In the modern times of online shopping, its easy to get catfished. Usually buyers tend to crave for warranty, in this case the closest thing to warranty is coverage done by other customers.
Suppose you are trying to purchase a product which has only 3 reviews and all of them are positive, but when you receive the product its quality is not as advertised, reviewed or not up to your expectations.
Chances are those reviews were added to lure customers, since it’s easy to deceive people with enchanting things.
This model tries to break down into the Reviews part of an online shopping site (Amazon).And suggests us reviews based on a reliability rate.
This model is split to:
- Study the relationship between online reviews and how customers approach to them.
- Deciding parameters from a human’s perspective.
- Categorising the statements based on various parameters
- Attaching values to each statement like a score.
- Categorizing the statements based on the reliability rate.
To understand what goes on in a person’s mind when he is captivated by the reviews we must try to look at all the items at display.
These attributes help gain his trust.
Considering parameters and attributes in a project where I need to understand thinking of the masses I decided to take a poll.
I took inputs from a google doc form which I have shared, before deciding these attributes.
From a raw perspective we get stars count, verified tag, people who found it helpful count, images and date.
Also we could access edited tag.
Each customer is different, he/she has a different expectations from the product and therefore a different opinions. But one thing common is that these opinions can be changed as they get biased after reading the reviews.
Another take-away from my analysis is that users must be shielded from unreliable reviews since it would waste their time, efforts and decrease their trust on the seller or the website.
This model was built on assumptions made on front end part of Amazon. The display part. Naturally if this was used by Amazon itself they could use it to sort the ratings again.
Also it could be used by, for and on other E-commerce platforms and companies after they tweak the working to suit themselves.
This model should not be limited to reviews itself. Consider a broader perspective With considerable tweaks it could be used on Platforms like Twitter, Youtube and Facebook.