Browsing by Author "Desrosiers, Chrsistian"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Solving the Sparsity Problem: Collaborative Filtering via Indirect Similarities
    (2008-12-10) Desrosiers, Chrsistian; Karypis, George
    Collaborative filtering is an important technique of information filtering, commonly used to predict the interest of a user for a new item. In collaborative filtering systems, this prediction is made based on user-item preference data involving similar users or items. When the data is sparse, however, direct similarity measures between users or items provide little information that can be used for the prediction. In this paper, we present a new collaborative filtering approach that computes global similarities between pairs of items and users, as the equilibrium point of a system relating user similarities to item similarities. We show how this approach extends the classical techniques based on direct similarity, and illustrate, by testing on various datasets, its advantages over such techniques.
  • Thumbnail Image
    Item
    Within-network classification using local structure similarity
    (2009-03-30) Desrosiers, Chrsistian; Karypis, George
    Within-network classification, where the goal is to classify the nodes of a partly labeled network, is a semi-supervised learning problem that has applications in several important domains like image processing, the classification of documents, and the detection of malicious activities. While most methods for this problem infer the missing labels collectively based on the hypothesis that linked or nearby nodes are likely to have the same labels, there are many types of networks for which this assumption fails, e.g., molecular graphs, trading networks, etc. In this paper, we present a collective classification method, based on relaxation labeling, that classifies entities of a network using their local structure. This method uses a marginalized similarity kernel that compares the local structure of two nodes with parallel random walks in the network. Through experimentation on different datasets, we show our method to be more accurate than several state-of-the-art approaches for this problem.