Turkish students develop AI-powered system to analyze customer reviews

A talented team from Ondokuz Mayis University’s (OMU) Bafra Vocational School of Computer Science, named Bafra 4.0, has developed a cutting-edge AI system designed to analyze customer reviews.

This innovative project earned them first place at the Teknofest Turkish Natural Language Processing Competition held in Kocaeli’s Gebze district.

Competing against 86 other teams, the Bafra 4.0 team emerged victorious with their Entity-Based Sentiment Analysis system. This remarkable achievement now has the students eagerly anticipating their award ceremony, which will take place during the Teknofest event scheduled for Oct. 2-6 in Adana.

Advanced AI for customer feedback analysis

The Teknofest competition focuses on creating technology solutions in the field of Turkish natural language processing (NLP). The goal of this project is to build international technology brands in this domain.

Students participating in the competition worked on generating data, utilizing open-source tools, developing datasets, and converting unmarked data into labeled formats.

The competition was divided into two categories: Free and NLP scenario. The Bafra 4.0 team, consisting of second-year computer science students Hicran Temel, Ahmet Anil Sariguzel, and Aysegul Topcu, under the guidance of lecturer Mustafa Kaya, secured first place in the NLP category.

Bringing innovation to NLP

Lecturer Mustafa Kaya expressed his delight in their achievement, highlighting the hard work that went into the project. “We participated in the scenario category because I believed I could guide my students well in this area,” Kaya explained.

The team was given a scenario and tasked with finding solutions tailored to it. Kaya elaborated, “The goal was to develop a system that detects customer feedback – whether positive, negative or neutral – about various companies. The key to success in such projects is creating a robust dataset.”

The students worked diligently to design a program capable of analyzing customer reviews effectively. Their innovative approach utilized a hybrid, transformer-based model, which ultimately led to their victory.

First-time competitors triumph

Competing for the first time, the Bafra 4.0 team outperformed 86 other teams, showcasing the effectiveness of their novel approach to natural language processing. “It was a great honor to win, especially as first-time participants,” Kaya said. “My goal is to inspire my students to love natural language processing and recognize the possibilities it offers.”

Student Aysegul Topcu described the process: “We gather comments from social media, analyze them, and determine whether the feedback about brands is positive or negative. Our system learns from the data we provide, enabling it to make accurate predictions about brand sentiment.”

Ahmet Anil Sariguzel, another team member, was responsible for creating the program’s user interface theme.

This impressive win marks the beginning of what promises to be a bright future for the Bafra 4.0 team as they continue to innovate and represent their university and city with pride.