Sentiment Analysis for Better Public Health Analysis: Vaping

Vaping Is a New Phenomenon Whose Effect on Public Health Is Still Not Completely Understood

Electronic cigarettes (also known as e-cigarettes or e-cigs) came on the US market in 2007 and have become a popular product. Basically, e-cigs contain liquid nicotine that is transformed into a vapor that is inhaled by the user.

E-cigs are like conventional tobacco cigarettes in the way they look and how it feels to use them. They do seem to reduce craving, but to date there haven’t been many studies on their health effects, particularly long-term ones. There have been many claims that they are a good smoking cessation aid, but again this has not, to date, been backed up by scientific evidence. E-cigs do contain fewer chemicals than conventional cigarettes, but little is actually known publicly about what chemicals they do contain. Whether they are harmful or not is still a matter for debate.

E-cigs are also very popular with young users because they are perceived as safer than traditional cigarettes, but there are worries about these users becoming nicotine-addicted at an early age. There is also research that suggests e-cigs may even lead to simultaneous use of tobacco cigarettes.

Social Media Is an Important Data Source for Studying Vaping

So the debate is still ongoing. An important new data source for developing greater understanding of vaping is that of social media. People, especially youth, go on-line and post about all aspects of their experience with e-cigs. This data can be analyzed so as to understand public attitudes towards the subject. Given the vast amount of social media data, it would not have been feasible for human researchers to study the data manually. They need a technology that can help them. Fortunately, there is one: Sentiment Analysis.

Using Sentiment Analysis to Research Public Attitudes towards Vaping

Sentiment Analysis finds the sentiments expressed in unstructured natural language, such as found in tweets. It locates words and expressions that communicate sentiment, e.g., “E-cigarettes are better than regular cigarettes.”  Sentiment Analysis analyzes sentiment automatically and also at great scale, something required to handle voluminous social media data.

Consider the following tweets:

• “E-cigs ease my anxieties”
  • Sentiment: Positive
  • Entity: e-cigs

• “e-cig prices are way high”
  • Sentiment: Negative
  • Entity: e-cig prices

Sentiment Analysis detects the sentiments expressed about e-cigs in each of these and also understands the polarity of the sentiments, organized in this case into a simple scale of POSITIVE and NEGATIVE.

Sentiment Analysis, in particular the more advanced Entity-based Sentiment Analysis, produces a structured representation of the sentiment, including the entity the sentiment is being expressed about. Recognizing the specific entity in a tweet that a sentiment is aimed at is extremely useful as it allows researchers to analyze many aspects of public attitudes towards vaping.

Sentiment Analysis Provides an Effective Tool for Gaining a Better Understanding of Public Attitudes towards Vaping

People comment on all aspects of vaping on social media. At a high level, Sentiment Analysis can give researchers a clear view of how popular vaping is among the population and how it evolves across time. In more granular fashion, Sentiment Analysis can allow researchers to zero in on specific complaints and praise (“Vaping is pretty hard on my throat”, “Vaping irritates my eyes”, “E-cigs taste so good”).

Recently the FDA announced a ban on menthol-flavored regular cigarettes (after banning of flavored e-cigarettes in 2019), but not on all e-cigarettes. The public health impact of vaping is gradually becoming clear, but more research is needed. Sentiment Analysis is a critical tool in enhancing our understanding of the public attitudes towards vaping.