Insights provided by facebook are quantitative not qualitative. Difficult for community leaders to track the semantic response of the members and to determine the type of overall response.

What it does

It analyse the comments on the community posts and indicate the positivity and negativity of that comment. Also, overall semantic analysis of the post is done and on the basis of positive and negative comments we can relate the keyword of the post to the sentiment it generated in the community. Then we can promote the topics of the posts which got an overall positive response and moderate the topics which got negative response from the community members.

How I built it

We used react to imitate the facebook post and comments to show a demo that how our system would look. Then the dashboard is build using Bootstrap and Javascript where the community leader can review the semantic analysis easily with super interactive and responsive graphs. The UI/UX is very important to us as most of the community leaders are not technical person so they are intersted in the content rather than technical data so we spent most of the time to improve the UI of dashboard so it's more intutive and more user friendly and help admins and community leaders to make their important decisions. Also, I extracted important keywords and transform into word cloud so that community leaders can get an idea about the content just by viewing.

Challenges I ran into

The group API gave us short lived tokens so we couldn't load the dynamically into the system. Instead we saved the response and used that data to show a demo of our prototype.

Accomplishments that I'm proud of

On taking to various community leads I discovered that this idea solved many of their problems. Mainly how to engage more and more people. This can impact communities on a very wide scale by building a positive environment and promoting positive topics.

What I learned

How to work around a specific problem and try to think from someone's else perspective to build solutions for then, here, community leads.

What's next for Content sentiment analysis

We discussed with some Facebook people and they were really excited to think that this feature might pair well with the new Upvote/Downvote feature that Facebook is rolling out soon.

Built With

Share this project: