A platform for Zhuanzhuan R&D and industry peers to learn and exchange technology, regularly sharing frontline experience and cutting‑edge topics. We welcome practical discussions and sharing; contact waterystone with any questions.
This article describes a two‑stage pipeline that first collects and processes product pair data, uses a large language model to score pair compatibility, and then fine‑tunes Qwen‑7B and distills its knowledge into a BERT model to enable fast, online recommendation of well‑matched product combinations.
This guide explains how to communicate effectively with the Cursor AI coding assistant by crafting high‑quality prompts, using context cues, managing multi‑turn dialogues, and configuring persistent rules, offering practical principles, examples, and actionable recommendations to boost development productivity.
This article investigates a puzzling MySQL behavior where a row updated within a transaction sometimes returns its previous value, explains the role of InnoDB's snapshot reads and ReadView, reproduces the issue, and offers practical ways to prevent the “update disappearance” scenario.
This article chronicles the evolution of a quality‑inspection system from a simple finite state machine to a self‑developed, lightweight workflow engine, detailing the motivations, design decisions, implementation steps, challenges faced, and future roadmap for flexible, configurable process orchestration.
This article introduces MyBatis-Plus, explains its core zero‑injection CRUD capabilities, demonstrates powerful condition builders, plugin extensions, and best practices for secure, efficient data access, including custom query objects, micro‑service integration, and advanced mapper customization, helping developers boost backend productivity.
Facing massive daily customer service data, ZhiZhi built the AI-driven Minos quality inspection system, combining inspection items, plans, and tasks with large models, regex and programmatic checks, achieving a 26‑fold detection boost and processing over 20,000 interactions per day.
This article details how an AI-driven Voice of Customer (VOC) recommendation system was designed, trained, and deployed to automate tag selection and conversation summarization, resulting in a 37.9% reduction in selection time, a 428% boost in recommendation accuracy, and significant labor savings for customer service agents.
This article details how Zhaozhuan transformed its customer‑service robot from relying on external NLP providers to a fully self‑developed AI solution, covering technical choices, model training, RAG integration, deployment strategies, and future directions.
This article introduces Easy-Cache, a Spring‑AOP based caching framework that eliminates repetitive cache code by offering annotation‑driven operations, multi‑level Redis and local caches, dynamic upgrade/downgrade, elastic expiration, and Lua‑script‑backed consistency mechanisms for high‑availability distributed systems.
This article explains how Event Storming—a collaborative, sticky‑note‑driven workshop—helps teams break down complex business processes into events, commands, aggregates and policies, dramatically speeding up requirements analysis, improving model quality, and turning the resulting model directly into code and living documentation.
