I wanted to see what the customers of my local IKEA store think about, so I copied some of its reivews from google maps. Google restricts how many reviews can be loaded, so the data set is rather small and not representative for all reviews! Content: text: the review text or empty if there was no text in the review empty: whether the the review is empty or not rating: the star rating of the review days: Google's display of the reviews age date: approximate conversion of days into a date The days and date variable is unfourtunately pretty useless, since google doesnt display the actual publishing date. 4 star rating reviews are heavily underrepresented because they are in the "middle": google only let's you load about 1,300 reviews, so I got the 1,300 best and worst reviews. The almost 4,000 reviews in the middle are missing..... Possible questions: - classifying the sentiment/rating - classifying reviews into different categories - what do customers think and want?