![](https://img-blog.csdnimg.cn/20201014180756928.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
Text As Data
JYY_JYY_
这个作者很懒,什么都没留下…
展开
-
Uog Text as Data Lecture 7
Word Vectors & Contextual Word Embeddings1.Word Vectors andWord meaning用one-hot 向量表示句子是有歧异的,因为相同word有不同meaning一个解决方法是WordNet:a knowledge base containing lists of synonym sets and hypern...原创 2020-04-02 01:28:15 · 154 阅读 · 0 评论 -
UoG Text as Data Lecture 5
Text Classification分类应用:Topic categorization: Is this document about [technology / sports / lifestyle, etc..]Spam detection: Is this comment/email spam?Sentiment analysis: Does this tweet expr...原创 2020-03-28 01:15:28 · 195 阅读 · 0 评论 -
UoG Text as Data Lecture4
Language Models目的:Predict the probability of a sentence or sequence of words in a document. 预测某个单词在某个句子中或者某个句子在某个文档中出现的概率应用场景:定义定义:something that specifies the following two quantities, for...原创 2020-03-27 23:11:28 · 179 阅读 · 0 评论 -
UoG Text as Data Lecture3
Clustering在TaD中,我们会把相似度高的document聚为一类,可以用来给用户作推荐系统。之后的课程中,会把term聚类聚类算法是unsupervised的,即没有labelled training data,也没有正确答案。聚类过程:1.Derive a document representation – Typically, vectors of weighted...原创 2020-03-27 00:22:30 · 257 阅读 · 0 评论 -
UoG Text as Data Lecture2
1.Term Frequency & Bag-of-words与上一期的one-hot encoding不同,bag-of-words会记录每个term的frequencybag-of-words假设:If a term occurs lots in a document it should imply something about what that document is a...原创 2020-03-26 23:25:09 · 220 阅读 · 0 评论 -
UoG Text as Data Lecture1
1. Text Processing首先 token 的概念1)Segmenting/tokenzinge.g. “House of Tartan sells blankets” -------->“House”, “of”, “Tartan” “sells”, “blankets”2)Normalizee.g., “House”, “of”, “Tartan” “se...原创 2020-03-26 22:40:17 · 239 阅读 · 0 评论