简体繁体 English

本地语言的情感分析（尼泊尔语）

[英]Sentiment Analysis for local languages (Nepali)

原文 2016-07-05 07:28:47 6 1 localization/ nlp/ sentiment-analysis/ naivebayes/ senti-wordnet

I would like to do sentiment analysis on document level. 我想在文档级别上进行情感分析。 But I am try to do sentiment analysis Nepali. 但是我试图做尼泊尔的情绪分析。 So, I dont have any resources. 因此，我没有任何资源。 I can't do Naive Bayes Classifier as I don't have any labelled data and I can't do vai wordnet as no nepali wordnet exist. 我无法执行朴素贝叶斯分类器，因为我没有任何标签数据，也不能执行vai词网，因为不存在尼泊尔词网。 Papers I read generally had labelled data or senti-wordnet for other languages. 我阅读的论文通常带有其他语言的标签数据或标记字网。

I would like know these things: 我想知道这些事情：

Which approach should I use in above case for sentiment analysis? 在上述情况下，应该使用哪种方法进行情感分析？
Is there any method for me to dynamically generate labels for data? 我有什么方法可以动态生成数据标签？

1 个解决方案

Since you don't have any labelled data, Have a look at this GitHub Repo, feel free to fork. 由于您没有任何标签数据，因此请查看此 GitHub Repo，请随意进行分叉。

It has the code for neural network for Handwriting recognition in Java. 它具有用于Java中手写识别的神经网络代码。 Jeff Heaton has done it easy for us, with a nice UI, you can train this model to recognize Nepali. 杰夫·希顿（Jeff Heaton）通过良好的用户界面为我们简化了工作，您可以训练该模型来识别尼泊尔语。

And for sentiment Analysis, you can try using Opennlp which has some good support, this blog for Beginner's . 对于情绪分析，您可以尝试使用Opennlp ，该博客提供了一些很好的支持，该博客适用于Beginner's 。

Also DL4J is a good library for deep learing for Java which can be used for Sentiment Analysis. DL4J还是一个很好的库，可用于Java的深度学习，可用于情感分析。 It has a good Word2Vector Implementation and has a lot of support. 它具有良好的Word2Vector实现并具有很多支持。

These resources will help you, any futher doubts-feel free to comment. 这些资源将帮助您，如有任何疑问，请随时发表评论。