AI-powered content decoding app for a media startup to drive emotional engagement with customers

Client

A US-based startup

Industry

Media, Advertising, Technology

Services

Product Development, AI, NLP

Tech

AWS, REST API, React, TypeScript, Python, NLTK Library, Flask, Recharts, Swagger Codegen, TSLint

Challenge

Creators of an AI-enabled methodology that applies humanistic psychology to decipher emotions in content were looking for a vetted AI-development partner that could turn their innovation into a successful commercial product. The client wanted to enable media companies to detect character traits, such as creativity, social intelligence or leadership, in both content and audiences for reaching out with emotionally intelligent messages. They turned to ITRex for assistance with developing a fully functional web application using natural language processing and artificial intelligence.

Our team took on the following challenges:

●

Provide a business analyst to gather requirements and formulate a full-scale vision of the future product on the development progress

●

Rebuild the client’s existing app by splitting the monolith infrastructure into back end and front end for separate development and maintenance

●

Refactor the back end to improve its architecture and prepare documentation for API

●

Implement Natural Language Processing for data mining and analysis

●

Refactor a MATLAB AI algorithm for decoding content

●

Pack the system with multiple features

●

Implement a visualization solution to display content analysis results

Solution

●

A minimum viable product (MVP) with a new back-end architecture and logic

●

A refactored AI algorithm that deciphers emotions in content and users’ data, tags both the content and users with specific character traits based on a scientific classification system of more than 20 character traits and common language taxonomy, and matches content to audiences for targeting

●

A Python program based on NLTK for working with human language data

●

REST APIs hosted on AWS for scalability and availability

●

A light-weight API using Flask

●

A powerful visualization of analyzed content and data on an open-source radar chart, customized for the product

●

The user’s ability to add text corpora and train the models through their analysis, as well as add pre-trained models and export character visualizations

●

Display of audience sizes with specific character traits, links to other research on the text and recommendations on engaging content for the audience

Impact

●

As global entertainment leaders are stepping into a new era of emotional engagement, the app enables them to plug emotions into relationships with customers, including in content management, advertising and promotion, and sales

●

The solution empowers creative executives to predict the emotional response of their audiences to specific content and target them better with emotionally framed messages

●

Advertisers can buy audiences with character traits they are interested in for increased profits