岡田 龍太郎

Through technology, it is essential to seamlessly bridge the divide between diverse speaking communities (including the signer (the sign language speaker) community). In order to realize communication that successfully conveys emotions, it is necessary to recognize not only verbal information but also non-verbal information. In the case of signers, there are two main types of behavior: verbal behavior and emotional behavior. This paper presents a sign language recognition method by similarity measure with emotional expression specific to signers. We focus on recognizing the sign language conveying verbal information itself and on recognizing emotional expression. Our method recognizes sign language by time-series similarity measure on a small amount of model data, and at the same time, recognizes emotion expression specific to signers. Our method extracts time-series features of the body, arms, and hands from sign language videos and recognizes them by measuring the similarity of the time-series features. In addition, it recognizes the emotional expressions specific to signers from the time-series features of their faces.

In this paper, we present an emotion estimation method using heart rate variability parameters of vital data. Recently, as sensors have become more precise and smaller, it has been possible to obtain users' vital data in real-time quickly. In our method, ECG (electrocardiogram) data are measured beforehand while listening to a story with voice narration that evokes emotions and based on the trends obtained through the measurement, the emotions that have a high correlation with the newly acquired ECG data are estimated to be the emotions expressed in the ECG data. With the implementation of our method, it is possible to estimate the user's emotions based on ECG data. In this paper, we also represent the application of our method to chat icons that see users' emotions in real-time. By realizing this application, users will see the changes in their emotions and control their mental health.

In general, when a place depicted in a novel exists, users may like to visit the site that appears in the story. There is a need to efficiently visit tourist spots in or around the places depicted in the novel when visiting them. When we realize to retrieve appropriate tourist spots in areas in stories, it is possible to increase the number of new sightseeing opportunities for users. This paper presents a tourist spot recommendation method corresponding to place names appearing in novel contents. Our system recommends some actual tourist spots corresponding to a novel selected by a user. The feature of our proposed method is to recommend tourist spots within the range that users can visit in a day, based on the place names that appear in the novel. We apply this method to realize seamless linking media between the creative world like a novel and the real-world.

In this paper, we present a method for discovering of historical transition in aesthetic notions of waka poetry over by using changes in co-occurrence words. The structure and the words used in waka poetry change as time passes. By analyzing the chronological changes in the structures and words of waka poetry, we can clarify the historical changes in the aesthetic notions of Japanese people. In this paper, we focus on the three major anthologies of Japanese poetry, Manyoshu, Kokin Wakashu and Shin Kokin Wakashu, and on Kago, which is the central elements in Japanese poetry. Kago is a word and expression often used in Japanese poetry. In our method, we derive differences in the frequency of occurrence of Kago categorized according to their meanings in each collection and extract important co-occurrences of Kago as context words. By our method, we clarify the differences in the context words of Kago and show the historical transition of aesthetic notions.

In this paper, we present gaze-music transformation realization method by realizing gaze emotion detection system and interconnecting the detection system and an automatic music media creation system. This method automatically creates a music media contents by a detected emotion from human gaze. When it is possible to realize a system that acquires human gaze data from a camera device, detects emotions from the data, and creates media content in response to those emotions, it will contribute to creative activities corresponding to human emotion. Our method connects a gaze emotion detection system and an automatic music creation system by the similarity of words that represent emotions and impressions. It is possible to compare emotions by comparing the emotions expressed by each different vocabulary using the similarity of words.

In this paper, we present a dialogue structure analysis method to visualize the transition of topics in a meeting as the one of dialogue process representation. Our method extracts topics in a meeting on time series. In addition, we define an index to assess the importance of the whole meeting in each phase. By this index, we can represent important phases in the meeting. In organizations such as companies, it is important to improve the efficiency of a meeting, because the meeting time occupies a large proportion in business hours. We should analyze contents and flows of remarks in dialogue on meetings in order to improve efficiency of a meeting. Generally, improving the efficiency of a meeting is improving the form of a meeting, such as pre-sharing of documents, keeping time, clarification of roles of members, and appointing a facilitator. Our method provides the one of the visualization for the flow of remarks in dialogue in a meeting. In this paper, we also represent some preliminary experiment by using text data for actual meetings.

本稿では，メディアコンテンツを対象とした統計的一般化逆作用素の構成方式を示す．これは，与えられた印象を表す単語とその重みで表される印象メタデータからメディアコンテンツを生成する機構を実現するものである．本方式は，従来，我々が提案してきた手法であるメディアコンテンツから印象を表す言葉をメタデータとして抽出する，メディアコンテンツを対象としたメタデータ自動抽出方式手法の逆演算として構成される．しかしながら，本逆演算においては一般的に，不良設定問題が発生する．この不良設定問題を解決するために，制約条件としてそのメディアの種類に応じた統計情報やそのメディアを対象とした研究成果・理論を用いる．これにより，メディアコンテンツと言葉の間の相互変換を可能とする．さらに，本方式を楽曲メディアコンテンツに適用し，印象語で表現された印象メタデータから楽曲メディアコンテンツを生成するシステムを実装する．これを用いて実験を行い，提案方式が実現されていることを示す．In this paper, we present a construction method of Stochastic generalized inverse operator for media contents. This method realizes automatic media contents creation depending on impression words as an inverse operation of our automatic metadata extraction method which we have proposed. This is an inverse operator which extracts words from media contents. However, this creation mechanism contains ill-posed problems. In order to solve them, we introduce stochastic data about the media and it's studies. Moreover, we construct a mutual conversion between impression words and media contents. Furthermore, we apply our method to music data creation system. We performed verification experiments, and showed the effectiveness of our method.

In this paper, we propose a method of image feature selection for integration of image classification by Bag-of-Keypoints method and efficient search method. Our method is integration of image classification, which provides high-precision classification for various images. Therefore, in order to select a type of image feature in accordance with various purposes, it is necessary to select method easily. In addition, our method integrates various kinds of image feature by Bag-of-Keypoints method. Classification with a combination of some detectors and descriptors is more effective than with single detector and descriptor. To realize combination of some detectors and descriptors, we propose an integrating method. This paper is LATE BREAKING PAPERS for CSCI-ISAI.