Why Predict the next word?

Around the world people are using mobile devices for a whole range of activites. This means a significant amount of typing text.

The small screen and touch screen keyboards are not as ergonomic making typing a text beyond a short paragraph challenging.

Any app which can facilitate this process is helpful.

The app described here predicts the next word using the previous words typed in a sentence. This Shiny application is available on:

The Implementation

The dataset includes 3 large text files including blogs, news and twitter messages.
Blogs represent colloquial English, the news represents formal English and Twitter messages represent abbreviated informal English.
The dataset can be obtained from:

Creating a large corpus of text: A corpus consists of a large body of structured text. It is analysed in order to validate linguistic rules or as part of Natural Language Programming. Creating a corpus involves necessary steps such as removing punctuation, capital letters and white spaces.
Creating Ngrams: unigrams, bigrams, trigrams, tetragrams and pentagrams are created using the Ngram Tokenizer package. Due to limitation in amount of storage and working memory for the proposes of this project only 10-percent of the available body of data was used to create the NGrams. which were stored as .rds files
The predict word app: uses ngrams and Back off method to predict the next word in a piece of text being typed. The user provides a word or a phrase of n words and the app gives a list of 10 possible words which best match the repos.

You can access the model in the following Git-hub repository: