Hiroki Asai

Microstructural study of electroless-plated CoNiP ternary alloy films for perpendicular magnetic recording media

The microstructure of electroless CoNiP films with various H/sub c/( perpendicular to ) values for use as perpendicular magnetic recording media was investigated. It was observed that the H/sub c/( perpendicular to ) values of the films varied according to their microstructure. The film which showed the highest H/sub c/( perpendicular to ) value contained large columnar grains with high crystallinity, whereas such a clear columnar structure was not observed in the films with lower H/sub c/( perpendicular to ) values. The shape anisotropy due to the columnar structure effectively contributes to the magnetic properties of the films as well as magnetocrystalline anisotropy associated with preferred crystallographic orientation. The results of compositional analysis of ultra-small regions suggested that the formation of low M/sub s/ and low H/sub c/ regions at the initial deposition stage was due to a preferential deposition of Ni. >

Detecting student frustration based on handwriting behavior

Detecting states of frustration among students engaged in learning activities is critical to the success of teaching assistance tools. We examine the relationship between a students pen activity and his/her state of frustration while solving handwritten problems. Based on a user study involving mathematics problems, we found that our detection method was able to detect student frustration with a precision of 87% and a recall of 90%. We also identified several particularly discriminative features, including writing stroke number, erased stroke number, pen activity time, and air stroke speed.

A challenge of authorship identification for ten-thousand-scale microblog users

Internet security issues require authorship identification for all kinds of internet contents; however, authorship identification for microblog users is much harder than other documents because microblog texts are too short. Moreover, when the number of candidates becomes large, i.e., big data, it will take long time to identify. Our proposed method solves these problems. The experimental results show that our method successfully identifies the authorship with 53.2% of precision out of 10,000 microblog users in the almost half execution time of previous method.

Detecting Learner's To-Be-Forgotten Items using Online Handwritten Data

An effective learning system is indispensable for human beings with a limited life span. Traditional learning systems schedule repetition based on both the results of a recall test and learning theories such as the spacing effect. However, there is room for improvement from the perspective of remembrance-level estimation. In this paper, we focus on on-line handwritten data obtained from handwriting using a computer. We collected handwritten data from remembrance tests to both analyze the problem of traditional estimation methods and to build a new estimation model using handwritten data as the input data. The evaluation found that our proposed model can output a continuous remembrance-level value of zero to 1, whereas traditional methods output a only binary decision. In addition, the experiment showed that our proposed model achieves the best performance with an F-value of 0.69.

Predicting various types of user attributes in Twitter by using personalized pagerank

Predicting various types of user-attributes in social networks has become indispensable for personalizing applications since there are many non-disclosed attributes in social networks. However, extracted attributes in existing works are limited to pre-defined types of attributes, which results in no extraction of unexpected-types of attributes. In this paper, we therefore propose a novel method that extracts various, i.e., unlimited, types of attributes by adopting personalized PageRank to a large social network. The experimental results using over 7.9 million of Japanese Twitter-users show that our proposed method successfully extracts four types of attributes per-user in average with 0.841 of MAP@20.

EA Snippets: Generating Summarized View of Handwritten Documents Based on Emphasis Annotations

Owing to the recent development of handwriting input devices such as tablets and digital pens, digital notebooks have become an alternative to traditional paper-based notebooks. Digital notebooks are available for various device types. To display a list of text documents on a device screen, we often use scaled thumbnails or text snippets summarized through natural language processing or structural analyses. However, these are ineffective in conveying summaries of handwritten documents, because informal and unstructured handwritten data are difficult to summarize using traditional methods. We therefore propose the use of emphasis-based snippets, i.e., summarized handwritten documents based on natural emphasis annotations, such as underlines and enclosures. Our proposed method places emphasized words into thumbnails or text snippets. User studies showed that the proposed method is effective for keyword-based navigation.