Sentinal - Human Sentiment Analyser

Where technology becomes human.

Comment

Inspiration

Cyberbullying or cyberharassment has become increasingly common these days, especially among teenagers, where social acceptance is such an important part of the experience. The issue becomes even more critical when it leads to teen depression and if it continues over a long enough period of time, it holds potential for some seriously adverse consequences. As per reports, 30 per cent of the children who have been bullied online have suicidal thoughts. Various attempts have been made in the past on varied scales in form of laws and regulations to mitigate this problem. However, finding an effective solution for this burning issue still remains the need of the hour.

This is the reason why the project was conceived.

What it does

The first part of the project is a sentiment analyser. Built on the Aylien API using Python, the project uses Machine Learning to provide a polarity for a given query which is of three types - Positive, Negative and Neutral. If the polarity of a query is negative, it is deemed unsuitable.

The second part of the project is for Brand/Product Analysis and is based on the previously discussed model. Here, instead of taking queries from the user, it takes data from social media platforms like Twitter, analyses them, creates a CSV file and then provides a pie chart of the sentiments for the same.

How I built it

The project is built on Python and used Aylien-API Client link for predicting the Polarity values. The reason for using this API is its accuracy and speed. For brand/product analysis, the project uses APIs like Tweepy link for gathering tweets, Aylien API link client for the predictions, Numpy link and Pandas link for working with data and Matplotlib link for plotting the outputs.

Challenges I ran into

When it comes to sentiment analysis challenges, there are quite a few things that we had to struggle with in order to obtain human sentiment analysis accuracy. Sentiment or emotion analysis can be arduous in natural language processing simply because machines have to be trained to inspect and understand emotions as a human brain does in addition to understanding the shades of different languages. One of the major challenges that we ran into was when working with compound sentences. Although there were ML models available but the accuracy was quite low. To overcome this challenge, Aylien API Client was used.

Accomplishments that I am proud of

Although it started off as a personal project, I soon realized it has the potential to save lives and create a better world as the sentiment analyser has the potential to gain a deeper understanding of a person's mindset. By automatically sorting the sentiment behind social media conversations, reviews, and more, a lot of disastrous situations can be prevented. Not only this but by understanding people's mentality, the youth can be saved from several scams and hazardous schemes having to potential to ruin their lives. This to me is an accomplishment.

What I learned

This project helped me a lot in understanding behavioural patterns of both the general mass, scammers and the people who peddle polarising content. I was also able to learn more about product placements and branding. Building this project sharpened my programming skills too and I got to learn various new technologies as well.

What's next for Sentinal - Human Sentiment Analyser

Sentiment Analyser

  • One of the future work plans for this project is its integration into web browsers. Once the program is integrated via an extension, the program can parse through each line of text on the website, check its polarity and strike out the things that are offensive (polarity= ‘negative’).
  • One more way of ensuring a safe internet is by image analysis. An image on a website is taken in as a query and a polarity is generated which can then be checked for its content. But once again, it’s a part of the future work plan.
  • One more work plan that we have is a port for Indian Language.

Brand/ Product Analyser

  • I am also thinking of creating a GUI based business suite for entrepreneurs for the easy use of the software for brand/product analysis.

Feedback Form Analysis

  • I am also thinking of creating a GUI based application for analyzing and inferring useful outcomes from the feedback forms.

Built With

Share this project:

Updates