The company, established in late 2015, said its QA system is being used not only for digital chatbots but also for the Smart Watch from broadband carrier LG U+ and smart speakers from telecom giant KT.
The venture’s foray into Britain came after it was accepted to the Global Entrepreneurship Program run by Britain’s Department of International Trade. 42Maru said its deep learning-based system is deemed superior to existing ontology-based or ruled-based solutions.
At the heart of 42Maru’s QA system is the Machine Reading Comprehension engine, which had joined the MRC competition organized by Stanford University and competed with the Google AI Team in November 2018. The MRC engine allows computers to understand the meaning of a passage and to find answers to any given question.
42Maru said it recently provided its QA system to Daewoo Shipbuilding & Marine Engineering to help deal with a number of technical inquiries and contract-related issues in a timely manner.
“We have already completed deep learning (sessions) on about 100,000 sets of data,” a Daewoo official said. “Since the system becomes smarter as we continue to use it, we expect it to evolve into a more accurate and powerful system through additional learning processes.”