1. Introduction to NLP
Chilly_Rain
posted @ 2014年3月08日 16:13
in Coursera_NLP
with tags
NLP
, 1230 阅读
之前在Coursera上找到了一门NLP的课，也跟过一段时间，但是没跟完。。。本来想再开课时再跟一次，但是从网站上看貌似是短期内不会重开了，好在教学视频都在，只好自学一下了，顺便把笔记记在这里。
第一周的课分成了三个部分，这里是第一部分，对NLP的简单介绍，由于不涉及具体的NLP算法，所以我也只是简单带过了。
什么是自然语言处理
- NLU：自然语言理解，把languge作为computer的输入，让后者理解输入的是啥；
- NLG：自然语言生成，把language作为computer的输出，即让后者生成自然语言；个人感觉这应该是在NLU的基础上才能做的。
具体应用的举例
- 机器翻译：理解源语言，生成目标语言，这个应该是NLU＋NLG。
- 信息抽取：从一段非结构化的自然语言的文本中抽取出关键信息，如这段话描述的时间，位置，事件等结构化信息（这些结构后信息通常存于db中），换句话说就是想把一段文本对应到一条数据库记录上。应该场景可以是复杂的search query或者statistical query。
- 摘要：同样也是NLU＋NLG。需要先理解这段文本的意思，然后进行压缩，输出摘要内容。
- 对话系统：接受自然语言的问题，理解之，再做出回答，也是NLU＋NLG。
基础的NLP问题：这些问题是解决上面具体应用的基石
- Tagging：标注问题，把一段文本做为输入，为每个单位打上一个标记，输出对应的标记序列。包括POS Tagging（词性标注），Named Entity Recognition（命名实体识别）。
- Parseing：解析问题，以一句话做为输入，辨识出它的语法结构，输出是一棵语法树。这两个问题后面都有章节具体地讲。
为什么NLP是个难题
一句话可能有各种层次上的歧义，包括acoustic level（like your vs lie cured），syntactic level（不同的语法树，从而造成不同的意义），semantic level（一词多义）, discourse level（指代不明）。
课程包含的内容
- NLP问题：POS tagging，parsing，词义消歧
- 机器学习技术：PCFG，HMM，EM，log-linear model，estimation/smoothing techniques
- 应用：信息抽取，机器翻译等
2022年8月07日 19:01
New India Assurance corporation limited does offer insurance plans for vehicles which include commercial vehicles and private vehicles. The insurance policy does protect them from third-party liabilities and unwanted expenses which might arise through accidental damage or due to any kind of loss to the vehicle. New India Assurance Renewal Online Payment Having proper insurance will cover your entire vehicle damage as per their insured value, as well having regular insurance will help you have good value as it gets changed every year with a calculated percentage.
2022年9月23日 23:47
The students who study in Central board schools can download NCERT 5th Class Computer Science Sample Paper 2023 with answers to get the new exam scheme for all question patterns conducted under Term-1 & Term-2 such as SA1, SA2, FA1, FA2, FA3, FA4 and Assignments. By downloading and practising, NCERT Computer Sample Paper Class 5 these model papers help to get an analysis of the question paper pattern, and important questions.The students who study in Central board schools can download NCERT 5th Class Computer Science Sample Paper 2023 with answers to get the new exam scheme for all question patterns conducted under Term-1 & Term-2 such as SA1, SA2, FA1, FA2, FA3, FA4 and Assignments. By downloading and practising.
2023年2月06日 15:43
The best facility for every prepaid mobile customer is online recharge to feel proud for managing recharges from own hands with new offers provided by Bharat Sanchar Nigam Limited. This may possible through a new website call as BSNL Quick Recharge Portal or with My BSNL App. BSNL Quick Recharge This is one of the best service to avail Full Talktime Top Up, STV’s or any latest prepaid mobile plan recharges online. It is like an instant service, but from anywhere on just log in to BSNL Recharge Portal.
2023年7月01日 18:56
Dedicated webmasters and bloggers founded Pavzi Post with the goal of creating engaging information that is truthful, fascinating, and worth reading. We resemble a virtual community where you may find a variety of resources, information, and themes about everyday events or the news. With the assistance pavzipost.com of the editorial and content teams, we strive to give you the best web information on each and every topic.
2023年8月11日 19:36
SCERT Assam Board Follows NCERT Curriculum These Textbooks are Updated as per the Syllabus Prescribed by Assam Board. Students of 3rd Class Should follow Prescribed Textbooks while Preparing for Exam. Our Team Refer to the Respective Subject Textbook while Preparing the Final Important questions. Students Best Practice Study Materiel about Textbooks are the Fact that they are so Assam 3rd Class Textbook 2024 Comprehensible that it does not require the aid of a Subject Literate.SCERT Assam has Developed the new Textbooks at the Elementary Standard in Prepare by Senior experts. Assam Board once Publishes the Assam 3rd Standard Textbooks 2024