[IR] Probabilistic Model
If user has told us some relevant and some irrelevant documents, then we can proceed to build a probabilistic classifier, such as a Naive Bayes model.
Can we use probabilities to quantify our uncertainties?
Ranking method:
Rank by probability of relevance of the document w.r.t. information need.
P(relevant | document i, query)
Bayes’ Optimal Decision Rule: x is relevant(相关的)iff p(R|x) > p(NR|x)
C - cost of retrieval of relevant document
C’- cost of retrieval of non-relevant document
C ⋅ p(R | d) + C ′ ⋅ (1− p(R | d)) ≤ C ⋅ p(R | d′ ) + C ′ ⋅ (1− p(R | d′ ))
for all d’ not yet retrieved, then d is the next document to be retrieved
- How do we compute all those probabilisties?
- 二值独立模型 - Binary Independence Model
(q位置没有变,odds 优势率)
分母约去。
Query相关的话,文档Vecdor如此的概率是多少?需要估计。
思考:针对一个Query,某单词是否该出现在文档中呢?
假设 (重要):
pi = p ( xi = 1 | R , q );
ri = p ( xi = 1 | NR , q );
(去掉xi = 0后,乘的变多了,多了xi =1, qi =1的部分。在前一个连乘中乘以倒数,达到平衡。)
两个常量:
query能获得有效返回的概率。
every query 与vocabulary中的each word的相关的概率。
一个变量:
Retrieval Status Value
So, how do we compute ci ’s from our data ?
For each term i look at this table of document counts:
(Term与doc的关系:出现但不一定相关;相关但不一定出现,比如computer与IBM)
pi = s / (S-s)
ri = (n-s) / (N-n-S+s)
Add 1⁄2 Smoothing
结论:一篇新文档出现,遂统计every Term与该doc的关系,得到Ci。
- Okapi BM25: 一个非二值的模型 (略)
[IR] Probabilistic Model的更多相关文章
- Intro to Probabilistic Model
概率论复习 概率(Probability) 频率学派(Frequentist):由大量试验得到的期望频率(致命缺陷:有些事情无法大量试验,例如一封邮件是垃圾邮件的概率,雷达探测的物体是一枚导弹的概率) ...
- 本人AI知识体系导航 - AI menu
Relevant Readable Links Name Interesting topic Comment Edwin Chen 非参贝叶斯 徐亦达老板 Dirichlet Process 学习 ...
- [IR] Information Extraction
阶段性总结 Boolean retrieval 单词搜索 [Qword1 and Qword2] O(x+y) [Qword1 and Qword2]- 改进: Gallo ...
- PGM:概率图模型Graphical Model
http://blog.csdn.net/pipisorry/article/details/51461878 概率图模型Graphical Models简介 完全通过代数计算来对更加复杂的模型进行建 ...
- [IR] Word Embeddings
From: https://www.youtube.com/watch?v=pw187aaz49o Ref: http://blog.csdn.net/abcjennifer/article/deta ...
- 深度学习基础 Probabilistic Graphical Models | Statistical and Algorithmic Foundations of Deep Learning
目录 Probabilistic Graphical Models Statistical and Algorithmic Foundations of Deep Learning 01 An ove ...
- FAQ: Machine Learning: What and How
What: 就是将统计学算法作为理论,计算机作为工具,解决问题.statistic Algorithm. How: 如何成为菜鸟一枚? http://www.quora.com/How-can-a-b ...
- ### Paper about Event Detection
Paper about Event Detection. #@author: gr #@date: 2014-03-15 #@email: forgerui@gmail.com 看一些相关的论文. 1 ...
- [ML] I'm back for Machine Learning
Hi, Long time no see. Briefly, I plan to step into this new area, data analysis. In the past few yea ...
随机推荐
- 通过Ruby On Rails 框架来更好的理解MVC框架
通过Ruby On Rails 框架来更好的理解MVC框架 1.背景 因为我在学习软件工程课程的时候,对于 MVC 框架理解不太深入,只是在理论层面上掌握,但是不知道如何在开发中使用 MVC ...
- ARM Cortex Debug Port Access Port DP AP JTAG-DP SW-DP SWJ-DP JTAG-AP MEM-AP
- React中props.children和React.Children的区别
在React中,当涉及组件嵌套,在父组件中使用props.children把所有子组件显示出来.如下: function ParentComponent(props){ return ( <di ...
- 正在使用MJRefresh & MJExtension的App
框架地址:https://github.com/CoderMJLee已经有上百个App用到了MJRefresh & MJExtension框架(只列出了其中一部分App):
- APK重签名总结
keytool -genkey -alias aeo_android.keystore -keyalg RSA -validity 20000 -keystore aeo_android.keysto ...
- 40页PPT勾画“互联网颠覆性思维”----诠释互联网思维
本文PPT内容涉及移动互联网的三个分支——移动电商.在线教育和新媒体. 不同领域一直是可以相互借鉴.相互渗透.相互学习的,在盈利模式和思维方式上有很多是共通的.
- Road to the future——伪MVVM库Q.js
模仿Vuejs的伪MVVM库,下面是使用说明 项目地址:https://github.com/miniflycn/Q.js 相关项目:https://github.com/miniflycn/Ques ...
- Conway's Game of Life: An Exercise in WPF, MVVM and C#
This blog post was written for the Lockheed Martin Insight blog, sharing here for the external audie ...
- Android 模拟器检测
参考链接:https://github.com/MindMac/HideAndroidEmulator 从多个方面识别模拟器1.用户习惯:联系人数量.短信数量.相册里面照片数量.安装的应用2.从IME ...
- C#排序比较
与C#定义了相等性比较规范一样,C#也定义了排序比较规范,以确定一个对象与另一个对象的先后顺序.排序规范如下 IComparable接口(包括IComparable接口和IComparable< ...