Inspiration

When working on production support, sometimes it happens that the client gets frustrated and support needs to handle the conversation well. One of the tools that can help spot frustration early on is sentiment analysis on the comments. We can then react to the ticket accordingly before it graduates to the state where we could even lose the client.

Sentiment analysis can also play an important part of project management. Happy teams work better together and deliver higher-value and achieve goals easier then unhappy, demotivated teams.

Analyzing and tracking the sentiment of issues can be a great addition to the project analysis tools. If you spot some issue is being problematic you can quickly check the Issue sentiment and even see the sentiment of people involved in the comments.

So you don't need to read through all comments anymore to see if the discussion is healthy or heated and take quick actions.

What it does

Comment Sentiment analyzer looks through the comments and shows sentiment in the 'Comments tone analyser' section in the right panel of the issue.

Here you can see:

the average sentiment of the issue (includes all comments in the issue)

the average sentiment per user

How I built it

A small but useful Forge app that pulls user input text from issue comments and description and processes using sentiment library and the AFINN-165 wordlist to identify the tone of voice.

Challenges I ran into

There is no straight forward way to apply styling in Forge so we had to use some SVG wizardry to make the tone list show the way it does

Limitations in Forge, we wanted to add the average sentiment emoji in a lozenge on the Glance tab. This would’ve enabled user to quickly understand if the issue needs attention. This is not currently possible with Forge.

Accomplishments that I'm proud of

Making the list of users and their tone assessment look as per mockups

What I learned

The lack of the ability to manipulate DOM elements and styles limits significantly what you can achieve with Forge

That there’s no way to send dynamic data to the glance tab

What's next for Comment Sentiment analyzer

Sentiment analysis is definitely a big topic. A great addition to this app would be to analyze sentiment on the project over time and provide a chart for the dashboard. Then project managers could monitor the chart and if any problems come up (sentiment drops) they can drill down and see what issues influenced sentiment drop most in this period.

Then they could use Issue level of sentiment analysis to see shat caused the troubles and who they need to speak to.

Built With

Share this project:

Updates