☞☞☞AI 智能聊天, 问答助手, AI 智能搜索, 多模态理解力帮你轻松跨越从0到1的创作门槛☜☜☜
{\"type\":\"doc\",\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"322102da-fc53-432a-a749-f406b3123740\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"大家好，我是小耶，写功课只是为了我踩过的坑，你们别再踩了！\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"c852f756-db34-4085-9b0a-829b104760a8\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"传统的数据库执行sql很“机械”：你写什么，它就执行什么。但最近几年，数据库开始变得更“聪明”——它不再逐字逐句执行你的指令，而是尝试“理解”你真正想查什么。这就是所谓的\"},{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"语义执行\"},{\"type\":\"text\",\"text\":\"。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"28132848-3a12-4578-b2ec-deb3af35b206\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"从运营转行做dba，我经常遇到这种情况：业务方说“我要查上个月的活跃用户”，他描述的是一个业务语义，但写出来的sql可能扫了全表。如果数据库能理解“活跃用户”背后的含义（比如最近30天有登录记录），自动优化查询路径，很多慢查询根本不会发生。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"be1d60fb-96ec-4d42-939f-b40d5c92b993\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"传统执行方式的局限性\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"2ae6fec1-bc7d-440a-a4b4-ebcb27c96dc6\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"传统数据库执行sql的过程大致是：解析→优化→执行。优化器基于表统计信息和代价模型，选择它认为最快的执行计划。但优化器并不“理解”业务语义。比如你写\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"select * from orders where order_date \u003e '2026-01-01'\"},{\"type\":\"text\",\"text\":\"，优化器只知道这是一个范围扫描，不知道你真正想要的是“今年的订单”。如果\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"order_date\"},{\"type\":\"text\",\"text\":\"列有索引还好，没有索引就全表扫描。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"6f092d9e-8896-469f-b5f4-c1ac1f5a68ca\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"更麻烦的是，当sql写得不那么“标准”时，优化器可能选错执行计划。比如用子查询而不是join，或者用\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"not in\"},{\"type\":\"text\",\"text\":\"而不是\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"not exists\"},{\"type\":\"text\",\"text\":\"。数据库只是机械执行，不会帮你“纠正”写法。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"00519f14-2327-4920-8c87-18c15a86250c\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"语义执行与传统执行方式的对比\"}]},{\"type\":\"table\",\"attrs\":{\"id\":\"7f1100a7-c691-48b3-9e05-3950c22a1878\",\"ishoverdraghandle\":false},\"content\":[{\"type\":\"tablerow\",\"content\":[{\"type\":\"tableheader\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap; font-weight: 600;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"468a56d0-3536-427f-8cd0-5e917d093d52\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"对比维度\"}]}]},{\"type\":\"tableheader\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap; font-weight: 600;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"8c48fb58-507c-4748-a10b-3e1963c67405\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"传统执行\"}]}]},{\"type\":\"tableheader\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap; font-weight: 600;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"7abc667a-eda9-4ebf-b357-8920f9e8b640\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"语义执行\"}]}]}]},{\"type\":\"tablerow\",\"content\":[{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"ec89d72b-391c-416f-a59d-f3ccd4bb0e63\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"理解层次\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"cff026c1-64fc-4bee-9f6e-879876fa1667\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"语法级别\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"451aa366-1d78-4d8f-985f-bf8f05b48996\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"语义级别\"}]}]}]},{\"type\":\"tablerow\",\"content\":[{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"5423f067-58ae-44d4-ba8d-c43220025a3f\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"优化依据\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"bfd34eda-a73d-48b9-9127-d9bf6a7e619b\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"固定代价模型\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"2309f505-d958-4490-9161-53f3aa793dec\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"历史反馈+机器学习\"}]}]}]},{\"type\":\"tablerow\",\"content\":[{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"719d5712-a1e2-447a-8d2e-3c2285b671a6\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"查询方式\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"135948b5-50d2-4086-9607-b5afe24238c4\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"精确匹配\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"640af7ea-7500-41bb-bf4c-713e64fa3dce\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"支持近似查询、相似性查询\"}]}]}]},{\"type\":\"tablerow\",\"content\":[{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"73e4d64e-6ccd-493b-b700-521f9af3e98d\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"用户交互\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"ea573c83-8c67-4684-b4b1-b42cff17dd29\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"必须写sql\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"f57e1bc5-3bed-432b-a406-7c15a353c29c\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"支持自然语言\"}]}]}]},{\"type\":\"tablerow\",\"content\":[{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"55bd7865-df02-4183-ae1e-0f85c5af4635\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"典型技术\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"30119f45-a5ac-4ed4-8d3c-3b17376a1c90\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"基于规则的优化器\"}]}]},{\"type\":\"tablecell\",\"attrs\":{\"colspan\":1,\"rowspan\":1,\"colwidth\":null,\"background\":null,\"color\":null,\"textalign\":null,\"style\":\"padding: 6px 13px; border-color: rgb(223, 226, 229); word-break: normal; white-space: nowrap;\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"0c969b2c-12a4-4f96-bbf8-27ea8f617b75\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"ai代价模型、向量检索、nl2sql\"}]}]}]}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"228f1088-6d6c-4bc7-a7f4-f3fd637c1816\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"语义执行的核心技术\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"debe315d-cddc-434c-b160-e2a874c5884f\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"1. 