Content-based book recommending using learning for text categorization

By Raymond J. Mooney, Loriene Roy. 1999

Recommender systems improve access to relevant products and information by making personalized suggestions based on previous examples of a user's likes and dislikes. Most existing recommender systems use collaborative filtering methods that base recommendations on other users' preferences. By contrast,content-based methods use information about an item itself to make suggestions.This approach has the advantage of being able to recommend previously unrated items to users with unique interests and to provide explanations for its recommendations. We describe a content-based book recommending system that utilizes information extraction and a machine-learning algorithm for text categorization. Initial experimental results demonstrate that this approach can produce accurate recommendations.

How researchers see this paper

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

Citations in the corpus

How algorithmic confounding in recommendation systems increases homogeneity and decreases utility (2017, background): “forts have been made to understand whether or not various recommender systems impact diversity by reinforcing the popularity of already-popular products or by increasing interactions with niche items [2, 11, 15, 19, 20, 46, 48, 50, 62, 67, 68]. These systems also have the potential to create “echo-c”
Ask the GRU: Multi-task Learning for Deep Text Recommendations (2016, methodology): “Such methods are unable to fully leverage the text content, being limited to bag-of-words suﬃcient statistics [17], and furthermore unsupervised learning is unlikely to focus on the aspects of text relevant for content recommendation.”
Deep reinforcement learning for page-wise recommendations (2018, background): “Another common approach is content-based filtering [21], which tries to recommend items with similar properties to those that a user ordered in the past.”
Explainable Restricted Boltzmann Machines for Collaborative Filtering (2016, background): “Bilgic and Mooney (Bilgic & Mooney, 2005) proposed three forms of explanations: keyword style, neighbor style and inﬂuence style, as separate forms of explanation approaches from their recommender system, a hybrid approach, called LIBRA, that recommends books (Mooney & Roy, 2000).”
Content-boosted collaborative filtering for improved recommendations (2002, methodology): “The choice of the threshold 50 is based on the performance of the content-based predictor, which was evaluated using 10-fold cross-validation (Mitchell 1997).”
Explaining Recommendations: Satisfaction vs. Promotion (2005, methodology): “LIBRA was originally developed as a purely content-based system [9] and has a database of approximately 40,000 books.”
A unified approach to building hybrid recommender systems (2009, methodology): “This information can be meta-data about the item such as actors appearing in movies for movie recommendation or information derived from the item such as counts of words in documents in the case of document recommendation [21].”
A Comprehensive Survey of Neighborhood-based Recommendation Methods (2011, background): “recommendation of books and CDs [47, 53], music [45, 70], movies [31, 51, 5], news [6, 41, 76], jokes [23], and web pages [7, 52].”
Impact of Similarity Measures on Web-page Clustering (2000, background): “Most classical techniques, and even fairly recent ones proposed in the data mining community (CLARANS, DBSCAN, BIRCH, CLIQUE, CURE, WAVECLUSTER etc. (Rastogi & Shim 1999)), are based on distances between the samples in the original vector space.”
Interpretable Policies for Dynamic Product Recommendations (2016, methodology): “This problem arises in many domains such as recommender systems [21, 6] and personalized online advertising [8, 16, 7]; similar methods for matching customers to products are even used for email classiﬁcation and spam detection [20].”
Dynamic Attention Deep Model for Article Recommendation by Learning Human Editors' Demonstration (2017, background): “On the other hand, content-based recommendation [4, 24, 32] fully depends on the item aributes or content thus has no cold-start problem but may not provide personalized recommendation.”
Exploring social influence for recommendation: a generative model approach (2012, background): “Content-based techniques [23] make recommendations by matching a user’s personal interests (or profile) to descriptive information of items.”
Tagsplanations: explaining recommendations using tags (2009, methodology): “The Libra book recommender, for example, extracts keywords from the book text based on their predictive power in a naive Bayesian classiﬁer and uses them to explain the recommendation [7].”
Hybrid Systems for Personalized Recommendations (2003, background): “This is a way to expand the capabilities of a well-understood and well-tuned system, by adding new kinds of features into the mix [2, 24 ].”
MeLU: Meta-Learned User Preference Estimator for Cold-Start Recommendation (2019, background): “Content-based systems are introduced [15, 17] to solve the cold-start problem.”
Tree2Vector: Learning a Vectorial Representation for Tree-Structured Data (2018, background): “Previous work on books [35] and authors [36] mainly rely on”