如何在mahout中使用朴素贝叶斯分类器创建自己的情感分析模型？

Question

I am a beginner in mahout. i don't know how to create our own model for sentiment analysis using naive bayes classifier in mahout. I want to create my own model to do sentiment analysis on top of some log data. Is there a step by step procedure for doing this. Like what are the classes we have to implement and how to create model or how we can use existing models in mahout. Any help would be appreciated. Thanks in advance.

Answer 1

You can see in this presentation a good guide about the steps to follow for an analysis using Naive Bayes classifier in Mahou. It's kind an step by step procedure.

A Deep Dive into Classification with Naive Bayes. Along the way we take a look at some basics from Ian Witten's Data Mining book and dig into the algorithm....

Answer 2

So did you look at the quickstart ?

The first step is to asses your corpus. How is your log data labeled? How much data do you have? If you have a labeled corpus then just follow the quickstart and substitue your corpus for the ones in the example.

Before you even start to write code you have to have a high quality corpus. Ensure that your examples are balanced and that you have enough data to train on. You can take a look at some research corpora for a general idea on what is required to train. I would sugest the Reuters-21578 corpus or if you can get it the RCV-1 corpus.

Answer 3

您可以查看以下博客，在此详细说明了逐步过程： http : //instantjavasolutions.blogspot.in/2014/10/how-to-train-my-own-model-using-mahout.html