Lianjia's Big Data Platform Practice: Building a One‑Stop Data Service for Real Estate

During the QCon Global Software Development Conference in Beijing, platform big‑data architect Li Xiaolong (a 2012 graduate of Wuhan University who later worked at Baidu and joined Lianjia in 2016) delivered a talk titled “Lianjia Big Data Platform Practice,” sharing the overall construction and operation of Lianjia’s data platform.

Lianjia’s data is divided into three major categories—House (over 80 million detailed property records with more than 300 attributes), People (150,000 professional agents and over 40 million app users), and Behavior (massive online and offline interaction data such as browsing, clicks, follows, comments, viewings, and signings).

The platform supports four primary application scenarios: BI (business intelligence for all business lines), AI (person‑, house‑, and client‑level profiling, personalized search, recommendation, and knowledge‑graph construction), 2C (consumer‑facing features like house valuation and market trends in the Lianjia app), and 2B (RealData Smart Real‑Estate solutions for developers, financial institutions, and governments).

To meet rapidly growing data demands across 28 cities, Lianjia built a one‑stop big‑data platform consisting of five layers—business, data, platform‑service, data‑product, and user layers. The platform now stores nearly 10 PB of data, handles over 50 million daily API calls, runs more than 25 000 scheduled tasks per day, and manages over 13 000 metadata tables.

Li also introduced the DIKW pyramid model: the data layer is managed by a data‑management platform; the information layer provides scheduling and ad‑hoc query services; the knowledge layer delivers unified metric reporting and visual analytics, enabling the transformation of raw data into actionable insights.

Looking ahead, Lianjia plans a three‑phase roadmap: (1) completing a company‑wide platform to enhance data capabilities and user experience; (2) opening the platform for co‑construction across the Lianjia group; and (3) evolving into an industry‑level real‑estate data platform that becomes the benchmark for comprehensive, accurate property data.