Project Marshal

Inspiration

Project marshal is a project aims to solve one of the most pressing problem of social webpages. Humans are always been emotional creatures and we tend to expresses our self to others in order to get understand. The most under-rated statement is screaming of comments on social media platforms like YouTube, Facebook and twitter etc. For this project we chose YouTube as our platform. Comments of any YouTube video can reach not only thousands but also millions in the number. Project marshal is the best way to understand and summarize those millions of comments to understandable and more meaning full state at which the admin and the uses can understand public opinion in a better prospective. Through this project I hope to make a new and whole-sum view of the content that is created by on the social media platforms. The reach of the project marshal will not be limited to one platform but extends its wings of justice to over platform where there are comments and personal opinion involve.

What it does

CSE is a wonderful field where engineers thrive to solve real life problems with the help of technology they learn for 4 years. Observation, determination and knowledge is fuel to life of automation. This project is based on the gap that been growing alone with the development of technologies like internet and social media.

Social media is a platform of various people express where views. Especially platform like YouTube where admins of several channels try to build content for the public to view. By digitalization of this era the number of users of internet is being increased by a vast amount. I personally watch YouTube for more than 8 hours a day. It can be for different uses, but most for the educational purpose. The particular difficulty is to choose the right video to follow because one video for any development can be pretty long on average of 4 hours, thus in order to understand where that video is useful to me takes a lot of time and effort. Not only this, let’s see the problem from different perspective of admins. Lot of channel admins are struggling to understand the opinion of their viewers. We all know that comments are best way to find the opinion of the viewers.

Thus, it raises a major problem. Let’s take an example, channel pewdiepie and T-series these channels compete with each other on the subscriptions. Pewdiepie has 100 million subscriptions, let’s think 1% of those subscribers' comment on his video that’s 1 million and now admin have to read one million comments to understand the public opinion.

Project marshal is the perfect solution for this problem.

How I built it

With the help of streamlet which is a package in python that provide the very basis User interface for the project. User is going to type the url of the Video she/he is going to analyze and with the help of selenium web driver which loads the link in an automated testing environment which allow us to collect the comments which are under comment tag in YouTube webpage.

The most difficult part in marshal is interaction of python with browser which is completely build on html, CSS and JS but development of Marshal is on python. Thus, selenium was the best way as it supports several languages like c+, java, python and ruby.

After the completion of the gathering of comments user is provided with options like tokenization, summarization and sentimental analysis.

What I learned

optimisation is an important part of any ML model

What's next for Project Marshal

To deploy an fully functional ML model using tensorflow that is specifically oriented for Mashal needs.

Built With

• machine-learning
• natural-language-processing
• python
• sklearn
• streamlit