.Div([ html.Small("ANOMALY DETECTION", className="text-muted fw-bold"), html.H2("12", className="text-white fw-bold mt-2"), html.Div(style={"height": "2px", "width": "40px", "background": "#dc3545", "margin": "10px 0"}), html.P("Validated Risk Sigils", className="text-muted small") ], className="modern-card p-4", style={"borderLeft": "1px solid #dc3545"}), width=4 ), dbc.Col( html.Div([ html.Small("SOTA BENCHMARK (F1)", className="text-muted fw-bold"), html.H2("93.4%", className="text-white fw-bold mt-2"), html.Div(style={"height": "2px", "width": "40px", "background": "#9d50bb", "margin": "10px 0"}), html.P("DeBERTa-v3 Engine Performance", className="text-muted small") ], className="modern-card p-4", style={"borderLeft": "1px solid #9d50bb"}), width=4 ), ], className="mb-5 g-4"), # --- Methodology & Execution --- dbc.Row([ dbc.Col([ html.Div([ html.H4("Experimental Protocol", className="text-white mb-3", style={"fontFamily": "serif"}), html.P([ "The detection engine utilizes cross-attention mechanisms to parse semantic irregularities. " "Select 'Launch Detection' to initiate the inference pipeline on raw unstructured text data." ], className="text-muted mb-4"), dbc.Button("Launch Inference Engine", href="/detection", className="px-5 py-2", style={ "background": "transparent", "border": "1px solid #00d2ff", "color": "#00d2ff", "borderRadius": "0", "textTransform": "uppercase", "letterSpacing": "2px" }) ], className="modern-card p-5", style={"border": "1px solid rgba(255,255,255,0.05)"}) ], width=12) ]) ])