Knowledge-based recommender systems: overview and research directions

Recommender systems are decision support systems that help users to identify items of relevance from a potentially large set of alternatives. In contrast to the mainstream recommendation approaches of collaborative filtering and content-based filtering, knowledge-based recommenders exploit semantic user preference knowledge, item knowledge, and recommendation knowledge, to identify user-relevant items which is of specific relevance when dealing with complex and high-involvement items. Such recommenders are primarily applied in scenarios where users specify (and revise) their preferences, and related recommendations are determined on the basis of constraints or attribute-level similarity metrics. In this article, we provide an overview of the existing state-of-the-art in knowledge-based recommender systems. Different related recommendation techniques are explained on the basis of a working example from the domain of survey software services. On the basis of our analysis, we outline different directions for future research.

Minimal Cardinality Diagnosis in Problems with Multiple Observations

Model-Based Diagnosis (MBD) is a well-known approach to diagnosis in medical domains. In this approach, the behavior of a system is modeled and used to identify faulty components, i.e., once a symptom of abnormal behavior is observed, an inference algorithm is run on the system model and returns possible explanations. Such explanations are referred to as diagnoses. A diagnosis is an assumption about which set of components are faulty and have caused the abnormal behavior. In this work, we focus on the case where multiple observations are available to the diagnoser, collected at different times, such that some of these observations exhibit symptoms of abnormal behavior. MBD with multiple observations is challenging because some components may fail intermittently, i.e., behave abnormally in one observation and behave normally in another, while other components may fail all the time (non-intermittently). Inspired by recent success in solving classical diagnosis problems using Boolean satisfiability (SAT) solvers, we describe two SAT-based approaches to solve this MBD with multiple observations problem. The first approach compiles the problem to a single SAT formula, and the second approach solves each observation independently and then merges them together. We compare these two approaches experimentally on a standard diagnosis benchmark and analyze their pros and cons.

Temporal-Fault Diagnosis for Critical-Decision Making in Discrete-Event Systems

Since its appearance in AI, model-based diagnosis is intrinsically set-oriented. Given a sequence of observations, the diagnosis task generates a set of diagnoses, or candidates, each candidate complying with the observations. What all the approaches in the literature have in common is that a candidate is invariably a set of faulty elements (components, events, or otherwise). In this paper, we consider a posteriori diagnosis of discrete-event systems (DESs), which are described by networks of components that are modeled as communicating automata. The diagnosis problem consists in generating the candidates involved in the trajectories of the DES that conform with a given temporal observation. Oddly, in the literature on diagnosis of DESs, a candidate is still a set of faulty events, despite the temporal dimension of trajectories. In our view, when dealing with critical domains, such as power networks or nuclear plants, set-oriented diagnosis may be less than optimal in explaining the supposedly abnormal behavior of the DES, owing to the lack of any temporal information relevant to faults, along with the inability to discriminate between single and multiple occurrences of the same fault. Embedding temporal information in candidates may be essential for critical-decision making. This is why a temporal-oriented approach is proposed for diagnosis of DESs, where candidates are sequences of faults. This novel perspective comes with the burden of unbounded candidates and infinite collections of candidates, though. To cope with, a notation based on regular expressions on faults is adopted. The diagnosis task is supported by a temporal diagnoser, a flexible data structure that can grow over time based on new observations and domain-dependent scenarios.

Diagnosing a Strong-Fault Model by Conflict and Consistency

The diagnosis method for a weak-fault model with only normal behaviors of each component has evolved over decades. However, many systems now demand a strong-fault models, the fault modes of which have specific behaviors as well. It is difficult to diagnose a strong-fault model due to its non-monotonicity. Currently, diagnosis methods usually employ conflicts to isolate possible fault and the process can be expedited when some observed output is consistent with the model’s prediction where the consistency indicates probably normal components. This paper solves the problem of efficiently diagnosing a strong-fault model by proposing a novel Logic-based Truth Maintenance System (LTMS) with two search approaches based on conflict and consistency. At the beginning, the original a strong-fault model is encoded by Boolean variables and converted into Conjunctive Normal Form (CNF). Then the proposed LTMS is employed to reason over CNF and find multiple minimal conflicts and maximal consistencies when there exists fault. The search approaches offer the best candidate efficiency based on the reasoning result until the diagnosis results are obtained. The completeness, coverage, correctness and complexity of the proposals are analyzed theoretically to show their strength and weakness. Finally, the proposed approaches are demonstrated by applying them to a real-world domain—the heat control unit of a spacecraft—where the proposed methods are significantly better than best first and conflict directly with A* search methods.