IIT-BHU hosts third round of Dark Patterns Buster Hackathon; 500 students participates

NEW DELHI: The consumer affairs department organised round 3 of the inter-college competition of Dark Patterns Buster Hackathon (DPBH-2023) at the Indian Institute of Technology (IIT) BHU. The hackathon is aimed towards innovation in designing and prototyping application or software-based solutions such as browser extensions, plugins, add-ons and mobile applications that can detect the use, type and scale of dark patterns for e-Commerce platforms.

In the finale of round 3, more than 500 students, 150 subject experts and jury members participated. Each team presented real-time working models capable of detecting 13 different types of illegal dark patterns as recognised by the union government. Rohit Kumar Singh, secretary DoCA, was the chief guest at the event which was organised in collaboration of the Department and the institute.

Launched in October 2023

The DPBH 2023 was launched by the consumer affairs department on October 26, 2023. This year, DPBH 2023 saw participation from more than 150 colleges and over 40,000 students. A total of 380 teams competed in the second round of the intra-college competition, from which 172 were shortlisted for round 3.

The solutions presented by participating teams were tailored to address 13 specified dark patterns identified by the Department of Consumer. Congratulating the students, secretary DoCA Rohit Singh said the solutions provided by participants will not only assist in detecting dark patterns but will be a major game changer in regulating other online deceptive practices such as fake reviews, online traps, misleading advertisements amongst others.

Also read IIT Delhi denies 132 PhD seats to SC, ST, OBC candidates: APPSC finds

Projects at DPBH-2023 round 3

Some of the projects presented by participating teams at round 3 of DPBH-2023 were following:

Large language models for dark pattern detection: A project proposing a browser extension to detect dark patterns on e-commerce sites.

Serverless architecture for real-time analysis: Solutions featuring a serverless setup using AWS Lambda and cross-encoder machine learning models, with Transformer.js enabling multi-lingual analysis, showcasing scalability and efficiency in identifying "misdirection" and "forced action."

Modular approach for comprehensive protection: Projects adopting modular approach, integrating text and visual analysis with You Only Look Once (YOLO) technology and react for real-time use alerts, tackling a wide range of dark patterns and prioritising user privacy.

Web and mobile platform solutions: Utilising random forest classifiers and Google's Vision API, some solutions are focusing on uncovering "hidden information" through a user-friendly platform developed with flask and react native.