# Data Sources

### Data Sources Overview

The Onairos platform integrates with multiple social media, communication and other platforms to collect user behavioral data, which users allow to train their personal AI model. \
\
Our system doesn't expose third-party platform APIs directly **OR RAW USER DATA**. Instead, the trained user model generates personalized insights and predictions valuable for your context window, ML algorithms and more.

### Supported Data Sources

* [Instagram](https://onairos.gitbook.io/docs/data-sources/instagram)
* [Youtube](https://onairos.gitbook.io/docs/data-sources/youtube)
* [X](https://onairos.gitbook.io/docs/data-sources/x-twitter)
* [Reddit](https://onairos.gitbook.io/docs/data-sources/reddit)
* [Pinterest](https://onairos.gitbook.io/docs/data-sources/pinterest)
* [Linkedin](https://onairos.gitbook.io/docs/data-sources/linkedin)
* [Gmail](https://onairos.gitbook.io/docs/data-sources/gmail)
* [Tiktok](https://onairos.gitbook.io/docs/data-sources/tiktok)

### Onairos AI Model Outputs

Our AI models process collected data to provide four core prediction types:

#### 1. Personal Traits Analysis

**What It Provides:** Unique personality traits and behavioral patterns specific to each user based on their digital footprint across platforms.

**Example Outputs:**

* Communication style preferences
* Content consumption patterns
* Social interaction tendencies
* Professional vs personal persona differences
* Risk tolerance and decision-making patterns

**API Endpoint:**

```
GET /api/v1/traits/{user_id}
```

***

#### 2. Sentiment Prediction

**What It Provides:** Neural network-powered sentiment scoring (0-1 scale) predicting user response to specific content combinations.

**Scoring System:**

* `0.98` = User will highly engage/like this content
* `0.65` = Moderate positive response expected
* `0.23` = Low engagement/negative response likely
* `0.05` = User will likely dislike/ignore this content

**Input Types:**

* Text + Image combinations
* Video content analysis
* Product recommendations
* Social media posts
* Marketing materials

**API Endpoint:**

```
POST /api/v1/sentiment/predict
{
  "user_id": "string",
  "content_type": "text_image|video|product",
  "content_data": {...}
}
```

***

#### 3. Emotion Recognition (Coming Soon)

**What It Will Provide:** Advanced emotional state detection and prediction based on multi-platform behavioral analysis.

**Planned Features:**

* Real-time mood detection from posting patterns
* Emotional response prediction to content types
* Stress and wellness indicators from communication patterns
* Emotional compatibility matching for social interactions
* Temporal emotion mapping (daily/weekly emotional cycles)

**Targeted Release:** Q4 2025

***

#### 4. Connection Intelligence (Coming Soon)

**What It Will Provide:** Advanced relationship analysis and social network intelligence that goes far beyond basic friendship mapping.

**Core Features:**

**Relationship Strength Analysis:**

* **Super Friends**: Ultra-close connections with daily interaction, shared interests, and mutual influence
* **Influence Networks**: Users who significantly impact decision-making and content preferences
* **Professional Allies**: Career-focused connections with collaboration patterns and professional support
* **Interest Tribes**: Users connected through specific hobbies, topics, or communities
* **Emotional Support Networks**: Connections that provide comfort, advice, and emotional stability

**Social Dynamic Mapping:**

* **Conversation Catalysts**: Users who spark engagement and drive discussion
* **Trend Amplifiers**: Connections who boost content virality and social reach
* **Opinion Leaders**: Individuals whose preferences influence user behavior
* **Social Bridges**: Connections linking different social circles or communities
* **Trust Indicators**: Relationship authenticity and genuine connection strength

**Interaction Pattern Analysis:**

* **Communication Rhythms**: How often, when, and in what context users interact
* **Content Sharing Patterns**: What type of content users share with specific connections
* **Mutual Influence Scoring**: How much each user influences the other's behavior
* **Social Context Recognition**: Understanding relationship context (work, family, hobby, etc.)
* **Emotional Resonance**: Which connections share similar emotional responses and triggers

**Privacy-First Implementation:** All connection analysis operates on anonymized interaction patterns and behavioral signals, never exposing personal communication content or private relationship details.

**Planned Applications:**

* Enhanced content personalization based on friend preferences
* Social recommendation systems for meaningful connections
* Community building and group formation assistance
* Professional networking optimization
* Emotional well-being through relationship health insights

**Targeted Release:** Q4 2025