FunRec 概述
序言
这是一篇 datawhale 的相当优秀的推荐系统教程,因此特别废了九牛二虎之力把 FunRec 的半套内容,较为完整的移植到了本 wiki 中。
为什么要专门移植这篇?
zzm 个人以为推荐系统是一个非常有趣的横向和纵向都有很多应用的领域(放到外面是因为放到某一个模块下会因为次级链接太多把 wiki 撑爆了)
若你想尝试一个新领域,也许这是一个不错的切入点。更何况,如果你想足够完整的构建一个有实际价值的推荐系统,可能需要你去了解相当全面的知识。
在学习了基础内容之后,如果你想向着科研领域进发,也许对你而言最好的方式或许是选择一个大佬然后去 follow 他的进度。
如果你想去找相关的工作,你可以自行去深入学习有关本教程内容的实践部分,甚至是阅读算法面经。
同时只放上半部的原因是毕竟本偏内容是人工智能大类下的内容,后续可能会涉及一些前后端以及一些更为深入的东西,如果你只是想大致了解一下,那么阅读放在本片的内容被也许是一个不错的选择。
再次感谢 Datawhale 的大伙做出了如此卓著的贡献
正文
本教程主要是针对具有机器学习基础并想找推荐算法岗位的同学。教程内容由推荐系统概述、推荐算法基础、推荐系统实战和推荐系统面经四个部分组成。本教程对于入门推荐算法的同学来说,可以从推荐算法的基础到实战再到面试,形成一个闭环。每个部分的详细内容如下:
- 推荐系统概述。 这部分内容会从推荐系统的意义及应用,到架构及相关的技术栈做一个概述性的总结,目的是为了让初学者更加了解推荐系统。
- 推荐系统算法基础。 这部分会介绍推荐系统中对于算法工程师来说基础并且重要的相关算法,如经典的召回、排序算法。随着项目的迭代,后续还会不断的总结其他的关键算法和技术,如重排、冷启动等。
- 推荐系统实战。 这部分内容包含推荐系统竞赛实战和新闻推荐系统的实践。其中推荐系统竞赛实战是结合阿里天池上的新闻推荐入门赛做的相关内容。新闻推荐系统实践是实现一个具有前后端交互及整个推荐链路的项目,该项目是一个新闻推荐系统的 demo 没有实际的商业化价值。
- 推荐系统算法面经。 这里会将推荐算法工程师面试过程中常考的一些基础知识、热门技术等面经进行整理,方便同学在有了一定推荐算法基础之后去面试,因为对于初学者来说只有在公司实习学到的东西才是最有价值的。
特别说明:项目内容是由一群热爱分享的同学一起花时间整理而成,大家的水平都非常有限,内容难免存在一些错误和问题,如果学习者发现问题,也欢迎及时反馈,避免让后学者踩坑! 如果对该项目有改进或者优化的建议,还希望通过下面的二维码找到项目负责人或者在交流社区中提出,我们会参考大家的意见进一步对该项目进行修改和调整!如果想对该项目做一些贡献,也可以通过上述同样的方法找到我们!
为了方便学习和交流,我们建立了 FunRec 学习社区(微信群 + 知识星球),微信群方便大家平时日常交流和讨论,知识星球方便沉淀内容。由于我们的内容面向的人群主要是学生,所以知识星球永久免费,感兴趣的可以加入星球讨论(加入星球的同学先看置定的必读帖)!FunRec 学习社区内部会不定期分享 (FunRec 社区中爱分享的同学) 技术总结、个人管理等内容,跟技术相关的分享内容都放在了 B 站上面。由于微信群的二维码只有 7 天内有效,所以直接加下面这个微信,备注:Fun-Rec,会被拉到 Fun-Rec 交流群,如果觉得微信群比较吵建议直接加知识星球!。