智能化查询优化器\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"4666c54f-f720-4cc1-af85-65ab9cb96282\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"传统优化器基于固定的代价模型。新一代优化器引入机器学习，根据历史执行反馈动态调整模型。比如，某条sql在过去一周的执行计划都是a，今天突然有个新的统计信息，优化器会评估切换计划b的风险，而不是机械地选择代价最小的。这有点像推荐算法——根据历史行为预测最优路径。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"c47c46c6-e3a7-4cf5-9dfc-f0f075c36cc6\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"2. 近似查询与结果估算\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"d529e693-4648-494b-9a7d-655b9946d0de\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"有些查询不需要精确结果，只需要“大概”。比如“上个月的销售额大约多少”。传统数据库会老老实实扫全表，语义执行可以返回一个估算值（误差1%以内），耗时从分钟级降到秒级。这在bi分析和dashboard场景非常实用。postgresql的\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"tablesample\"},{\"type\":\"text\",\"text\":\"、金仓的近似聚合函数都提供了这类能力。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"ddb33fa0-3e37-4648-b5c1-9347fdb0310f\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"3. 自然语言查询（nl2sql）\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"a7943781-8124-4bc8-8494-62769c05972a\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"最典型的语义执行是让用户用自然语言提问，数据库自动生成sql。比如输入“查去年销量前十的商品”，系统理解“去年”=2025年，“销量前十”=按销量降序取前10，生成对应的sql。虽然目前准确率还有提升空间，但趋势已经很明显：数据库正在从“sql引擎”变成“语义引擎”。开源工具如vanna、chat2db可以集成到内部平台，让业务方自助取数。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"ceac4646-f3c3-4ac7-ab3b-1bc63219686b\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"4. 向量检索与相似性查询\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"0d193332-eff4-406a-b930-fa90fba9d81d\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"传统查询是精确匹配：\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"where name = '张三'\"},{\"type\":\"text\",\"text\":\"。语义执行支持相似性查询：\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"where embedding \u003c-\u003e '[向量]'\"},{\"type\":\"text\",\"text\":\"，找出最相似的记录。这在以图搜图、智能推荐、知识库问答场景中广泛应用。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"52daf395-5cb4-4ac7-a55d-b86be7f06db1\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"实际运用：dba能做什么？\"}]},{\"type\":\"bulletlist\",\"attrs\":{\"id\":\"9f4c76a3-1557-4f5f-b615-ff78addefa72\",\"ishoverdraghandle\":false},\"content\":[{\"type\":\"listitem\",\"attrs\":{\"id\":\"8b2b54c3-83c6-45e0-a98e-e47f915bda45\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"dde23348-f413-49cc-b4c1-038818b052c8\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"\"},{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"利用近似查询\"},{\"type\":\"text\",\"text\":\"：对于报表类的“大概数据”，主动建议业务方使用近似查询，而不是每次都精确计算。\"}]}]},{\"type\":\"listitem\",\"attrs\":{\"id\":\"5447974e-187c-4c4d-a300-e093891a26d3\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"5ad5f51e-2e87-4770-ab5c-32800db11a57\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"\"},{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"用好执行计划反馈\"},{\"type\":\"text\",\"text\":\"：mysql 8.0的\"},{\"type\":\"text\",\"marks\":[{\"type\":\"code\"}],\"text\":\"explain analyze\"},{\"type\":\"text\",\"text\":\"能输出实际执行信息，结合慢查询日志，可以给优化器“反馈”，让它下次选对计划。\"}]}]},{\"type\":\"listitem\",\"attrs\":{\"id\":\"c81872d9-bd2e-42f1-b805-6a7c15dfb109\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"37fff2d5-3472-4893-9cb3-ee63ec0220cf\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"\"},{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"关注nl2sql工具\"},{\"type\":\"text\",\"text\":\"：像vanna、chat2db等开源项目，可以集成到内部平台，让业务方自助取数，减少dba的临时查询负担。\"}]}]},{\"type\":\"listitem\",\"attrs\":{\"id\":\"2352ed41-c174-4f36-8ddc-e82e160ac8dd\"},\"content\":[{\"type\":\"paragraph\",\"attrs\":{\"id\":\"479b771e-0d01-41da-aa1f-353adcd2b829\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"\"},{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"理解向量检索原理\"},{\"type\":\"text\",\"text\":\"：当公司需要做ai应用（如智能客服、推荐系统）时，dba可以给出数据库层面的选型建议——是用专用向量数据库，还是用现有数据库的向量扩展。\"}]}]}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"8db945ee-13ca-4f04-9af2-3d8ef9b261b3\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"一点总结\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"ee55698b-ed44-460a-8e43-5a4e7a3a6336\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"语义执行是数据库智能化的重要方向。它不是说dba要被取代，而是让数据库帮我们做更多“理解”的工作。作为dba，了解这些趋势，可以更好地选择数据库产品、设计数据模型，甚至在团队中推动从“写sql”到“描述意图”的转变。\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"10fdbd6f-570b-4b5c-95cb-556090db4714\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"marks\":[{\"type\":\"bold\"}],\"text\":\"小耶在手，sql 不愁\"}]},{\"type\":\"paragraph\",\"attrs\":{\"id\":\"bd0763c7-b870-4ae7-91f7-14a9ce9f8cbd\",\"textalign\":\"inherit\",\"indent\":0,\"color\":null,\"background\":null,\"ishoverdraghandle\":false},\"content\":[{\"type\":\"text\",\"text\":\"还有什么想了解的，欢迎留言！小耶一定知无不言言无不尽……我们下次见~\"}]}]}
