This article introduces causal analysis, outlines its major challenges such as correlation versus causation, confounding factors, and selection bias, explains a three‑step framework (association, intervention, counterfactual), and details how Beike applied these principles in a smart client‑management tool with rigorous A/B experiments.
The article details Beike's OLAP platform built on Druid, covering its four‑layer architecture, selection criteria, performance comparisons with Kylin, data ingestion workflows, custom improvements for data import and real‑time distinct counting, and operational measures such as caching, dynamic throttling, and HDFS storage optimization.
Beike adopted Flutter to address rapid growth and staffing shortages, achieving over 80% hybrid development, native‑like performance, one‑second hot‑reload cycles, extensive community package usage, and a systematic Flutter infrastructure across dozens of mobile applications.
This article introduces the concept of user preference mining, discusses challenges such as accurate expression, interpretability, and high-dimensional preferences, reviews statistical and model-based approaches including weighting, decay, XGBoost, DNN, LSTM, Seq4Rec, and Deep Interest Network, and describes their practical implementation at Beike.
This article details the evolution of Beike's commercial strategy algorithm platform, describing its business scenarios, bidding mechanisms, architecture redesign across online, near‑real‑time, and offline layers, model training, vector retrieval, service governance, and the performance and stability improvements achieved.
The article details Beike's intelligent recommendation platform, describing its C‑end and B‑end user scenarios, the challenges of handling numerous recommendation scenes and material types, and how a strategy‑driven, multi‑stage architecture—from rapid V1.0 construction to V4.0 deep‑model integration—has been evolved, optimized for stability, real‑time processing, and future search‑recommendation convergence.
This article analyzes Beike's search and recommendation platform evolution—from a simple Solr service to a unified cloud‑based mid‑platform—detailing architectural stages, efficiency gains, cost reductions, stability improvements, and future plans for a one‑stop optimization platform.
Beike’s OLAP platform has progressed from a basic Hive‑MySQL batch pipeline to a Kylin‑based single‑engine solution, and now to a flexible multi‑engine architecture that uses a query‑engine layer to route metrics across Kylin, Druid, ClickHouse and Doris, dramatically cutting cube‑build times, supporting real‑time ingestion, and paving the way for further engine consolidation and automated performance routing.
This article details Beike's journey of storing and querying a 480‑billion‑triple industry graph in milliseconds, covering graph database fundamentals, a comparative evaluation of JanusGraph and Dgraph, the design and deployment of a Docker‑K8s based Dgraph platform, data ingestion pipelines, a custom Graph‑SQL layer, performance testing, optimizations, and future roadmap.
The article details Beike's Data Management Platform (DMP) built since May 2018, covering its overall architecture, data collection, processing, real-time profiling, storage solutions, application scenarios, achieved performance metrics, and future development directions.
The article presents a comprehensive case study of Beike's Hermes real‑time computing platform, detailing its business evolution, Hermes architecture, SQL V1/V2 editors built on Spark and Flink, large‑scale deployment statistics, monitoring, diverse business use cases, and planned future enhancements.
This article presents an in‑depth overview of Beike's four‑layer streaming data platform, covering the foundational infrastructure, capability aggregation, data content, and output layers, as well as the challenges of metadata management, real‑time processing, and productization through the Ark and Tianyan systems.
