Data and text mining is a science that statistics, machine learning, and data bases technologies are unified into for the purpose of knowledge discovery from a large amount of data. Specifically among the advanced topics in this area we pick up the issues of statistical outluer detection, outlier filtering rule generation, time series mining, and reputation-analysis on the web, and introduce an approach to them from the pointview of information-based induction sciences.

View full abstract

The rapid development of modern information technologies has led to important progress in the collecting, processing, and dissemination of data. This has resulted in the accumulation of tremendous amount of business data in databases. The ability to leverage the data has become a key success factor in an increasingly competitive global market. Knowledge discovery in databases or data mining is a new technology or methodology that seeks to automatically extract meaningful knowledge from business data. However, it is not a trivial task. Also, in order that data mining techniques can be used in real business, it is crucial to extract meaningful knowledge that can be turned into business action. The purpose of this paper is to demonstrate how useful knowledge can be extracted by new techniques of analyzing string patterns that were originally developed for areas such as genetic sequencing. We shall demonstrate a few experimental results.

View full abstract

As an approach to data mining, we focus on technologies for learning in neural net-works., In this paper, after explaining two discovery algorithms developed for modeling complex phenomena, we describe an approach to improving scientific models that are cast as sets of equations. As an application using these methods, we review one such model for aspects of the Earth ecosystem, then recount its application to revising parameter values, intrinsic properties, and functional forms, in each case achieving reduction in error on Earth science data while retaining the communicability of the original model.

View full abstract

We consider the problem of finding neighboring class sets. Objects of each instance of a neighboring class set are grouped using their Euclidean distances from each other. For example, we have a database containing large number of access records of a location-based service. Records of the objects may consist of "requested service name," "number of packet transmitted" in addition to x and y coordinate values indicating where the request came from. The algorithm presented here efficiently finds sets of "service names" that were frequently close to each other in the spatial database. For example, it may find a frequent neighboring class set, where "ticket" and "timetable" are frequently requested close to each other. By recognizing this, location-based service providers can promote a "ticket" service for customers who access the "timetable."

View full abstract

We explain automatic recommendation techniques by which computer systems can automatically select goods or documents preferred by each user based on his/her buying or accessing history. Especially, we focus on the collaborative filtering method using weighted majority prediction algorithm developed by the authors. From both theoretical and experimental points of view, we explain how excellent the method is compared to the method using correlation coefficient, which is the most popular collaborative filtering method. We also address issues that we should consider in order to make our method practically useful.

View full abstract