-
Notifications
You must be signed in to change notification settings - Fork 91
Home
Case Recommender was developed to provide flexibility and extensibility in research environments, while maintaining high performance, providing a variety of recommendation and clustering algorithms, as well as functions for data manipulation. Its primary concern is to maximize usefulness for research and education, instead of large-scale commercial operations. The framework is also designed to support a wide variety of recommendation approaches, including content-based, collaborative filtering and hybrid approaches. Case Recommender is published in PyPi, a international repository of software for the Python programming language, as an MIT License, making it easy for third parties to contribute with additional implementations and features.
The important features while designing our framework were to enable the computation of recommendations in large-scale, easy creation and extension of algorithms for different types of filtering and scenarios. Another feature is to support sparse and large datasets in a way that there is as little as possible overhead for storing data and intermediate results. We propose this framework to be extensible and flexible, as developers can implement new recommendation algorithms while using the available data structures and routines. According to the application scenario, developers can choose between using one of the available recommender algorithms, combining multiple recommendations using one of the available ensemble techniques, or develop their own algorithm using the BaseRatingPrediction or BaseItemRecommendation classes.