Kenro Aihara

DVHMM: variable length text recognition error model

This paper proposes a text recognition error model called the dual variable length output hidden Markov model (DVHMM) and gives a parameter estimation algorithm based on the EM algorithm. Although existing probabilistic error models are limited to substitution (1, 1), insertion (1, 0), and deletion (0, 1) errors, the DVHMM can handle error patterns of any pair (i, j) of lengths including substitution, insertion, and deletion.

Enhancing creativity through reorganising mental space concealed in a research notes stack

In this article, we propose a method and its implementation to aid the process of creative thinking. The objective of this research is to enhance human scientific creativity with computers. In order to aid the process of creative thinking, we have implemented a system named En Passant 2, which stores the users research notes and gives the triggers to recall his/her memories in current context. En Passant 2 has a function to deal with indices and a time attribute of the users thought explicitly, and shows the users notepads related to his/her awareness of the issues.

Quality enhancement in information extraction from scanned documents

When constructing a large document archive, an important element is the digitizing of printed documents. Although various techniques for document image analysis such as Optical Character Recognition (OCR) have been developed, error handling is required in constructing real document archive systems. This paper discusses the problem from the quality enhancement perspective and proposes a robust reference extraction method for academic articles scanned with OCR mark-up. We applied the proposed method to articles appearing in various journals, and these experiments showed that the proposed method achieved a recognition accuracy of more than 94%. This paper also discusses manual correction and investigates experimentally the relationship between extraction accuracy and cost reduction.

Owlery: a flexible content management system for “growing metadata” of cultural heritage objects and its educational use in the CEAX project

With the Educational use of Cultural heritage Archives and Cross(X) search (CEAX), we have investigated how to establish a framework for managing various kinds of information on cultural heritage objects and how to utilize them for educational purposes. To achieve this goal, we propose a conceptual framework in this paper called “Growing Metadata” and a flexible content management system called Owlery. Growing Metadata includes not only factual descriptions of objects but also various annotations about the objects, such as metadata for children, course materials prepared by school teachers, classroom reports, etc., and are reusable for search and educational purposes. Owlery is a software platform to create, share, utilize and reuse the Growing Metadata, and in which various metadata and annotations are managed in different levels of authenticity, authorship, and user groups. As a result of the experimental classes for 89 6th-grade children, our framework was found to be efficient and accepted by the content creators, like museum experts, content annotators and shool teachers.

Adaptive Replication Method Based on Peer Behavior Pattern in Unstructured Peer-to-Peer Systems

In this paper, we propose replication methods to perform effective document sharing on a peer-to-peer(P2P) system where peers frequently join and leave the system. The proposed method uses the relevancy and usefulness of peers to determine how many replications should be made, and where to locate these replications. This paper shows empirically that the proposed method improves the efficiency of query processing.

Crowdsourced Mobile Sensing for Smarter City Life

This paper introduces the ongoing project that aims to develop a mobile sensing framework to collect sensor data reflecting personal-scale, or microscopic, roadside phenomena by crowd sourcing and also using social big data, such as traffic, climate, and contents of social network services like Twitter. To collect them, smartphone applications are provided. One of the typical applications is a driving recorder that collects not only sensor data, such as acceleration, gyro, compass, and speed, but also recorded videos from the drivers view. To extract specific roadside phenomena, collected data are integrated and analyzed at the service platform. The framework also provides tools for interactive analysis to support city administration.

Information extraction from scanned documents by stochastic page layout analysis

We propose a stochastic context-free grammar for extracting information from scanned document images. The grammar is designed to disambiguate layout analysis and utilize both layout and text features. We applied this grammar to the problem of extracting bibliographic information from scanned academic papers and found that it can accurately extract information.

The Dynamic Construction of Knowledge-Based Systems

Publisher Summary This chapter discusses the reasons for consideration of dynamism for knowledge-based systems while also discussing the methods to construct such systems. Human activities, excluding routine work, have some dynamic aspect. A human–machine environment that supports such activities should correspond to the dynamism. This chapter emphasizes especially the dynamism of knowledge and our mental world. The knowledge one uses in the real world contains an element—called “tacit knowledge” in many cases—that is not clearly explained even by domain experts. This is because of the nature of knowledge as well as the lack of a clear world mental model on which the knowledge is based. This chapter attempts to construct an environment where the nebulous mental world is clarified and new ideas can be found. It proposes two different methodologies, where each example corresponds to a type of activity. Although each methodology is discussed separately, they are complementary to each other.

Do Strollers in Town Needs Recommendation?: On Preferences of Recommender in Location-Based Services

When we discuss about recommendation especially in Location-Based Services LBS, we need to reveal whether users really want recommendations or not in fact while they are strolling in town, prior to evaluate each recommendation model. In this paper, a Location-Based Service, called nicotoco, is shown. nicotoco is an iPhone-based LBS in Futako-tamagawa area, Tokyo, Japan and provides information about stores and events to users. In the experiment using nicotoco, recommendations may be preferred more than rankings which was made from access counts.

Traffic Surveillance System for Bridge Vibration Analysis

The vibration response of a damaged bridge is known to have changed characteristics. To analyze the response, we start by collecting waveforms of the vibration immediately following the passage of a vehicle. We then need to isolate just those vibrations caused by a single heavy vehicle, if the vibration characteristics are to be accurate. In this paper, we propose a traffic-vibration analysis system that interacts with a surveillance camera. The system identifies a vehicle from the video by combining a moving-object detector with a neural-network-based object detector, thereby estimating automatically the bridges natural frequencies and damping ratios as features that characterize the bridges damage.