Leobit
AI-Powered Sentiment Analysis Solution
Custom development of a PoC that classifies the tone and the emotional context of text messages
ABOUT the project
- Client:
- Leobit's Internal Project
- Location:
-
USA
- Company Size:
- 100+ Employees
- Industry:
-
Information Technology
- Solution:
- Custom Software
Leobit has developed a PoC for an AI-powered solution capable of analyzing text and identifying its overall mood and tone. An AI model processes a message and assigns a confidence score to major tones of voice, as well as identifies the general mood of the text.
We effectively leveraged the capabilities of Azure Language API for analyzing text mood and sentiment. The solution can be easily customized, and we can configure it to recognize even more types of emotions and tones of voice across texts of different sizes and complexity.
Customer
It was Leobit’s internal project designed to explore the capabilities of Azure Language API, as well as support our internal workflows, including those of our sales and HR teams.
Business Challenge
Understanding the mood of the message is a very important feature across industries, especially those that rely heavily on customer interactions. By reviewing communication patterns, businesses can understand the intent of potential customers, HR managers can assess the soft skills of employees and potential candidates, and public service organizations can evaluate overall citizen sentiment.
Project
in detail
Our team developed the solution in around a week. Most of the work was dedicated to configuring the solution’s logic.
Our specialists came up with an idea of the solution that would use Azure Language API to analyze text sentiment. We planned the solution’s functionality for sending HTTP requests to the service, as well as the core tech stack for the product.
We used Ruby to build the solution’s back end. This language is known for its simplicity and rapid development, which were ideal parameters for building a PoC fast. We also used React.js and Tailwind CSS to build a simple yet user-friendly PoC interface.
Our specialists created a complex integration layer capable of aggregating data from various sources (emails, tickets, chat messages, etc.), processing them, and orchestrating their interaction with Azure Language API. The service was custom-configured to identify text tone, style, and emotional context, playing a key role in the tool’s sentiment analysis capabilities.
In-Depth Analytics on Text Mood and Tone
The solution’s AI core processes messages and assigns confidence scores to key tones of voice:
- Positive
- Negative
- Neutral
- Mixed
It also determines the overall sentiment by identifying the text’s primary emotion (e.g., disappointed or friendly). Such capabilities help employees in customer-facing roles accurately assess the tone of client messages, understand intent, and adjust their communication style accordingly. This leads to more effective interactions. The tool can also be used in HR management, recruitment, public sentiment analysis, and other domains where understanding and managing tone of voice is important.
User-friendly UI
The solution features a simple, efficient UI that makes its functionality easy to explore. Users can enter text manually or analyze email tone and mood by selecting sample messages from the “Emails” tab. Upon processing the message, the tool provides a detailed report in a clear format, along with a visual representation of the text’s mood.
Rich Potential for Application Across Industries
We have a detailed roadmap for expanding the solution to a production-grade cloud-based solution. It can be relevant for businesses across multiple domains, including:
- HR managers can determine employees’ mood and soft skills
- Recruiters can use the tool’s assistance to evaluate candidates’ soft skills
- Marketing teams can use the tool for monitoring social media sentiment
- E-commerce businesses can use the tool’s assistance to analyze customer intent and mood
- Government and public service organizations can benefit from the solution’s capabilities for analyzing social sentiment
Technology Solutions
- Custom configuration of Azure Language API allowing the service to identify text mood, tone, and emotional context.
- A complex integration layer that uses off-the-shelf HTTP requests and a Puma 6.5 web server, which ensures effective processing of concurrent HTTP requests.
- Minimalistic yet efficient UI/UX design built with React.js and Tailwind CSS.
- Roadmap for the continuous expansion of a solution to a full-fledged cloud-based system using Docker containerization.
Value Delivered
- Effective recognition of mood and tone of voice across various types of texts.
- Successful examination of AI capabilities for in-depth text analysis.
- A framework that can be reused and expanded for delivering a production-grade solution for businesses across industries.
