The Natural Language Prediction Model application predicts the next word from the input of a sentence or phrase using probabilities of two-word or three-word combinations determined from a large dataset provided by Swiftkey.
2024-12-13
Language Prediction Model Description
Language Prediction Model Flow
- Model Input:
- Enter a sentence or phrase into the text input box.
- Press the “Predict Next Word” button.
- The system captures the input.
- Enter a sentence or phrase into the text input box.
- Model Processing
- The user’s text is passed to the model for processing.
- The input is “cleaned” of extraneous words and split into words.
- The prediction function is called using probabilities to determine the next word.
- Model Output
- The Input is displayed along with the most likely predicted next word.
Example Model Input & Output
The user enters the text then clicks the “Predict Next Word” button.
The text string is passed to the prediction function for a predicted next word.
Memory Management and Performance
- A 20% random sample from the data set was taken and cleaned for a total of 538.2 MB
- The percent can be adjusted based on memory capacity.
- There is a hesitation for the first output for the application.
- Additional output is returned in less than a second.
- Additional output is returned in less than a second.
- Application access from: https://bmerr.shinyapps.io/Prediction/