Does your child explain Bluey episodes to you but you have no idea what episode they are talking about and can’t handle flicking through all 131 of them?
Problem solved.
This website lets you type in the vague descriptions of a small child and it will return a mathematically ranked list of closest matching Bluey episodes.
https://deanmarchiori.shinyapps.io/blueysearch/
Development notes
The website is a Shiny app (deployed to shinyapps.io), which contains all 130 episode titles, descriptions and thumbnails.
All of the episode descriptions were tokenized into ‘terms’ using the {tidytext} R package and formed into a Document-Term-Matrix. Rather than use the typical term counts, a binary indicator was used if a term appears in an episode. This was preferred as the user’s text input is unlikely to really mimic the detail of an episode description, which threw the similarity measure out a bit.
Once a user inputs text, the Shiny app dynamically forms a new term vector and compares it to the all-episode’s matrix using cosine distance. The episodes are then ranked based on smallest cosine distance and displayed to the users.
For the source code visit: https://github.com/deanmarchiori/bluey-search