Inferring Networks of Substitutable and Complementary Products

By Julian McAuley, Rahul Pandey, Jure Leskovec. Knowledge Discovery and Data Mining, 2015

818 citations recorded by the source corpus.

To design a useful recommender system, it is important to understand how products relate to each other. For example, while a user is browsing mobile phones, it might make sense to recommend other phones, but once they buy a phone, we might instead want to recommend batteries, cases, or chargers. In economics, these two types of recommendations are referred to as substitutes and complements: substitutes are products that can be purchased instead of each other, while complements are products that can be purchased in addition to each other. Such relationships are essential as they help us to identify items that are relevant to a user's search. Our goal in this paper is to learn the semantics of substitutes and complements from the text of online reviews. We treat this as a supervised learning problem, trained using networks of products derived from browsing and co-purchasing logs. Methodologically, we build topic models that are trained to automatically discover topics from product reviews that are successful at predicting and explaining such relationships. Experimentally, we evaluate our system on the Amazon product catalog, a large dataset consisting of 9 million products, 237 million links, and 144 million reviews.

Subfields: Artificial Intelligence, Information Systems, Marketing

View Paper: https://www.semanticscholar.org/paper/6505a9bd862fafcde0315a87695c504ced6c7df6

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

In Information Systems

In Marketing

Citations in the corpus

Cluster-GCN: An Efficient Algorithm for Training Deep and Large Graph Convolutional Networks (2019, methodology): “To test the scalability of GCN training algorithms, we constructed a much larger graph with over 2 millions of nodes and 61 million edges based on Amazon co-purchasing networks [11, 12].”
Jointly Learning Explainable Rules for Recommendation with Knowledge Graph (2019, methodology): “The item associations are extracted from user log on Amazon (same as [20]).”
Explainable Product Search with a Dynamic Relation Embedding Model (2019, methodology): “We use the 5-core data provided by McAuley et al. [37] where each user and product has at least 5 purchases and 5 reviews.”
Comparison of Naive Bayes, Random Forest, Decision Tree, Support Vector Machines, and Logistic Regression Classifiers for Text Reviews Classification (2017, background): “Amazon customers’ product-review data for Android Apps is selected for investigating (McAuley et al., 2015).”
Towards Conversational Search and Recommendation: System Ask, User Respond (2018, methodology): “As in previous work on product search and recommendation tasks [2, 35], we adopt the Amazon product dataset [22] as the experimental corpus.”
Extreme Multi-label Loss Functions for Recommendation, Tagging, Ranking & Other Missing Label Applications (2016, background): “It has been argued that substitutable products can be inferred from the “also viewed” items while complimentary items can be inferred from the “also bought” items [27, 29].”
Dual Attention Network for Product Compatibility and Function Satisfiability Analysis (2017, methodology): “Complementary products are studied in the context of recommender system (McAuley, Pandey, and Leskovec 2015).”
The Emerging Trends of Multi-Label Learning (2020, methodology): “For example, [194] develops an MLC model to automatically recommend bid phrases to an advertiser from a given ad landing page; [195] approaches the item-to-item recommendation task on Amazon, which aims”
Learning to Rank Features for Recommendation over Multiple Categories (2016, methodology): “We choose the Amazon 2 [7, 8] and Yelp 3 datasets for experiments.”
Identifying Unknown Unknowns in the Open World: Representations and Policies for Guided Exploration (2016, background): “5 ) achieves slightly smaller regret than UUB on the Amazon reviews data set.”
Automated Crowdturfing Attacks and Defenses in Online Review Systems (2017, background): “download reviews, or large review datasets made public by researchers [15, 35, 38].”
GPU Kernels for Block-Sparse Weights (2017, methodology): “, 2017] and train byte-level generative models on the Amazon Reviews corpus [McAuley et al., 2015].”
The MovieLens Datasets: History and Context (2016, background): “Amazon review datasets7 [McAuley et al. 2015b; McAuley et al. 2015a] are notable for several reasons.”
Probabilistic Label Tree for Streaming Multi-Label Learning (2021, background): “For example, in webpage categorization [17], labels (categories) are collected in Wikipedia and one needs to annotate a new webpage with all relevant labels from such a big candidate set; in image annotation [4], one wishes to tag each individual picture with multiple labels from such a candidate la”
Robust Learning from Untrusted Sources (2019, background): “Our ﬁrst set of experiments is on the "Multitask dataset of product reviews" 2 (Pentina & Lampert, 2017), containing customer reviews for 957 Amazon products from the "Amazon product data" (McAuley et al., 2015a;b), together with a binary label indicating whether each review is positive or negative.”
Learning to Generate Reviews and Discovering Sentiment (2017, methodology): “The model chosen for the large scale experiment is a single layer multiplicative LSTM (Krause et al., 2016) with 4096 units.”