Augmented Analytics and Alibaba Mama's AutoInsight System: Concepts, Architecture, and Applications

Augmented analytics, a concept introduced by Gartner in 2017, represents the next-generation data and analytics paradigm that leverages machine learning to automate data preparation, insight discovery, and insight sharing for business users, operations staff, and citizen data scientists. It enhances human capability to evaluate data through statistics, AI, and ML, moving beyond traditional tools.

The technology roadmap divides augmented analytics into three categories: enhanced data preparation (visual data interaction and automated relationship discovery), enhanced data analysis (automated insights, automated visualization, natural language query and generation), and enhanced machine learning (AutoML and automated online learning platforms). These capabilities aim to reduce manual effort in data workflows.

Automated insights, a core function, replace part of analysts’ work by detecting associations, anomalies, performing diagnostic analysis, and providing intelligent recommendations, thereby enabling deep business analysis and decision-making.

Alibaba Mama’s intelligent insight system AutoInsight implements these ideas. Its architecture consists of four layers: data source access & ETL module, metadata configuration service, InsightCore service layer, and copy display service layer. InsightCore includes anomaly detection, insight operators, and TopK‑Insights intelligent recommendation, following a three‑step strategy: anomaly discovery, fluctuation/trend analysis, and significance scoring.

To achieve high automation and personalized diagnosis, AutoInsight addresses multi‑dimensional drill‑down with dimension pruning, metric contribution analysis (including additive, ratio‑based, and multiplicative metrics), and insight personalization via cold‑start statistical rules and machine‑learning models trained on user feedback.

Two application cases are presented: (1) intelligent diagnosis on advertising placement platforms, where AutoInsight provides data analysis and opportunity insight modules for pre‑, mid‑, and post‑campaign optimization; (2) intelligent warning on mobile placement platforms (e.g., DingTalk’s QianNiu app), delivering real‑time metric anomaly alerts to improve timeliness of business decisions.

The article concludes that augmented analytics brings greater automation and innovative insight to marketing analysis, and that continued development of AutoInsight will expand its capabilities, improve analytical efficiency, and enable personalized insight sharing through intelligent strategies and ML algorithms.