Sentiment Analysis of Animal Crossing Reviews
After the craze about Stardew Valley, I thought I would never go back to any game that requires farming/hoarding/fishing/mining. But here I am: spending a total of 100 hours on Animal Crossing: New Horizons (ACNH) since I bought it on March 22, and this is more than hours than BotW. Genuinely, ACNH is not a game I thought I would enjoy: nothing too exciting, just a bunch of grinding. But it is cute, and it just came at the perfect time for me (and many people) to connect with friends while social-distancing. And there is this article on Polygon about how the grinding element in this kind of game can provide a certain level of relaxation and comfort. So, it is not surprising to see the sales of ACNH just sky-rocketed since its release on March 20, so as the number of google searches about ACNH.
This graph also shows that the rise in interest in ACNH also increases the interest in Nintendo Switch. And, Nintendo Switch is almost sold out everywhere right now, including in Japan. News has reported the ACNH has broken the sales record of the Nintendo Switch title in Japan. The price of ACNH’s physical copy has almost doubled on Amazon too. My bold inference is that there is a positive correlation between the increase of Nintendo Switch sales and the increase of ACNH sales.
It almost seems like ACNH is very well-received by the players. However, when I looked up the reviews on some gaming websites, such as GameSpot and Metacritic, the average user scores are 6.4 and 5.2/10, respectively, despite the website’s official rating of 9/10. So, why do players rate this game worse than the official reviewers? I am surprised by the user scores (I would give a 7/10, better than Pokémon Sword & Shield), and I am curious about what do the users talk about ACNH.
So, I did a quick sentiment analysis (since I just learned it from Prof. Chris Maurer’s text analytics class this week). I first scrapped the review data from Metacritic using BeautifulSoup on Python. Then I formatted the CSV file a bit before diving into sentiment analysis using R. I compared several lexicons from the lexicon package and decided that Hu-Liu, Jockers-Rinker, and NRC lexicons are best suited for my use in analyzing Metacritic reviews due to their consistencies in sentiment scores.
Average Sentiment Scores of User Reviews
The average sentiment scores generated by three lexicons have shown that user reviews generally are less positive (0 being neutral, 1 being positive). Still, there is an increasingly more positive trend in the recent reviews. The less positive sentiment score in the reviews are not surprising as out of 3,450 user ratings, almost 50% of them are negative. The highly positive reviews can be canceled out by the highly negative reviews, resulting in a relatively neutral score.
Positive and Negative Words in User Reviews
So, what are these Metacritic users talking about in their reviews? I compiled the lists of negative and positive words generated by Hu-Liu and Jockers-Rinker Lexicon (not shown) and made word clouds according to their frequency.
I guess people do not like the idea of capitalism in this game and describe Tom Nook as a dictator, and we (the players) as his slaves (to pay our housing loan oops). ‘limits” is the most frequently appeared negative words. I inferred this frequent occurrence due to users’ complaints about the limitation of one island per console (maybe?). ‘disappointment’ appears, showing the high expectation from these users due to the hype.
This positive word cloud has “fun” as the most frequently appeared positive words. Words like “enjoy,” “enjoyed,” and “enjoys” often appeared enough to infer that some users do enjoy the game.
These two-word clouds just show that those who like the games praise the game for its fun and creativity elements, thinking that this game is amazing (etc.). They enjoy the game enough to give positive reviews and high ratings. Those who do not like the games to the extent they gave really low ratings (read “0”), and use harsher words like “dictator,” “slaves,” “disappointment” in their reviews. These just show that ACNH is a game that receives polarized opinions.
Emotions in User Reviews
I also looked at the emotions associated with these reviews using the NRC lexicon. Although users used trust-related words most frequently, words like “system,” “account,” and “series” do not carry real insights that describe users’ attitudes towards the game. Anticipation words like “long,” “wait,” and “start” also do not carry much meaning here. Words that mean joy seem to appear more frequently too. Sadness, disgust, anger, and fear are the negative emotions detected in this collection of reviews. They are most likely associated with those reviews with low ratings.
I did this analysis out of curiosity. It seems like the Metacritic users have polarized opinions about ACNH, as shown in the neutral sentiment score, positive and negative word clouds, and the rating percentage on the website. But this sentimental analysis does not tell the whole story of ACNH’s reviews. I am aware of the limitations of this analysis, especially the one about emotion. Metacritic users also do not represent the ACNH players well, as it is just one of the few game review websites. YouTube could be another possible platform where players talk about the game and share their experiences.
