> For the complete documentation index, see [llms.txt](https://web3-growth-agent-wga.gitbook.io/whitepaper_ver2.0_en/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://web3-growth-agent-wga.gitbook.io/whitepaper_ver2.0_en/.-wga-solution-overview-and-structure2/3.-core-components.md).

# 3. Core Components

* **Data Collection Layer**

The data collection layer gathers behavioral data from Web2 and Web3 environments through SDKs, APIs, and pixels.

* **SDK Integration**

Partner apps transmit encrypted user events to the network. The SDK manages event tracking, session management, and initial data validation. Personally Identifiable Information (PII) is hashed.

* **API & Pixel**

These methods are used where SDK integration is not feasible. On-chain data is collected via chain-specific indexers.

* **Privacy & Compliance**

Data collection adheres to GDPR and CCPA regulations. Users retain the right to opt-out, which immediately excludes their data from processing.

* **AI Validation Engine**

  The AI validation engine is the core of WGA. It consists of four independent yet interconnected analytical modules.

  * **Fraud Detection Module**

  1. Bot Pattern Recognition: Detection of anomalous velocity, rhythmic repetition, and sequences diverging from standard human behavioral patterns.

  2. Duplicate Account: Clustering of redundant accounts utilizing device fingerprinting, behavioral pattern similarity, and network connectivity analysis.

  3. Automation Detection: Identification of script-based interactions and signatures associated with automated software tools.

  4. Incentive Gaming: Detection of patterns that satisfy only the minimum requirements for reward eligibility and identification of irregular acquisition paths.

  * **Sentiment Analysis Module**

  1. Interest Intensity: Quantitative measurement of interest levels based on dwell time, scroll depth, and return intervals.

  2. Engagement Quality: Differentiation between incidental clicks and intentional participation through behavioral context analysis.

  3. Response Level: Evaluation of response latency, task completion rates, and connectivity to subsequent actions.

  * **Pattern Recognition Module**

  1. Behavioral Consistency: Assessment of the frequency, stability, and predictability of user actions.

  2. Participation Depth: Stratified analysis of behavioral layers to distinguish between superficial and substantive engagement.

  3. Churn Prediction: Identification of exit indicators, declining engagement patterns, and points of systemic friction.

  * **Network Analysis Module**

  1. Traffic Source: Systematic tracking of acquisition paths, referrers, and campaign origins.
  2. Viral Coefficient: Differentiation between organic and artificial diffusion to measure authentic network effects.
  3. Community Impact: Analysis of roles, influence, and connectivity structures within the community ecosystem.

  These four modules operate independently, with their outputs aggregated within an integrated scoring layer. A single behavioral data point may satisfy the criteria for Fraud Detection while receiving a low score in Pattern Recognition, or vice versa. The final interpretation is derived from the weighted average of the outputs from all four modules.

* **Blockchain Settlement Layer**

Verification results are recorded on the blockchain, consisting of three primary components.

1. Validation Proof: Documentation specifying the AI models and criteria utilized for the verification process
2. Metric Summary: The resulting KPI values derived from the interpretive analysis
3. Timestamp & Hash: Verification timing and cryptographic proof of data integrity.

This record is publicly accessible but immutable. The architecture enables transparent tracking of the origin and verification history of all analytical results.