Global Fast-Food Sentiment DB: Deep Learning Analysis of Twitter Feedback

Description

The 'Global Fast-Food Sentiment DB: Deep Learning Analysis of Twitter Feedback' dataset comprises data collected from Twitter about three major fast-food restaurants - McDonald’s, Burger King, and KFC - which were listed among the top 25 most valuable restaurant brands by Brand Finance in 2022. The dataset includes English and Turkish tweets collected within a specific date range. Turkish tweets were translated into English to enhance classification accuracy. These brands were selected due to their high recognition both domestically and internationally. Online reviews about these brands were gathered from Twitter. Tweets written in English and Turkish between September 9, 2022, and October 9, 2022, were collected, resulting in 145,551 English and 3,392 Turkish tweets. The data collection was conducted using the Python programming language, with the Snscrape library utilized for scraping tweets. To augment the number of Turkish tweets, additional data from June 8, 2022, onward were collected, totaling 12,391 Turkish tweets. The dataset also includes the location, date, and time information of the tweets. The raw tweets underwent specific preprocessing steps. Polarity and subjectivity values were then derived using the TextBlob library. Based on polarity, the sentiment of the tweet content was classified as positive, negative, or neutral. The attributes of the created dataset include Text, Translated text (Only in the Turkish dataset), Cleaned text, Location, Date and time, Polarity value, Subjectivity value, Language, and Sentiment. In the dataset composed of English tweets, McDonald’s is the most tweeted about brand with a total of 77,388 tweets. In contrast, in the dataset consisting of Turkish tweets, Burger King is the most mentioned brand with 2,334 tweets. The numbers of tweets gathered for each brand are as follows: For the English dataset, McDonald’s: 77,388 tweets, KFC: 24,904 tweets, and Burger King: 15,309 tweets; for the Turkish dataset, Burger King: 2,334 tweets, KFC: 1,831 tweets, and McDonald’s: 1,119 tweets. These numbers reflect the frequency of discussions surrounding each brand on social media, indicating a higher volume of English tweets for McDonald’s and a predominant Turkish conversation about Burger King within the context of our study.

Steps to reproduce

English and Turkish tweets have been collected, along with their location information. Based on the collected Twitter data which includes English and Turkish tweets with location information. The United States (USA) appears as the top location for English Tweets. Other prominent locations include cities like Los Angeles and Chicago, as well as states such as California and Florida, indicating that the majority of the English tweets were posted from the USA. The most frequent location for Turkish tweets is Istanbul, Turkey. Other locations also include cities like Istanbul, Izmir, Ankara, showing that the majority of Turkish tweets were posted from Turkey. Consequently, the relationship between obesity and sustainability in Turkey and the USA will be examined. In the English dataset, an example entry features a cleaned-up text version, "falafel wrap good kfc good," derived from the original "falafel wrap not good - kfc that’s good." This entry is associated with London, England, and was recorded on September 13, 2022, at 18:10:27. The sentiment analysis of this text reveals a polarity value of 0.7 and a subjectivity value of 0.6, indicating a positive sentiment in English. Similarly, in the Turkish dataset, an entry's original text, "Beni burger king mutlu eder tabii ki," translates to "Burger king makes me happy of course," and is further simplified to "burger king makes happy" in the cleaned version. This entry comes from Ankara, Turkey, with a timestamp of October 1, 2022, at 15:57:57. The sentiment analysis categorizes it as positive, with a polarity value of 0.8 and a subjectivity value of 1, reflecting the sentiment in Turkish.

Institutions

Categories

Funding

Licence