R学习之——R用于文本挖掘（tm包）

首先需要安装并加载tm包。

---

1、读取文本

x = readLines("222.txt")

2、建立语料库

 > r=Corpus(VectorSource(x))

 > r

 A corpus with 7012 text documents

3、语料库输出，保存到硬盘

> writeCorpus(r)

4、查看语料库

> print(r)
A corpus with 7012 text documents
> summary(r)
A corpus with 7012 text documents

The metadata consists of 2 tag-value pairs and a data frame
Available tags are:
  create_date creator 
Available variables in the data frame are:
  MetaID 

  > inspect(r[2:2])
  A corpus with 1 text document

  The metadata consists of 2 tag-value pairs and a data frame
  Available tags are:
  create_date creator
  Available variables in the data frame are:
  MetaID

  [[1]]
  Female; Genital Neoplasms, Female/*therapy; Humans

  > r[[2]]
  Female; Genital Neoplasms, Female/*therapy; Humans

5、建立“文档-词”矩阵

> dtm = DocumentTermMatrix(r)
> head(dtm)
A document-term matrix (6 documents, 16381 terms)

Non-/sparse entries: 110/98176
Sparsity           : 100%
Maximal term length: 81 
Weighting          : term frequency (tf)

6、查看“文档-词”矩阵

> inspect(dtm[1:2,1:4])

7、查找出现200次以上的词

> findFreqTerms(dtm,200)
 [1] "acute"          "adjuvant"       "advanced"       "after"         
 [5] "and"            "breast"         "cancer"         "cancer:"       
 [9] "carcinoma"      "cell"           "chemotherapy"   "clinical"      
[13] "colorectal"     "factor"         "for"            "from"          
[17] "group"          "growth"         "iii"            "leukemia"      
[21] "lung"           "lymphoma"       "metastatic"     "non-small-cell"
[25] "oncology"       "patients"       "phase"          "plus"          
[29] "prostate"       "randomized"     "receptor"       "response"      
[33] "results"        "risk"           "study"          "survival"      
[37] "the"            "therapy"        "treatment"      "trial"         
[41] "tumor"          "with"          

7、移除出现次数较少的词

inspect(removeSparseTerms(dtm, 0.4))

8、查找和“stem”的相关系数在0.5以上的词

> findAssocs(dtm, "stem", 0.5)
 stem cells 
 1.00  0.61 

 9、计算文档相似度（用cosine计算距离）

> dist_dtm <- dissimilarity(dtm, method = 'cosine')
> head(dist_dtm)
[1] 1.0000000 0.7958759 0.8567770 0.9183503 0.9139337 0.9309934

10、聚类

> hc <- hclust(dist_dtm, method = 'ave')
> plot(hc,xlab='')