# 6. Beacons: Working with Unstructured Data

Traditional analytics platforms excel at processing **structured data**—numerical values, categories, and predefined metrics. However, **unstructured data**—such as **customer reviews, call transcripts, support tickets, and social media comments**—remains an untapped resource for many businesses.

**Beacons** are our cutting-edge mechanism that **leverages Generative AI** to extract insights from **unstructured data**, transforming raw text into **meaningful, actionable intelligence**.

Unlike other analytics platforms, Infinity allows users to **not only analyze structured data** but also **extract deep insights from free-text content**. This provides a **powerful, competitive advantage**—enabling businesses to understand **customer sentiment, identify recurring themes, and uncover hidden trends** that traditional analytics tools cannot capture.

***

## **Use Case: Extracting Sentiment & Themes from Disneyland Reviews**

Imagine you have a dataset named **disneyland\_reviews** containing the following columns:

| review\_id | reviewer\_name | date       | rating | comments                                                                        |
| ---------- | -------------- | ---------- | ------ | ------------------------------------------------------------------------------- |
| 1          | Alice          | 2024-02-20 | 5      | Loved the rides, but food was overpriced.                                       |
| 2          | Bob            | 2024-02-21 | 3      | The wait times were unbearable, but the staff was friendly.                     |
| 3          | Charlie        | 2024-02-22 | 2      | Food was terrible, and the roller coasters broke down.                          |
| 4          | Daisy          | 2024-02-23 | 4      | Great shows, but too crowded to enjoy properly.                                 |
| 5          | Eve            | 2024-02-24 | 1      | The worst experience! Staff was rude, food was cold, and long lines everywhere! |

#### **The Problem with Traditional Analytics**

Most analytics platforms can **only plot numerical ratings**, such as:\
📉 Average Rating Over Time\
📊 Ratings by Month or Reviewer

However, these insights **lack context**. Why did a customer give **1 star**? Was it due to **bad food, rude staff, or broken rides**? Traditional tools **cannot extract themes or sentiment** from the reviews themselves.

#### **How Beacons Solve This Problem**

With **Beacons**, Infinity automatically processes **commentary data** to extract:\
✅ **Sentiment Analysis** (Positive, Neutral, Negative)\
✅ **Key Themes** (Food complaints, Ride issues, Wait times, Staff friendliness, etc.)

**Beacon-Generated Insights**

| review\_id | sentiment | themes extracted                  |
| ---------- | --------- | --------------------------------- |
| 1          | Positive  | Good rides, Overpriced food       |
| 2          | Neutral   | Long wait times, Friendly staff   |
| 3          | Negative  | Bad food, Ride malfunctions       |
| 4          | Positive  | Great shows, Crowded              |
| 5          | Negative  | Rude staff, Cold food, Long lines |

#### **Integrating Insights with Other Data**

Once Beacons process the reviews, users can:\
🔗 **Merge the sentiment data with structured tables** (e.g., CRM or sales data)\
📊 **Plot trends over time** (e.g., complaints about food increasing over summer)\
📉 **Analyze correlations** (e.g., do negative reviews correlate with staff shortages?)

For example, if you **merge the review insights with CRM data**, you can see if unhappy customers **had previous complaints, lower lifetime value, or requested refunds**.

***

## **Why Beacons Are a Game-Changer**

🚀 **No other analytics platform on the market** can automatically extract **themes and sentiment** from unstructured text and **make that data usable** in traditional BI workflows.

📈 With **Beacons**, Infinity users can **unlock hidden insights** and transform unstructured data into **actionable business intelligence**—without requiring manual tagging, complex NLP models, or external AI tools.

This is the **future of analytics**—beyond numbers, into the **realm of understanding and prediction**.