You meet someone, exchange numbers and start chatting with them. You want to make a strong first impression and keep the conversation going - not let the person feel like you don't know abut something or haven't caught up on the news yet. What if you could get live prompts about the topic being dicussed and sentimental feedback on how good your conversation is going? You would totally own the conversation! Being able to translate messages within your chat window is just another awesome feature that this application provides. In short, never screw up your chances of getting a date ever again!

What it does

This application brings smart topic suggestions and prompts to your chat in real time. It also helps you gauge the overall mood of the conversation to help you improvise on your conversation style. We've also added a translator to give you instant access to translating anything to the language of your choice. With all this, our users will never feel lost in a conversation ever again!

How I built it

This chat application uses Moxtra's chat client and Microsoft's Cognitive APIs to help you get the upper hand in your chat conversation. It makes use of Micrsoft's Text Analytics and Bing Search APIs to automatically detect and present the user with snippets of information relevant to the topic of conversation. The Moxtra client provides the basis of the chat and we build on its collaborative capabilities with our features. For every incoming message, we use the Micrsoft's Sentiment Analysis API to give the user a feel of overall conversation so far. We also use the Keyword Extraction and the Topic Detection APIs from the Cognitive API suite, to extract keywords from the incoming messages and detect their categories (context) to be able to use Bing search to give the user an access to the latest news and web articles relevant to the topic. We also use Wikipedia's API to display text snippet's on the most significant topic in the conversation. This application also uses Micrsoft's Language Detection and Translate APIs to help the user translate text at his finger tips to a language of his choice.

Challenges I ran into

  • Fixing CORS issues - using a proxy server
  • Working with the restrictions of using a chat client in an iframe - it couldn't be modified
  • Creating test users for Moxtra on Azure

Accomplishments that I'm proud of

  • Completing all 3 features - topic detection and relevant suggestions, sentiment analysis, language translation
  • Enhancing UI with additional features - Live Preview to view the content of the recommend links

What's next for chatwise

  • Extending the same support for images and voice messages as well
Share this project: