My application Shiny

General Information: In this project, data from tweets, blogs and news collected from heliohost.org on September 30, 2016 and retrieved via Wayback Machine on April 24, 2017 was used to generate a predictive model that predicts the next word in a sentence. The application can be accessed at: https://tester-tester.shinyapps.io/Magic_Quote/

The next slides cover:

• Overview of the application;
• Details about the model used in the application;
• Points for improvement.

The application predicts the next word of any sentence in English that contains at least 3 words.

• The data used in this app was collected from heliohost.org on September 30, 2016 and retrieved via Wayback Machine on April 24, 2017. The data is composed of news, blogs and tweets; -he data was cleaned and then n-gram were created; -A Shiny application was also developed to host the model (n-grams). That interface collects the user input that is then processed and, using the last words in the input, matched to an n-gram developed using the corpus of text mentioned in the previous slide.

Though a substantial effort was put into developing the application, it can be improved in many ways. N-grams are not considered the state-of-the-art in NLP anymore and a better model would use embeddings to predict the next word.