December 12, 2020
Web scraping, data science, javascript, Python, Flask, HTML, and CSS are most of the technologies used to build this project. It is not fully featured, but I was able to learn what I needed to build it and deploy it on Heroku.
Heroku has removed its free tier, so this link is no longer active.
For this project, I wanted to build a recommendation engine for the dubbed anime that I watch on Funimation. I was pretty close to hitting that goal, but the list is not perfect as I could not directly scrape the FUNimation website. To build the dataset, I pulled titles from FUNimation along with additional information from IMDB and MyAnimeList.
With the data, I built the recommendation engine using a content-based filtering approach. A content-based filtering approach ultimately takes all of the titles and metadata and determines how similar or dissimilar they might be. When a user selects a title, it finds the most similar in the ordered listing of similarity.
To build the web application, I used the python FLASK web framework with HTML, CSS, and Bootstrap. There is a little javascript incorporated for the type-ahead function against the available titles on the landing page’s input field.
The hosting for the app is Heroku’s free tier, so the performance is very variable.
The logic for this application happens in real-time, and the data is all hosted on flat files. To improve this solution and continue my learning, I want to place the recommendation logic into an API. I also would like to put all of my data about the anime titles into a database. Additionally, since I now enjoy anime from a host of sources, I want to expand the anime listings’ scope.