土方 嘉徳

情報推薦の一手法として協調フィルタリング（CF）がよく知られているが，このCFの中でもMatrix Factorization（MF）は，評価値の欠損が多い現実のデータセットに対して優れた結果を出していることで有名である．ただし，MFは評価値の欠損に強いものの，評価値が極端に少ない場合には適用することが困難である．この問題に対して，評価値だけでなくテキスト情報も利用したCollaborative Topic Regression（CTR）というモデルが近年注目されている．CTRでは，トピックを抽出する情報源として，アイテムに関するテキスト情報を用いている．しかし，ユーザに関するテキスト情報を用いた場合の評価は行われてこなかった．本研究ではCTRにおいて，アイテムに関するテキスト情報を用いた場合と，ユーザに関するテキスト情報を用いた場合で，推薦の性能にどれだけの違いがあるのかを調査する．この両者とベースラインであるMFに対して推薦結果の正確性（再現率）と利便性（被覆率・多様性）に関して定量的な比較を行う．実験の結果，ユーザに関するテキスト情報を用いた場合は正確性の高い推薦が実現でき，アイテムに関するテキスト情報を用いた場合は利便性の高い推薦が実現できることが分かった．Collaborative filtering is one of the popular methods for selecting contents or items suited to user's interest or preference from a content (item) set. Among the existing methods of CF, Matrix Factorization (MF) is superior for the dataset with much missing data. However, even MF cannot perform well when the number of users' ratings are extremely small. To solve this problem, an approach that uses not only users' rating values but also text information of items are becoming popular recently. We focus on one major model of this approach, Collaborative Topic Regression (CTR). In CTR, text information of items is used as information source to extract topics. However, the case using text information of users has not been evaluated. In this study, we compare the case using text information of items and the case using text information of users in CTR to know the performance difference. We compare the both cases and MF (which is a baseline) according to accuracy metrics and usefulness metrics. We found that using text information of users realizes the recommendation with high accuracy and using text information of items realizes the recommendation with high usefulness from the experiment.

<p>Maintaining one's personal space is quite important in leading a comfortable social life. However, it is difficult to maintain an appropriate interpersonal distance all the time. If we can control the visual distance to reduce the discomfort, the problem would be solved. To confirm the effectiveness of this solution, we made an interpersonal distance control system with a video see-through system, consisting of a head-mounted display (HMD), depth sensor, and RGB camera. The system controls the interpersonal distance by changing the size of the person in the HMD view. We conducted an experiment to confirm the capability of the system and found that the system can reduce the discomfort because of the inappropriate interpersonal distance by controlling the visual interpersonal distance.</p>

<p>In recent years, research on the development of cloud databases for sharing robot motion data has been conducted. In this work, we propose a search framework for this kind of databases. This framework has been designed for end users that do not have any knowledge on robotics. It is composed of an specific input format (search template) and a search algorithm. To verify the effectiveness of the proposed framework experiments were carried out, where end users search for a given motion and evaluated the ease of use of the search template.</p>