NEHA October 2022 Journal of Environmental Health

18 Volume 85 • Number 3 A D VANC EME N T O F T H E SCIENCE These factors commonly are segmented into the number of reviews a restaurant has, the restaurant’s ranking, food quality, service quality, value, atmosphere, and prior customer rating (Kim et al., 2016). While several studies have examined these factors in-depth, food safety and restaurant cleanliness usually have not been considered. One study posited that food safety issues can present a major concern to a restaurant’s continued success, as the study determined that food safety and restaurant cleanliness issues can decrease repeat patronage (Barber et al., 2011). Our study aimed to utilize text mining to explore the content within customer-generated online reviews through the lens of food safety and restaurant cleanliness. To our knowledge, no prior studies have utilized text mining as part of the methodology to monitor FBI and restaurant cleanliness reporting and no studies have determined the relationship between customer satisfaction and food safety—thus presenting a gap in the scientific literature. Methods To address the two research objectives, we performed text mining and business analytics using XLMiner, Wordstat 8.0.8, and IBM SPSS Statistics (version 26.0; Figure 1). Our study utilized a data set of 231,381 reviews from Yelp from 2005–2017 for 954 restaurants in Houston. For each review in the data set, the following pieces of information were collected: review rating, review text, and customer satisfaction ratings. The initial steps of our study involved defining the research questions and cleaning the data to be used for business analytics. To address research objective 1, we developed dictionaries for text mining containing keywords related to 1) FBI events and 2) restaurant cleanliness issues. The majority of typical foodborne pathogens cause acute gastroenteritis in humans; thus, terms relating to typical symptoms were included (Lucado et al., 2013). Additional terms were added from an exploratory sampling of 100 postings on IWasPoisoned.com (n.d.), a crowdsourced online forum for food service customers who have experienced food poisoning. We selected words for the restaurant cleanliness dictionary for their relation to typical vectors of FBI (e.g., pests), facility conditions that could denote potential pathogenic activity (e.g., sticky, smelly), and violations of the Texas Food Establishment Rules (2021) recorded by health inspectors (e.g., hair in food, no gloves). Due to the nature of customer-generated reviews, common misspellings of keywords were added to the inclusion list. We excluded instances where keywords were preceded by negatives. The frequency of keywords within each individual review was recorded and totaled for each restaurant. The two dictionaries are shown in Table 1. To address research objective 2 and to analyze the relationship between the frequency of keywords and customer satisfaction on a restaurant-by-restaurant basis, we performed a Pearson’s correlation with the frequency of keywords as an independent variable and the restaurant’s average rating as a dependent variable. Furthermore, to analyze this relationship for each individual review, we performed a similar Pearson’s correlation, with the individual’s rating as the dependent variable. Results and Discussion The overarching goals of our study were to explore the usage of keywords related to FBI and restaurant cleanliness in online customer reviews and to examine the relationship this usage had with customer satisfaction. Text mining the reviews for the Houston marText Mining Dictionaries Dictionary Keywords Foodborne illness Vomit, vomiting, poisoning, food poisoning, ill, diarrhea, fever, puke, puking, nausea, cramps, throwing up, threw up, sick, nauseous Restaurant cleanliness Dirty, sticky, slime, slimy or slimey, roach, cockroach, rat, ammonia, hair in, fingernail or finger nail, crumbs, no gloves, smell, smelly TABLE 1 Number of Instances of Foodborne Illness and Restaurant Cleanliness Keywords Keyword # Foodborne illness Sick 418 Ill 285 Poisoning 218 Threw up 53 Diarrhea 50 Nauseous 49 Throw up 46 Vomit 46 Vomiting 43 Throwing up 26 Fever 23 Puke 22 Cramps 20 Nausea 18 Puked 8 Puking 7 Diarrhea 2 Restaurant cleanliness Dirty 532 Smell 530 Sticky 298 Crumbs 182 Hair in 159 Slimy 133 Roach 119 Smelly 62 Cockroach 59 Rat 47 Slime 33 Finger nail 13 No gloves 12 Ammonia 8 Note. From a data set containing 231,381 online customer reviews from 954 restaurants in Houston, Texas. TABLE 2

RkJQdWJsaXNoZXIy NTU5MTM=