Subtitles and closed captions, which are prepared for hearing-impaired users, are now widely used by users without hearing concerns. In this paper, we focus on the adaptation of subtitles and captions for non-hearing-impaired users, particularly the adaptation of the kanji ruby. From our experiments on non-hearing-impaired adults, Welch’s t-test was used to clarify whether listening to audio with the same content affects the necessity of kanji ruby. In addition, we proposed and evaluated an adaptive model to predict whether ruby should be added to kanji captions based on the experimental results. The experimental results suggest that not only the difficulty of the kanji and the user’s kanji ability, but also the content of the audio is important for the optimization of kanji ruby.

Interest in emotion recognition in conversations (ERC) has been increasing in various fields, because it can be used to analyze user behaviors and detect fake news.Many recent ERC methods use graph-based neural networks to take the relationships between the utterances of the speakers into account.In particular, the state-of-the-art method considers self-and inter-speaker dependencies in conversations by using relational graph attention networks (RGAT).However, graph-based neural networks do not take sequential information into account.In this paper, we propose relational position encodings that provide RGAT with sequential information reflecting the relational graph structure.Accordingly, our RGAT model can capture both the speaker dependency and the sequential information.Experiments on four ERC datasets show that our model is beneficial to recognizing emotions expressed in conversations.In addition, our approach empirically outperforms the state-ofthe-art on all of the benchmark datasets.

Taro Miyazaki, Kiminobu Makino, Yuka Takei, Hiroki Okamoto, Jun Goto. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

MOTIVATION: Long expansions of short tandem repeats (STRs), i.e. DNA repeats of 2-6 nt, are associated with some genetic diseases. Cost-efficient high-throughput sequencing can quickly produce billions of short reads that would be useful for uncovering disease-associated STRs. However, enumerating STRs in short reads remains largely unexplored because of the difficulty in elucidating STRs much longer than 100 bp, the typical length of short reads. RESULTS: We propose ab initio procedures for sensing and locating long STRs promptly by using the frequency distribution of all STRs and paired-end read information. We validated the reproducibility of this method using biological replicates and used it to locate an STR associated with a brain disease (SCA31). Subsequently, we sequenced this STR site in 11 SCA31 samples using SMRT(TM) sequencing (Pacific Biosciences), determined 2.3-3.1 kb sequences at nucleotide resolution and revealed that (TGGAA)- and (TAAAATAGAA)-repeat expansions determined the instability of the repeat expansions associated with SCA31. Our method could also identify common STRs, (AAAG)- and (AAAAG)-repeat expansions, which are remarkably expanded at four positions in an SCA31 sample. This is the first proposed method for rapidly finding disease-associated long STRs in personal genomes using hybrid sequencing of short and long reads. AVAILABILITY AND IMPLEMENTATION: Our TRhist software is available at http://trhist.gi.k.u-tokyo.ac.jp/. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Infantile unhealthy lifestyle increases the risk of their adult diseases and cost of medical care in their future. If children life-log is measured and their physical condition can be analyzed, it is expected that such information can be utilized for prevention of adult diseases and for the health-promoting. Therefore, activity recognition for children is treated in the present paper. Feature points for activity recognition are computed from accelerations which are measured by a 3-axis acceleration sensor attached to child's upper-arm. Two-phased classification was performed by using Self-Organizing Map. As a result, it was confirmed that the classification accuracy was 65.92[%].

This paper proposed the robot motion planning which considers an utterance timing. Motion and utterance timing of autonomous robot is generated by using the SOM which was trained by human normative data in practical communication. Effective and ineffective behavior of the robot is generated, and an availability of behavior of the robot is evaluated by experiment. As a result, although effective and ineffective behavior of automatically robot are realized, that is, the influence of individual differences is appeared in human-robot communication.

On-demand services for TV program, which provide users with past programs on demand, are becoming popular. It is therefore necessary to find a means of efficiently searching for programs that users want to view, from huge program archives. This paper proposes an automatic method of retrieving programs related to the one being viewed by the user. To that end, we compute similarity between program summaries and closed captions obtained from broadcasting by weighting significant words such as compound words and named entities. Additionally our method provides inter-program relationship labels to indicate why the results of relevant programs were chosen. The results of an evaluation showed that the method recommended relevant programs with higher accuracy than baseline methods and indicated appropriate relationship labels for related programs.

BACKGROUND: During this recent decade, microarray-based single nucleotide polymorphism (SNP) data are becoming more widely used as markers for linkage analysis in the identification of loci for disease-associated genes. Although microarray-based SNP analyses have markedly reduced genotyping time and cost compared with microsatellite-based analyses, applying these enormous data to linkage analysis programs is a time-consuming step, thus, necessitating a high-throughput platform. RESULTS: We have developed SNP HiTLink (SNP High Throughput Linkage analysis system). In this system, SNP chip data of the Affymetrix Mapping 100 k/500 k array set and Genome-Wide Human SNP array 5.0/6.0 can be directly imported and passed to parametric or model-free linkage analysis programs; MLINK, Superlink, Merlin and Allegro. Various marker-selecting functions are implemented to avoid the effect of typing-error data, markers in linkage equilibrium or to select informative data. CONCLUSION: The results using the 100 k SNP dataset were comparable or even superior to those obtained from analyses using microsatellite markers in terms of LOD scores obtained. General personal computers are sufficient to execute the process, as runtime for whole-genome analysis was less than a few hours. This system can be widely applied to linkage analysis using microarray-based SNP data and with which one can expect high-throughput and reliable linkage analysis.

Aiming to close the digital divide in the television viewing environment, we are developing a TV system with an agent that controls the TV and peripherals on behalf of the user and provides information to the user. We propose a TV system function that answers viewers' questions about TV programs by calling upon multiple question-answering agents that search for relevant information.

This paper describes the practical application of robot vision techniques to electricity distribution work. To cope with variable outdoor lighting conditions and diverse backgrounds due to unstructured outdoor environments, we have developed two methods of detecting an object's 3D position using both 2D gray-scale image and 3D range image. The first method locates a target object as a 2D image taken at a distance. Although the resolution of a 3D range image taken from the same position is too low to extract the object's surface geometry, it is effective to eliminate the background section from the 2D image based on the distance threshold. The second method generates an online template image of the object to be worked on. This enables reliable real-time visual tracking by stereovision despite the variations of the object's shape and appearance. Here we show how these methods are implemented and effectively used in distribution work.

This paper focuses on the automation of insertion work, such as inserting a metal sleeve into a wire terminal, necessary in distribution work. The major technical issues for this task automation are considered as: the detection of a target object in an unstructured, complex outdoor environment; and the insertion executed at the work site with restricted clearance. In the former, the target object is detected by applying 3D image processing and template matching techniques, based on the position and posture relationship between CCD cameras, the target object, and the background. In addition, the manipulator's targeting position is updated by commands from the visual sensor system, applying the video-rate stereo camera method. This allows for realtime visual feedback control; thus, misalignment due to deviation arising from manipulator's mechanism is avoided. In the latter issue, a new algorithm was developed for insertion. This algorithm can estimate contact force, based on the deviation between the present position and command position of the manipulator. An experiment using a testing robotic system revealed that the proposed robotic technology was effective for performing insertion work without assistance by humans.