java - 使用stanford core nlp java代码获取情绪分析结果

Question

当我们在斯坦福演示页面上对其进行测试时:http://nlp.stanford.edu:8080/sentiment/rntnDemo.html

它给树每个节点的情感分数如下:

我正在尝试使用命令在我的本地系统上测试它:

H:\Drive E\Stanford\stanfor-corenlp-full-2013~>java -cp "*" edu.stanford.nlp.sen
timent.Evaluate edu/stanford/nlp/models/sentiment/sentiment.ser.gz test.txt

text.txt 有

这部电影不关心机智、机智或任何其他类型的智慧幽默。
那些在美丽的事物中发现丑陋意义的人是腐败而不迷人的。


产生结果:

谁能告诉我为什么它是空的？或者也许我在执行时犯了任何错误？我的目的是分析文本并获得带有分数的情感结果。

Answer 1

您使用的文件有误，命令也不完整。以下是您应该使用的命令。

java -cp "*"edu.stanford.nlp.sentiment.Evaluate -model edu/stanford/nlp/models/sentiment/sentiment.ser.gz -treebank 测试.txt

text.txt 文件不包含普通句子，而是包含树库

例如

(2 (3 (3 Effective) (2 but)) (1 (1 too-tepid) (2 biopic)))
(3 (3 (2 If) (3 (2 you) (3 (2 sometimes) (2 (2 like) (3 (2 to) (3 (3 (2 go) (2 (2 to) (2 (2 the) (2 movies)))) (3 (2 to) (3 (2 have) (4 fun))))))))) (2 (2 ,) (2 (2 Wasabi) (3 (3 (2 is) (2 (2 a) (2 (3 good) (2 (2 place) (2 (2 to) (2 start)))))) (2 .)))))
(4 (4 (4 (3 (2 Emerges) (3 (2 as) (3 (2 something) (3 rare)))) (2 ,)) (4 (2 (2 an) (2 (2 issue) (2 movie))) (3 (2 that) (3 (3 (2 's) (4 (3 (3 (2 so) (4 honest)) (2 and)) (3 (2 keenly) (2 observed)))) (2 (2 that) (2 (2 it) (2 (1 (2 does) (2 n't)) (2 (2 feel) (2 (2 like) (2 one)))))))))) (2 .))
(2 (2 (2 The) (2 film)) (3 (3 (3 (3 provides) (2 (2 some) (3 (4 great) (2 insight)))) (3 (2 into) (3 (2 (2 the) (2 (2 neurotic) (2 mindset))) (3 (2 of) (2 (2 (2 (2 (2 all) (2 comics)) (2 --)) (2 even)) (3 (2 those) (4 (2 who) (4 (2 have) (4 (2 reached) (4 (4 (2 the) (3 (2 absolute) (2 top))) (2 (2 of) (2 (2 the) (2 game))))))))))))) (2 .)))

收到的输出是

EVALUATION SUMMARY
Tested 82600 labels
  66258 correct
  16342 incorrect
  0.802155 accuracy
Tested 2210 roots
  976 correct
  1234 incorrect
  0.441629 accuracy
Label confusion matrix: rows are gold label, columns predicted label
       323      1294       292        99         0
       161      5498      2993       602         1
        27      2245     51972      2283        21
         3       652      2868      7247       228
         3       148       282      2140      1218
Root label confusion matrix: rows are gold label, columns predicted label
        44       193        23        19         0
        39       451        62        81         0
         9       190        82       101         7
         0       131        30       299        50
         0        36         8       255       100
Approximate Negative label accuracy: 0.912008
Approximate Positive label accuracy: 0.930750
Combined approximate label accuracy: 0.923128
Approximate Negative root label accuracy: 0.879081
Approximate Positive root label accuracy: 0.808266
Combined approximate root label accuracy: 0.842756

希望这有帮助:) !!