A unified approach to building hybrid recommender systems

By Asela Gunawardana, Christopher Meek. 2009

178 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Information Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yongfeng Zhang
2. Xu Chen
3. Xiangnan He
4. Qingyao Ai
5. W. Bruce Croft
6. Julian McAuley
7. Michael Bendersky
8. George Karypis

In Computer Vision and Pattern Recognition

1. Yongfeng Zhang
2. Xu Chen
3. Julian McAuley
4. Qingyao Ai
5. Xia Ning
6. George Karypis
7. Yue Shi
8. Alan Hanjalic

In Information Systems

1. Yongfeng Zhang
2. Xu Chen
3. Xiangnan He
4. Qingyao Ai
5. Julian McAuley
6. W. Bruce Croft
7. Yue Shi
8. Flavian Vasile

Citations in the corpus

• Modeling Consumer Preferences and Price Sensitivities from Large-Scale Grocery Shopping Transaction Logs (2017, background): “By combining multiple techinques, hybrid recommender systems can be developed to handle a variety of complex scenarios [8,17].”
• Combining content-based and collaborative recommendations: A hybrid approach based on Bayesian networks (2010, methodology): “Generally, collaborative systems report a better performance than content-based approaches, but their success relies on the presence of a sufficient number of user ratings [3,5,6,4,7].”
• Latent factor models with additive and hierarchically-smoothed user preferences (2013, background): “Hybrid recommendation models [9] [8] [6] that combine collaborative and content information have been proposed to overcome this issue.”
• Joint Representation Learning for Top-N Recommendation with Heterogeneous Information Sources (2017, background): “Different algorithms can be assembled by various strategies such as weighting, switching, mixing, cascading, or meta-level hybridization [7, 16].”
• Logistic Matrix Factorization for Implicit Feedback Data (2014, background): “Another popular approach is to utilize item content such as metadata, text, or audio content to improve collaborative filtering models.”
• Meta-Prod2Vec: Product Embeddings Using Side-Information for Recommendation (2016, background): “[10] leans the interaction weights between user actions and various features such as user and item metadata.”
• Recommender Systems (2016, unknown): “The work in [238] leans the interaction weights between user actions and various features such as user-item profile information and side information.”
• Collaborative Filtering beyond the User-Item Matrix (2014, methodology): “As another example, Boltzmann machines were applied to CF by Salakhutdinov et al. [2007], first using only the U-I matrix and later extending the approach to incorporate item content features into a unified recommendation model [Gunawardana and Meek 2009].”
• Sparse linear methods with side information for top-n recommendations (2012, background): “Gunawardana et al [6] proposed unified Boltzmann machines for top-N prediction, in which user-item profile information and side information are treated as homogeneous features, and interaction weights between such features and user actions are learned in a coherent manner so as to reflect how well s”
• Up next: retrieval methods for large scale related video suggestion (2014, background): “It is important to note the connection between this work and the ongoing research on hybrid recommender systems [4, 10, 15, 31].”
• Identifying and utilizing contextual data in hybrid recommender systems (2010, methodology): “Gunawardana and Meek [9] create a hybrid algorithm for recommending movies based on unified Boltzmann machines using movie meta data for hybridization which outperforms other non-hybrid approaches.”