|
|
--- |
|
|
task_categories: |
|
|
- text-classification |
|
|
size_categories: |
|
|
- 10M<n<100M |
|
|
--- |
|
|
# 说明 |
|
|
|
|
|
数据集来源于[AI Challenger 2018](https://github.com/AIChallenger/AI_Challenger_2018) |
|
|
|
|
|
sentiment_analysis_trainingset.csv 为训练集数据文件,共105000条评论数据 |
|
|
sentiment_analysis_validationset.csv 为验证集数据文件,共15000条评论数据 |
|
|
sentiment_analysis_testa.csv 为测试集A数据文件,共15000条评论数据 |
|
|
|
|
|
数据集分为训练、验证、测试A与测试B四部分。数据集中的评价对象按照粒度不同划分为两个层次,层次一为粗粒度的评价对象,例如评论文本中涉及的服务、位置等要素;层次二为细粒度的情感对象,例如“服务”属性中的“服务人员态度”、“排队等候时间”等细粒度要素。评价对象的具体划分如下表所示。 |
|
|
|
|
|
The dataset is divided into four parts: training, validation, test A and test B. This dataset builds a two-layer labeling system according to the evaluation granularity: the first layer is the coarse-grained evaluation object, such as “service” and “location”; the second layer is the fine-grained emotion object, such as “waiter’s attitude” and “wait time” in “service” category. The specific description is shown in the following table. |
|
|
|
|
|
|层次一(The first layer)|层次二(The second layer)| |
|
|
|---|---| |
|
|
|位置(location)|交通是否便利(traffic convenience)| |
|
|
|-|距离商圈远近(distance from business district)| |
|
|
|-|是否容易寻找(easy to find)| |
|
|
|服务(service)|排队等候时间(wait time)| |
|
|
|-|服务人员态度(waiter’s attitude)| |
|
|
|-|是否容易停车(parking convenience)| |
|
|
|-|点菜/上菜速度(serving speed)| |
|
|
|价格(price)|价格水平(price level)| |
|
|
|-|性价比(cost-effective)| |
|
|
|-|折扣力度(discount)| |
|
|
|环境(environment)|装修情况(decoration)| |
|
|
|-|嘈杂情况(noise)| |
|
|
|-|就餐空间(space)| |
|
|
|-|卫生情况(cleaness)| |
|
|
|菜品(dish)|分量(portion)| |
|
|
|-|口感(taste)| |
|
|
|-|外观(look)| |
|
|
|-|推荐程度(recommendation)| |
|
|
|其他(others)|本次消费感受(overall experience)| |
|
|
|-|再次消费的意愿(willing to consume again)| |
|
|
|
|
|
每个细粒度要素的情感倾向有四种状态:正向、中性、负向、未提及。使用[1,0,-1,-2]四个值对情感倾向进行描述,情感倾向值及其含义对照表如下所示: |
|
|
|
|
|
There are four sentimental types for every fine-grained element: Positive, Neutral, Negative and Not mentioned, which are labelled as 1, 0, -1 and-2. The meaning of these four labels are listed below. |
|
|
|
|
|
|情感倾向值(Sentimental labels)|含义(Meaning)| |
|
|
|---|---| |
|
|
|1|正面情感(Positive) |
|
|
|0|中性情感(Neutral) |
|
|
|-1|负面情感(Negative) |
|
|
|-2|情感倾向未提及(Not mentioned) |
|
|
|
|
|
数据标注示例如下: |
|
|
An example of one labelled review: |
|
|
>味道不错的面馆,性价比也相当之高,分量很足~女生吃小份,胃口小的,可能吃不完呢,。环境在面馆来说算是好的,至少看上去堂子很亮,也比较干净,一般苍蝇馆子还是比不上这个卫生状况的。中午饭点的时候,人很多,人行道上也是要坐满的,隔壁的冒菜馆子,据说是一家,有时候也会开放出来坐吃面的人。 |
|
|
|
|
|
|层次一(The first layer)|层次二(The second layer)|标注 (Label)| |
|
|
|---|---|---| |
|
|
|位置(location)|交通是否便利(traffic convenience)|-2 |
|
|
|-|距离商圈远近(distance from business district)|-2 |
|
|
|-|是否容易寻找(easy to find)|-2 |
|
|
|服务(service)|排队等候时间(wait time)|-2 |
|
|
|-|服务人员态度(waiter’s attitude)|-2 |
|
|
|-|是否容易停车(parking convenience)|-2 |
|
|
|-|点菜/上菜速度(serving speed)|-2 |
|
|
|价格(price)|价格水平(price level)|-2 |
|
|
|-|性价比(cost-effective)|1 |
|
|
|-|折扣力度(discount)|-2 |
|
|
|环境(environment)|装修情况(decoration)|1 |
|
|
|-|嘈杂情况(noise)|-2 |
|
|
|-|就餐空间(space)|-2 |
|
|
|-|卫生情况(cleaness)|1 |
|
|
|菜品(dish)|分量(portion)|1 |
|
|
|-|口感(taste)|1 |
|
|
|-|外观(look)|-2 |
|
|
|-|推荐程度(recommendation)|-2 |
|
|
|其他(others)|本次消费感受(overall experience)|1 |
|
|
|-|再次消费的意愿(willing to consume again)|-2 |