COFIDS: A BELIEF-THEORETIC APPROACH FOR AUTOMATED COLLABORATIVE FILTERING

COFIDS: A BELIEF-THEORETIC APPROACH FOR AUTOMATED COLLABORATIVE FILTERING

Abstract

    Automated Collaborative Filtering (ACF) refers to a group of algorithms used in recommender systems, a research topic that has received considerable attention due to its e-commerce applications. However, existing techniques are rarely capable of dealing with imperfections in user-supplied ratings. When such imperfections (e.g., ambiguities) cannot be avoided, designers resort to simplifying assumptions that impair the system’s performance and utility. We have developed a novel technique referred to as CoFiDS—Collaborative Filtering based on Dumpster-Shafer belief-theoretic framework—that can represent a wide variety of data imperfections, propagate them throughout the decision-making process without the need to make simplifying assumptions, and exploit contextual information. With its DS-theoretic predictions, the domain expert can either obtain a “hard” decision or can narrow the set of possible predictions to a smaller set. With its capability to handle data imperfections, CoFiDS widens the applicability of ACF to such critical and sensitive domains as medical decision support systems and defense-related applications. We describe the theoretical foundation of the system and report experiments with a benchmark movie data set. We explore some essential aspects of CoFiDS’ behavior and show that its performance compares favorably with other ACF systems.

Existing System

The DS-theoretic framework that offers a particularly convenient mechanism to represent a variety of data imperfections .DS-based techniques have been exploited in applications where the integrity of the decision-making process and its robustness against modeling errors caused by lack of precise information are critical. The tasks of identifying similar users are not easy when the ratings matrix is sparse.

Proposed system

Propose a technique that addresses these issues (in particular, the circumstance of data imperfections) and also allows one to exploit background knowledge that may be available in real-world applications. DS-Theoretic data model. Our system CoFiDS is based on the DS-theoretic framework that offers a particularly convenient mechanism to represent a variety of data imperfections (see Table 1). DS-based techniques have been exploited in applications where the integrity of the decision-making process and its robustness against modeling errors caused by lack of precise information are critical (e.g., in battlefield target tracking, situation awareness, etc.)  Incorporation of Contextual Information. Instead of relying on the entire user population, ACF systems often operate with a subset of users “similar” to the one whose ratings are to be predicted [36]. The tasks of identifying similar users are not easy when the ratings matrix is sparse. For instance, in the aforementioned HAART scenario, the number of drug cocktails prescribed to each patient is small compared to the number of available drug cocktails; and many drug cocktails might have never been rated at all. The number of items coated by more than one user is thus small, and the ossibilities to combine ACF with other recommendation systems and on explaining the predictions of ACF algorithms. One important aspect, though, seems to have escaped the attention of the research community: virtually, no previous work provides a mechanism to accommodate user rating imperfections (e.g., ambiguities, uncertainties, etc.). In certain application domains, this imposes too severe a restriction. For one thing, such ratings perforce are subjective, and research ought to address ways to deal with imperfect ratings.