Search
Close this search box.

MAIA

Multilingual Virtual Agents for Customer Service

Online conversational support – chat – is the fastest-growing customer service channel, being the preferred way for millennials to obtain customer service. Today, supporting international customers in this channel is mostly done using human agents that speak different languages – a scarce and costly resource.

The project MAIA will develop a multilingual conversational platform, supported by machine translation and dialogue systems, where AI agents assist human agents. This approach will overcome the limitations of existing customer service.

Keywords: AI . Machine Translation . Conversational support


Leading company in Portugal
UNBABEL, LDA

Project Start Date
01/04/2020

Project End Date
01/04/2023

MAIA, enabling chats to develop multilingual conversations

MAIA project will focus on improving chats [message applications] to develop multilingual conversations by expanding customer service professionals’ skills, using artificial intelligence, and making this support process more efficient. To do so, the MAIA consortium will develop a set of machine learning technologies that will allow automatic translation and responses. Among the main challenges is the need to consider the conversational context to ensure accurate, correct, and culturally appropriate translations

Whattoexpect

What to expect

The ultimate goal of MAIA will be to develop multilingual chats using AI, allowing companies to chat with their customers in 30 different languages

Meetthepartners

Meet the partners

 Promoter:
UNBABEL– João Graça and Paulo Dimas

Academic Co-promoters:
Instituto de Telecomunicações (IT) – André Martins
INESC-ID – Helena Moniz 

CMU:
Language Technologies Institute – Graham Neubig

Goals

Goals

MAIA will target the following scientific and technological goals:

  • New memory-efficient neural models for context-aware machine translation
  • New answer generation techniques to support the decisions of human agents
  • New techniques for conversational quality estimation and sentiment analysis
  • Integration of the scientific advances above into a full end- to-end product

Two demonstrators will be built to cover concrete use cases in the Travel and Tourism Industries.

Follow the project

IN THE NEWS

ON THE WEB

PUBLICATIONS