Helping Customers Buy The Right Product Online
Shopping online used to be very uncommon; however in recent years, online shopping is becoming a norm. Online shopping is becoming more and more popular each day.
Online shopping is becoming more and more popular because it is more convenient for customers to purchase products. Despite all the advantages that come with online shopping from convenience all the way to product choices, one main disadvantage that comes along with online shopping is you cannot see the product first hand.
When you are shopping online, you obviously do not have the ability to see the product first hand, so the closest thing the customer actually has to view the product is the product reviews by other customers. Now product reviews are not actually as useful as they should be. There are so many reviews for each and every product that it is impossible to read every single review, and what happens is the customer reads the first 10 reviews at max and then decides whether or not they want to purchase the product. So essentially, the customers are using a very limited amount of data points to make their decision, and this is what gives online shopping a disadvantage.
However, what if I told you that we can completely get rid of this disadvantage. We can actually mitigate this disadvantage and make shopping online basically perfect. This can be done using a model I built called Review Analyzer.
Review Analyzer is a model that I built leveraging Natural Language Processing (NLP). The model takes all the reviews for a product on Amazon ,for example ,and analyzes every single review, and then it outputs a word map of all the positives and negatives for the product. This allows the users too see the main positive and negative words describing the product, and also be able to see how often it was written from every single review for that product.
The Backend Of Review Analyzer
Now that we understand what Review Analyzer does, et’s understand how it actually works.
The first thing to understand is that we are using Natural Language Processing (NLP), which is essentially the interaction between human language and computers. NLP is actually used in your day-to-day life, ; forxample w,hen you type something and there is spell check being performed, that is NLP! NLP is super cool, and for this project, I used it to perform text analysis on the product reviews.
Now that we understand what NLP is and the uses of NLP, we can dive into the technicalities, the actual code behind the model. The entire back end of Review Analyzer works in three sections and was all programmed in python.
- Data Preparation And Input
In this section of the model, we first have to find the data for the given product. For example, we can choose a product like the iPhone XR. We would get a dataset of all the reviews for the product that arispecifically on Amazon only because that is where the customer is said to be purchasing its product. To find the dataset, you might have to look at a few resources, but I used Google Datasets, and I definitely recommend it for any datasets you are looking to find as well. Now that we have our dataset for the product, we can go on to the core part of the model.
2. The Brain + The Engine
This section is the main part of the program. In this section, the code first tells the model to look at the review dataset, and for each review that has a two-star rating or less, classify that as a bad review, and if the review has a 4-star rating or more then classify that as a positive review. The number of stars classification was done based on the fact that all the reviews on Amazon are out of 5 stars. Now the program runs two separate analysis programs on the positive and negative classified reviews. For each positive and negative set of reviews, the model performs text analysis and sees the most used terms used in both the positive and negative reviews. That concludes the second part of the model.
3. The Output and Results
This is the section where the user receives its results. The user will receive two-word maps, one for positive and one for negative reviews, and the way it works is the more frequent the word was used in either the positive or negative reviews, the bigger it will appear on the woworldap. Now with this output, the user is finally able to make a decision as good as if he were to go in-store, as he is able to get an idea of every single review that has been made for that product on the Amazon platform!
Next Steps…
This is not the end of this model! In the coming month, I am working on building this into a chrome extension, to make it more user-friendly for all the online shoppers out there! This model will help online shopping be as close as it can to being perfect! 🛍
Key Takeaways:
- NLP is a topic that is super popular right now in the world of machine learning.
- NLP is the interaction between human language and computers.
- There is a major flaw with online shopping, which is that there are so many reviews for each and every product that it is impossible to read every single review, the customer reads the first 10 reviews at max, not giving them enough data points to base their purchase decision off of.
- Using the Review Analyzer, which is an AI model that uses NLP, we can get rid of this flaw that online shopping currently has.
- The next steps are to implement this model into a chrome extension to make it more user friendly.
If you enjoyed this article:
- Share it with your network
- Follow my LinkedIn, Twitter, and Medium to keep updated with my progress in AI!
