Tomoko Murakami

Metrics for evaluating the serendipity of recommendation lists

In this paper we propose metrics unexpectedness and unexpectedness_r for measuring the serendipity of recommendation lists produced by recommender systems. Recommender systems have been evaluated in many ways. Although prediction quality is frequently measured by various accuracy metrics, recommender systems must be not only accurate but also useful. A few researchers have argued that the bottom-line measure of the success of a recommender system should be user satisfaction. The basic idea of our metrics is that unexpectedness is the distance between the results produced by the method to be evaluated and those produced by a primitive prediction method. Here, unexpectedness is a metric for a whole recommendation list, while unexpectedness_r is that taking into account the ranking in the list. From the viewpoints of both accuracy and serendipity, we evaluated the results obtained by three prediction methods in experimental studies on television program recommendations.

Friendly information retrieval through adaptive restructuring of information space

Although a technique of relevance feedback is common in the field of information retrieval (IR), the feedback is usually done by means of query refinement; restructuring of the information space has not been attempted yet. The restructuring not only allows useful applications such as clustering but also is indispensable for IR if a modeling function employs correlation of terms. In this paper we present a new method of relevance feedback through the restructuring of the information space. Our method adapts document space to the user’s mental model by manipulating a dictionary vector. Therefore, user’s viewpoint is preserved after a series of retrieval processes and reused for retrieval performed later. We show its effectiveness through the retrieval experiments on FAQ (Frequntly Asked Questions) documents.

Green behavior generation: A digital approach to reduce consumption

In this paper, we proposes a new approach called green behavior generation (GBG), which is designed to reduce the environmental burden without mental stress in daily life. A GBG system consists of sensing, modeling, and service elements. For sensing various kinds of data, a sensor home is constructed. The network system of the sensor home is based on advanced information and communication technology (ICT), such as networked home electric appliances and a ubiquitous sensor network, which are applied to collect energy consumption and context data simultaneously. An experimental example and the usefulness of the sensing environment are shown.

Information Space Optimization with Real-Coded Genetic Algorithm for Inductive Learning

New feature construction methods are presented. The methods are based on the idea that a smooth feature space facilitates inductive learning thus it is desirable for data mining The methods, Category-guided Adaptive Modeling (CAM) and Smoothness-driven Adaptive Modeling (SAM), are originally developed to model human perception of still images, where an image is perceived in a space of index colors. CAM is tested for a classification problem and SAM is tested for a Kansei scale value (the amount of the impression) prediction problem. Both algorithms have been proved to be useful as preprocess steps for inductive learning through the experiments. We also evaluate SAM using datasets from the UCI repository and the result has been promising.