TASC: topic-adaptive sentiment classification

This is a parallel implementation of multiple-label classification based on Spark MLlib.

Overview

内部循环选择添加非标注数据集
1.将训练数据split成文本特征和非文本特征训练数据
2.全局选择非标注数据集，已经被选择过的数据不再被选择
3.将每轮选择的非标注测试数据集添加到标训练数据集中
 
外部循环选择添加话题相关情感词
1.根据当前时刻选择的所有非标注数据集，选择话题相关情感词
2.然后根据配置文件中的map得到三元组：选择的非标注测试数据集，话题相关情感词 

build environment

1. As the former writer of this project changed some of the code of spark-mlib_2.10, spark-mllib_2.10-1.1.2-SNAPSHOT.jar should be replaced by the jar we upload here instead of the jar which is generated by the pom.
2. We recommend the spark-1.3.1-bin-hadoop2.6 to be your running environment.

sample

We upload a train sample of topic TeamFollowBack with its related files in folder sample. Please read the sample/SAMPLE_README.md for details.

Related works

Please cite the following references in your related work.

[1] Shenghua Liu, Fuxin Li, Fangtao Li, Xueqi Cheng, and Huawei Shen, “Adaptive co-training svm for sentiment classification on tweets” in Proc. of the 22nd ACM International Conference on Information and Knowledge Management, (CIKM ’13). New York, NY, USA, 2013, pp. 2079–2088.

[2] Shenghua Liu, Fuxin Li, and Fangtao Li, Xueqi Cheng, “TASC: Topic-Adaptive Sentiment Classification on Dynamic Tweets” IEEE Transactions on Knowledge and Data Engineering (TKDE), Vol. 27, issue 6, 2015, pp. 1696-1709.