IMDb is a widely known online movie platform that o ers movie information, allows to rate movies and recommends interesting movies to users. The IMDb movie recommendations do not however o er any means for interactivity or user control, which inherently limits their contextual adaptability. In this work we describe our Google Chrome extension { called MovieBrain { which o ers interactive movie recommendations and integrates the IMDb website for user rating data. Dynamic settings and genre lters are available, allowing users to manually ne-tune the recommendation process and its results.
While the Internet Movie Database (IMDb) website1 is a widely known and popular online movie platform, its integrated recommendations are static and non-interactive in nature. Users have no way of controlling, in uencing or ne-tuning the suggested movies other than by rating more movies and waiting for the recommendation results to change.
It has been shown however that interactivity and user
involvement in the recommendation process increases user
satisfaction and recommendation relevance [
The MovieBrain system consists of a 3-tier architecture which includes a calculation back-end, a middleware webserver and a front-end Chrome extension. All recommendation calculations are performed on a high-performance computing (HPC) infrastructure which iteratively retrains recommendation models in the background and in real time responds to user requests. A Google Chrome extension provides a visual web-based user interface, and a webserver links the back and front-end of the system while adding control, caching and data management. 3.
The MovieBrain recommender system is driven by a
dynamic hybrid recommendation strategy which we described
in previous work [
User ratings from IMDb are complemented with the
MovieTweetings dataset [
Browser plugins integrate seamlessly in people's everyday Internet activity (i.e., browsing the web), they allow to inject custom code into existing websites and track user browsing behavior. Our MovieBrain Chrome extension extends the IMDb website functionality by o ering customizable movie recommendations based on IMDb ratings that users have provided. The beauty in this work ow lies in the fact that for users interested in the MovieBrain service, ratings already available on IMDb can simply be re-used. Furthermore, a Chrome extension has an additional advantage of scalability. Since an extension is a self-contained le, hosted
at the client side, the impact on the webserver will be limited to HTTP calls to its API.
The user interface is based on a two-level visual design where movies are initially presented by their movie poster and title. Clicking a movie, triggers a popover information pane with more detailed movie information. For every recommended movie two action links are available which allow to hide a movie, or open its corresponding IMDb page. At the bottom of the page a more button allows to load more recommendations.
Users can change the individual importance weights of the integrated algorithms and lter movies by intuitively dragging genre buttons to an exclude or include area in the user interface as illustrated in Fig. 1. All requests to the middleware API are logged so that user interaction with the front-end and typical user behavioral patterns (e.g., implicit feedback) can potentially be analyzed. The MovieBrain extension source code can be found on Github2. Over 70 users currently have installed the extension. As more users install and use the extension, more data is collected which can ultimately be used to evaluate the user experience of our integrated recommendation algorithms and interaction process in a very realistic usage scenario.
While IMDb provides movie recommendations to users who have rated movies, the recommendations are static and can not be interacted with. We have created a Google 2http://github.com/sidooms/MovieBrain Chrome extension that integrates public IMDb ratings from users and provides an enhanced recommendation experience by allowing users to in uence the recommendation process using settings and lters. 6.
The described research activities were funded by a PhD grant to Simon Dooms of the Agency for Innovation by Science and Technology (IWT Vlaanderen). The experiments in this work were carried out using the Stevin Supercomputer Infrastructure at Ghent University, funded by Ghent University, the Hercules Foundation and the Flemish Government - department EWI.