Yelp Reviewers vs Professional Food Critics

produced using: R (packages – wordcloud, stringr, tm, snowballC, lsa)

Description

This project compares Yelp reviews to reviews written by professional food critics. The food critics were writers from the NYTimes, NY Magazine, and the Village Voice. There were in total 9 Michelin star restaurants used in this analysis (I chose Michelin star restaurants to control for quality). There were three 1-Michelin star restaurants, three 2-Michelin star restaurants, and three 3-Michelin star restaurants. In total, there were almost 27 reviews written by professional critics (the number is a little short of 27 because Village Voice and NY Magazine had a few missing reviews).

Since there were around 3 professional reviews per restaurant (one for each news outlet), I collected 3 Yelp reviews for each Michelin star restaurant as well. The reviews were taken from Yelpers who achieved “Elite” status, meaning that they are very active members on the website. I also made sure to collect only the three most recent reviews.

Wordclouds

There are two wordclouds shown below: the first one highlights the words that are different between the critics and Yelpers while the second one displays the most common words between the two groups.

(Comparison Cloud)

(Similarity Cloud)

In general it appears that Yelpers are more concerned with the actual meal and that critics are more interested in contextualizing the restaurant. As seen in the first wordcloud, the critics mention the name of the restaurant and the chefs in their reviews (“aquavit” and “eleven madison park” are restaurant names while “guidara” and “vongerichten” are chef’s names) while the Yelp reviews do not mention names at all. This can be interpreted as an attempt by the critics of creating a narrative for the restaurant in their reviews while Yelpers are more concerned with their personal dining experience by focusing on the “bread”, “flavor”, and “service.”

Relative Term Frequencies

Here are also the relative term frequencies for all the reviews. As seen below, words that critics tend to use more are “restaurant”, “sauce”, “french”, “italian”, and “room.” Words that Yelpers tend to use more are “course”, “food”, “meal”, “dessert”, and “chocolate”. It seems that professional critics are more likely to write about the space of the restaurant since they mention the “room” while also writing about the type of cuisine served since they mention “french” and “italian.” Yelpers on the other hand describe the taste of the food (using words like “good” and “delicious”) while also mentioning specific aspects of the meal by writing about “dessert” and “chocolate.”

Sentiment Analysis

Lastly, I conducted a sentiment analysis for all the reviews and then graphed the results using a bar chart. Both positive and negative words were determined using dictionaries that ascribed a variety of words in the English language with a “positive” or “negative” value. The sentiment score was calculated by subtracting the positive word count from the negative count and then dividing the results by the sum of the two counts.

As seen below, the review outlets tend to be positive in the aggregate since none of them have a negative score (this could be attributed to the fact that all restaurants in this sample have at least 1 Michelin star). Interestingly, both Village Voice and Yelp have the most positive reviews compared to the NYTimes and NY Magazine. Interestingly, the Yelp reviews have more positive sentiment than the combined articles for all the professional critics. It’s possible that this effect could be explained by differences in exposure to quality food between Yelp reviewers and critics since Professional food critics are more likely to eat at Michelin star restaurants on a more frequent basis. Since the experience is more normalized for the critics, it might explain why they do not use as much positive sentiment in their reviews when compared to the Yelpers