Bridge Language Gaps: Join Alibaba’s Cross‑Language Short Text Matching Challenge

What is Ali Xiaomi?

Ali Xiaomi is a service robot from Alibaba Group’s Intelligent Services Division, focused on e‑commerce. It leverages AI algorithms such as natural language processing, knowledge graphs, and deep learning to provide intelligent dialogue, assistance, decision‑making, and management, supporting multiple languages including English, Russian, Portuguese, Spanish, Indonesian, and Thai.

Background of the CIKM AnalyticCup

The competition, named CIKM AnalyticCup, is an international data‑mining contest organized in Turin, Italy, jointly hosted by Alibaba Cloud and the Shenzhen Meteorological Bureau last year. Participants built precipitation‑forecast models that improved RMSE by 25% over the baseline. This year, Alibaba Group and the Ali Xiaomi team co‑organize the event again.

Why a Cross‑Language Short Text Matching Task?

Chatbots rely heavily on text‑matching models to retrieve appropriate answers from FAQ datasets. Building such models requires large, manually annotated corpora, which are scarce for low‑resource languages (e.g., Indonesian, Thai, Filipino). The challenge aims to transfer knowledge from resource‑rich languages to these low‑resource languages, addressing data scarcity and the shortage of NLP researchers for those languages.

Recent Research from the Ali Xiaomi Team

In the past year, the team has advanced work in text matching and transfer learning, publishing papers at conferences such as ACL, WSDM, and CIKM. Notably, a 2018 WSDM paper introduced a transfer‑learning framework for text‑matching models to alleviate cold‑start problems in the same language.

Invitation to Participate

The competition is open to academia, research institutions, and industry. Participants are encouraged to focus on language characteristics and transfer ability rather than machine‑translation techniques, and to propose innovative models and solutions.

For details and registration, visit the Alibaba Tianchi platform.