Dark

Light

Dark

Light

Social Sentiment Analysis of CBD Consumption

Natural Language Processing

Problem

Despite the rising popularity of Cannabidiol (CBD) products, there is a lack of conclusive scientific evidence regarding its health benefits. Online communities are burgeoning with discussions about CBD, its uses, effects, and health benefits, making them a rich source of unstructured data. However, there is a significant gap in systematically analyzing this wealth of information to understand public sentiment and the prevalent uses of CBD at an epidemiological scale.

Solution

The project aimed to harness the power of Natural Language Processing (NLP) to analyze discussions from online communities, specifically on the Reddit platform, where health benefits and effects of CBD are frequently discussed. The approach involved two key methodologies:

  1. Bio-medical Entity Recognition: This technique was applied to submissions from the Reddit communities to identify and categorize medical entities being discussed. This step was crucial for understanding the context and focus areas of the discussions, particularly related to medical terms and CBD usage.

  2. Aspect-based Sentiment Analysis: Utilizing a BERT-based sentiment analysis model, the project analyzed both submissions and comments to gauge public sentiment surrounding four aspects of CBD: effectiveness, side effects, dosage, and administration methods.

Tools Used:

  1. BERT (Bidirectional Encoder Representations from Transformers): Leveraged as a state-of-the-art NLP model for understanding the context of words in search queries and performing aspect-based sentiment analysis, providing nuanced insights into public sentiment towards CBD.

  2. Bio-medical Entity Recognition Tools: Employed specialized NLP tools designed for recognizing and categorizing medical entities within large text datasets, essential for filtering and interpreting the discussions accurately.

Outcome

The analysis revealed that CBD is predominantly used for managing mental health conditions, particularly anxiety disorders, with sublingual administration being the most popular method. The overall sentiment towards CBD on these online platforms is low, with slightly positive views on its effectiveness but mixed feelings regarding side effects and dosage. This project underscores the potential of advanced NLP methods in mining social media data to track public sentiment and usage patterns related to health products like CBD. It highlights the importance of social media surveillance as a fast, convenient, and cost-effective method to complement traditional surveillance sources. The findings also suggest a need for more research and better public health communication regarding the effects and potential benefits of CBD consumption.