How Double Filtration of Review Works?

Published in

GigTricks

7 min readJul 20, 2018

Reviews are the most fundamental part of any freelancing platform because it helps people who acquire the services of the freelancers to share their experience and their views about the services as well as the freelancer. Reviews are powerful and they are the ‘word of mouth’ that will carry forward your reputation in the digital world, no matter what services or products you are selling. If a review is positive and constructive then it will add up to your service credibility but if you receive a negative review, then it can create a bad image of your service in potential prospects’ mind.

So the question here arises, what measure freelancing platforms take to ensure that only the concise, clear and constructive reviews are posted that is honest and doesn’t exploit any user’s professional image in an unfair manner? Well, for the majority of existing traditional freelancing platforms the answer will be, NO? So, does it really necessary? Do they even have a mechanism in place? Do they even care to empower freelancers? Most of the platforms are centralized and doesn’t really take into regard you as a person with needs to grow and earn, but a mean to earn money for their platform as they charge commission on your completed gigs. So the more the gigs, the more is the platforms’ earning, rest is all your business.

But what if I tell you, there is an ecosystem that has considered all scenarios and designed its review filtration system accordingly to create the maximum advantage for its users. Every review will go through a dual filtration process to make sure freelancers can have a ‘say’ in the review filtration process. At first, the automatic initial screening will be done and then the freelancer will perform manual filtration. And every review will be scored and that score will be added to the freelancer’s overall profile rating and the individual skill rating. GigTricks is that ecosystem, it is a blockchain based completely decentralized and integrated freelance and on-demand ecosystem.

So how exactly, GigTricks double review filtration process will work and what technology they are using to achieve maximum transparency.

What Will Happen?

Let’s say you are a graphic designer with a number of designing skills all recorded & verified in your GigTricks blockchain based Pro profile. An entrepreneur with a need to design his website logo checked your profile and hired you to design a spectacular logo for his cosmetics business. The client gave you his requirements and all the details you need to know about the business. You completed the order and upon achieving all elements of the smart contract, you received your payment.

However, at GigTricks, there will be an effective reward contributors’ system in place that will allow the system to reward people with high engagement, performing well anda for posting reviews. Therefore, when you will complete the order, there is a good chance that the client will post a review for your services. That review will go through some filtrations, whose mechanism is discussed in the later section of this article.

What are the Key Features?

Following are the key features of GigTricks review system:

1. All review data will be stored in the hyperledger fabric that is immutable, meaning that the data cannot be deleted or edited to benefit oneself, creating ultimate transparency within the ecosystem.

2. All reviews will go through an authentication process with the help of the automatic review filtration process using technologies such as Machine Learning, Neural Networks, and IBM Watson.

3. The seller will have the ability to either reject or accept a review as well as to raise a dispute, making it hard for deceitful clients to blackmail sellers to post a negative review.

Automatic Review Filtration Process

In the first step of the automatic review filtration process, the review will go through an initial screening to keep the ecosystem safe from malicious and scam users. For this purpose, neural networks and machine learning will be used. The machine learning is a great approach that will help us to identify scammers quickly by recognizing the patterns and anomalies. If you are wondering, why using machine learning is an effective approach, then read along as the later section of this article got all the answers:

Why Machine Learning & Neural Networks?

We intend to use machine learning because writing programs by hand cannot cope with the changing behaviors of the users, but using machine learning will allow us to gather examples of the correct outcome of a number of inputs. The example cases will be processed by machine learning algorithm and it will generate a program accordingly that will be more efficient than any handwritten algorithm and will adapt to the new set of data and train on it if the user behavior changes. In contrary, the neural network is created by taking inspiration from the human brain and it is a set of algorithms that have the ability to detect patterns from a set of data. Together both of these advanced technologies will help us in screening the review.

Once the initial screening is done the review will go through the IBM Watson Tone Analyzer and Watson Natural Language Understanding filters.

Detecting Buyer Moods, Emotions & Sentiments in Review

Since GigTricks plans to score each review posted by the buyers, it became crucial that each of the review goes through an in-depth scrutiny to make sure, the freelancers are scored and ranked properly for the services they offer, their professionalism and ethics. In order to do so, analyzing the sentiment, mood and emotion behind every review became a necessity.

Watson Tone Analyzer will help in detecting the mood of the buyer and help us in recording and scoring the review accordingly, Conversely, the Watson Natural Language Understanding will help the ecosystem to determine the sentiments and emotions of the buyer.

How IBM Watson Technology Works?

With the age of bigdata on the rise, it has become difficult to find value in the heap of available unstructured data. As a result, traditional data analyzing methods are failing. But cognitive computing allows organizations to find value from the inside of intertwined data. The IBM Watson Technology works on the same cognitive computing and the way, humans find the solution to a problem; the four key steps include, observe, interpret, evaluate and decide. Unlike the traditional data analyzing techniques, Watson can understand data that is not necessarily structured at an impressive speed and scale. Today, 80% of the data available is unstructured because of the increased use of the internet and social media platforms and Watson works on grammar, context, and culture to read this type of data. It reads and interprets data like a human by breaking down sentences structurally, relationally and pragmatically. It helps Watson to understand the intent behind a particular sentence or paragraph to draw on more accurate and relevant result. This is the feature that will help GigTricks review system to analyze mood and emotions behind a particular review.

This may seem a little intimidating to some people with a lesser tech background and you may be wondering, how exactly Watson learns all that. Consider this, in order to make Watson the perfect help for filtering reviews, with the help of human intervention relevant data will be fed for creating the corpus of knowledge in the relevant field. Mostly the human intervention is required to discard any useless, outdated and poorly regarded data. This process of forming the corpus is known as curating the content.

Once the corpus has been created, Watson will be required to be trained by the human experts. How? The experts will train Watson in machine learning by uploading data in the form of questionnaires along with answers. It will teach Watson the linguistic patterns and it will continue to learn from the future activities that it functions on. In order to boost Watson’s performance, a human will be in charge to keep a check on Watson’s working every now and then as well as to also update the information to help Watson become better and up-to-date with the changes in information.

Why GigTricks Plan to Use Custom Filters?

GigTricks is planning to make an authenticated review system that will not only help them in scoring the review to rate freelancer’s overall profile rating but allow the ecosystem to help in rating the freelancer on a micro level. Wonder what micro level rating mean here? The masterminds behind GigTricks want to create more benefit and open new doors of opportunities for the freelancers. In order to achieve this goal, freelancers will have two types of rating; one overall profile rating and individual skill rating. For instance, if you are a graphic designer, your GigTricks Pro profile may contain different skills with their respective ratings such as logo design, web design, banner design and so on.

In order to make the individual skill rating work, it is required that review score also add up to the skill rating that is why GigTricks seek to deploy numerous custom filters that will in scoring the review at both macro and micro level. Moreover, it will also the review system to detect any unconstructive review posted by the buyer to make sure freelancers are not negatively rated for it.

Manual Review Filtration Process

After the automatic filtration process is successfully completed, the review will then come to the freelancer for manual filtration. You, as a freelancer will be required to read the review and decide, whether the review posted by the buyer is fair or not. Based on your decision you will be able to either reject or accept the review. No matter what decision you take, the review will be visible to all ecosystem participants because all the review data is immutable and stored in the hyperledger fabric.

However, if in disagreement, you as a freelancer will have the ability to respond to the review and clear your position, similarly the client can also comment to shed light on his side of the story. Yet, if things go out of hand or on a bitter side, there is always a chance to raise a dispute that will be handled in a decentralized way by a community voting system.

GigTricks Pre-Sale is live with 50% discount on GigBit Tokens.