机器学习算法
我正在研究一种适合我的目的的算法，有人可以为以下场景建议一个好的学习算法：
用户可以在一组句子中搜索某个单词。然后我将根据该关键字返回前 10 个句子，我希望算法允许用户输入，即用户可以单击最佳句子，这些信息将帮助搜索算法将来返回更合适的结果。
I am looking to research an appropriate algorithm for my purpose, can someone suggest a good learning algorithm for the following scenario:
A user can search for some word in a set of sentences. I will then return the top 10 sentences based on that keyword, I want the algorithm to allow user input, that is a user can click on the best sentences and this information will help the search algorithm to return more appropriate results in the future.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（3）
似乎您想使用用户反馈来改进某种搜索结果。如果这是正确的，您应该看看 Rocchio。
您还可以为每个关键字维护一个“点击”句子的列表。这样您就可以提升“点击”的排名。
Seems like you want to use user feedback to improve some kind of search results. If that is right you should take a look at Rocchio.
You could, also, maintain a list of "clicked" sententes for each keyword. That way you can boost the "clicked" ones positions in the rank.
您可能会发现 Qi II 手册中的这一章很有用：
6.3 属性列表
本章讨论使用语义网络来存储和排序数据。我还推荐本章后面的练习；无论你用什么语言写作，他们都可能会给你一些想法。
You might find this chapter in the Qi II manual useful:
6.3 Property Lists
The chapter discusses the use of semantic nets to store and sort data. I also recommend the exercises at the back of the chapter; they may give you some ideas, no matter what language you're writing in.
研究某种矩阵分解技术，例如奇异值分解或非负矩阵分解。
Look into some sort of matrix factorization technique, like singular value decomposition or non-negative matrix factorization.