ARACHNET_SYS_V4.0
// PROJECT ARCHITECTURE SPECIFICATION

Arachnet Intelligence Terminal

Automated financial news aggregation, NLP clustering, and sentiment-driven market analytics pipeline

Python Flask SQLite scikit-learn NLTK VADER TF-IDF Plotly yfinance Live Data
01

Scraping Engine

Three-tier anti-bot evasion pipeline. Tier 1 cycles rotating real browser headers via requests. Tier 2 escalates to CloudScraper for TLS fingerprint spoofing. Tier 3 deploys headless undetected-chromedriver when static layers fail — ensuring maximum data ingestion resilience.

02

Parsing & ETL

Dual-strategy parsing engine: JSON-LD script extraction for structured page metadata and HTML DOM traversal for rendered layouts. Config-driven per-source, stored in arachnet.db. Pandas deduplication normalizes headlines cross-site before SQL insertion.

03

Clustering Engine

TF-IDF vectorization converts raw headlines and summaries into numerical feature matrices. Cosine similarity finds semantically related articles across sources. Levenshtein + Jaccard metrics provide precision verification. A 72-hour freshness gate prevents stale topic merging.

04

Scoring & Sentiment

VADER lexicon analysis produces compound sentiment scores bounded at ±1.0. A non-linear volume saturation curve prevents high-volume clusters from dominating. Historical scores undergo 33.3% daily exponential decay to surface emerging trends over stale narratives.

Pipeline Execution Sequence

Sources DB Scraping Engine Parsing Engine ETL / Dedup articles.db Clustering Engine Score & Store Flask Dashboard

Orchestrated by run_sentiment_analysis.py with OS-level singleton locking to prevent concurrent pipeline collisions.

Live Database Metrics

Total Articles 1051
Active Clusters 416
News Sources 2
Macro Categories 5