常规任务
从SemEval-2014 Task 4 发布的对ABSA任务的评测来看,大致细分了4个子任务
- Aspect term extraction:提取句子中的情感实体或者说情感对象
- Aspect term polarity:在第一个任务基础上,对情感实体进行极性分类(正面、负面、中性等)
举个例子:
I hated their fajitas, but their salads were great → {fajitas: negative, salads: positive}
fajitas, salads是情感实体,后面对应的就是情感极性
- Aspect category detection:预先会定义一组类别的集合如{food, service, price, ambience, anecdotes/miscellaneous},然后判别这个句子中包含了哪些类别
- Aspect category polarity:该任务是在上一个任务的基础上,已经预先确定了句子的 aspect 类别,需要做的工作就是确定 aspect 类别对应的情感极性
举个例子:
The restaurant was too expensive → {price: negative}
The restaurant was expensive, but the menu was great → {price: negative, food: positive}
待续。。。。