# Natural language understanding   through comments — Data

来源：https://www.kaggle.com/competitions/education-level-prediction-trough-comments/data

采集时间：2026-10-19T20:39:40Z

This dataset contains a list of comments delivered to content creators, it contains a list of feature extractions trough NLU that provide context to the particular text string, weather it is referencing particular individuals, countries, or organizations. 
Finally it gives a sentiment score of the particular string giving the context. Your goal is to as accurately as possible determine whether the comment was positive 1,  neutral 0 or negative -1. 



What files do I need? 
the training dataset will provide you with a base features and target of the sentiment evaluated by our algorithm,
the comment that was analyzed and some features of the context in the text, your goal is to tokenize, classify  and vectorize the words in the comments so they can be matched with positive neutral or negative emotions.
 
What should I expect the data format to be?  
straight foreword csv file. 


What am I predicting?  
the overall sentiment of comments summited to content creators, you must determine in a spectrum whether the sentiment behind the text is positive, negative or neutral.


What acronyms will I encounter?
video_id: YT video from witch the comment was extracted
comments: the comment you need to work with to determine the final sentiment analysis.
FACILITY,  entity recognition of a facility.
LOCATION, entity recognition of a person.
 GPE, entity recognition of a geopolitical event.
ORGANIZATION, entity recognition of a organization.
PERSON, entity recognition of a person or individual.
sentiment, determines the score of the analysis of the text.