Word prediction application

Gr3n4d3s
5/1/2018

Billions of texts, tweets, posts, etc… are created everyday.
Harnessing this data to create a fun and useful experience could be beneficial to market and consumer both.
Consumer engagement would include, Emoji's, Snapchats and mostly non-verbal textual messaging.
Market engagement includes, word/recommendation prediction, sentiment analysis and scale-ability.
The purpose of this application is to create a fun, fast, insightful and textual user experience. While still meeting our market engagement goals

Using word prediction algorithms we've created a simple button click selection model
The model finds the top next three predicted words and enables users to click the appropriate predicted word, or enter any new word they wish
In addition, a word cloud is generated that that dynamically holds more suggestions, based on user input.
In this example we also built a word limiter function so clouds wouldn't get out of hand, and a hard coded cut off at 50 (word cloud rendering is a bottleneck).

Simplicity here was paramount for speed, relativity and scale-ability.
Our algorithm uses knockoff principle, a way of phrase matching that removes the last word, then looks for extended last word matches
This is the best compromise of the previous three points, I.e. speed, relativity, scale-ability.
Ultimately a simple document corpus would be created and updated continuously from the user, however for this app we've developed our own from various social media sources.

We can create a quick and robust word predictor
Word prediction can be a user experience and a convenience
Although simple, this application is very scale-able, and an additional layer of sentiment analysis could branch this into several other markets…
Link to app
And as always, thanks for looking :)