We propose a system to make user’s ideas more diverse as a breakthrough during creating ideas. Discussions such as brainstorming are held to stimulate user’s idea, though such composition requires much effort. For solving the problem, the proposed system presents cross-category articles related to the user’s idea as the divergence support. When the users choose one of them and the keywords shown in the articles, the sentence prototype are generated with the keywords. Afterward, the sentence prototype or its revision is input as a query of searching for the analogy-recalling articles. By iterating the processes, the user’s idea goes diversely. For obtaining the cross-category articles related to the user’s idea, the article texts and user’s idea text is input to GPT2 models to convert to vectors. The category-dependent components extracted with correlation of the category codes of the article is subtracted from the vectors to generate the category independent vectors. The cross-category but similar articles are searched based on the category independent vectors. And also, the experimental evaluation results are shown in this paper.

Molecular fingerprints are widely used for predicting chemical properties, and selecting appropriate fingerprints is important. We generate new fingerprints based on the assumption that a performance of prediction using a more effective fingerprint is better. We generate effective interaction fingerprints that are the product of multiple base fingerprints. It is difficult to evaluate all combinations of interaction fingerprints because of computational limitations. Against this problem, we transform a problem of searching more effective interaction fingerprints into a quadratic unconstrained binary optimization problem. In this study, we found effective interaction fingerprints using QM9 dataset.

We present a practical people-flow simulation method for evaluating facility layout plans; and demonstrate its effectiveness in experiments using real people-flow data. To improve the robustness of our previous machine-learning-based method, we use a single model trained from entire trajectory data, considering the negative effect of dividing up the training data. We also introduce feature values that represent positional relations between a pedestrian's destination and obstacles. To evaluate the performance of our method, people-flow data were obtained by using LiDAR sensors in an office before and after its layout was changed. In the experiments, we trained the prediction model from only the data before the layout change and found that the people-flow simulation was accurate even after the layout change. We also confirmed that our method outperformed the existing ones. As a result, it was confirmed that our people-flow simulation method can be used to evaluate facility layout plans.

Simulating the movement of pedestrians is a challenging problem and becoming increasingly important in a variety of applications, such as determining potential safety hazards, evaluating operational performance, layout planning in public and commercial facilities. We propose a pedestrian-path prediction method for people-flow simulation, which takes into account the following two components: 1) dividing measured trajectory data into partial trajectories with each origin-destination pattern, and 2) training destination-wise models for pedestrian-path prediction with relative features based on velocities and distances We also discuss the path-prediction rate and people-flow simulation results with the proposed method using measured trajectory data in a check-in lobby at an airport.

Applications utilizing many types of location data, such as traffic congestion estimation and facility management using indoor pedestrian tracks, have been rapidly increasing. Such applications require the integration of various locational data from different data sources to produce more values. Efforts to ensure smoother data exchange are required for promoting the use of such applications because handling and integrating location data will enlarge the market for geo-spatial information. In response to this need, we had proposed a data encoding standard called `OGC®Moving Features' to contribute to smoother data exchange, and it was adopted as an international standard on Feb. 2015. We demonstrate in this work that OGC®Moving Features is an effective tool to advance technologies for applications using many types of location data, with referring "4Vs" to represent the most pressing bigdata issues.

For fast disaster estimation after a large-scale disaster occurs, this paper presents a fast spatio-temporal similarity search method that searches a database storing many scenarios of disaster simulation results represented by time-series grid data for some scenarios similar to insufficient observed data sent from sensors. The proposed method efficiently processes spatio-temporal intersection by using a spatiotemporal index to reduce the processing time for the spatiotemporal similarity search. Additionally, this paper presents the efficient spatio-temporal range search method by using this spatio-temporal index. The spatio-temporal range search is needed for the analysis and visualization in order to grasp a damage situation after spatio-temporal similarity search returns some scenarios similar to observed data. The results of the performance evaluation show that the proposed method has a shorter response time for the spatiotemporal similarity search than two conventional methods that use a temporal index and a spatial index. They also show that the response time is within about 30 seconds when the proposed method searches the database storing 50 billion time-series grid data items for some scenarios similar to 100 observed data items. As a result, the proposed method can be applied to a real environment in which a spatio-temporal similarity search needs to processed within 10 minutes. Additionally, the evaluation results show that the spatio-temporal range search method by using the spatio-temporal index can be also applied to a real environment.

A highly practical use case of pedestrian-track analysis by using LiDAR is presented in this paper. Many problems are caused by heavy crowdedness in the management of public facilities, i.e., shopping malls, airports, and so on. One solution is crowdedness equalization by controlling pedestrian flow. We conducted two experimental demonstrations at technical exhibitions to find that factors that determine pedestrian flow. Pedestrian tracks were obtained at an exhibition in 2013 by using a LiDAR-based pedestrian-tracking system first. As a result, new knowledge was gained; the layout of a technical exhibition should be designed to bend the path of a pedestrian flow toward areas where their attention is desired to be. The layout of an exhibition in 2014 was designed to bend the pedestrian path many times so that pedestrians' attention was located diversely. Therefore, pedestrian tracks were successfully obtained; as a result, it was confirmed that crowdedness was successfully equalized.

A "mixed autoregressive hidden Markov model" (MAR-HMM) is proposed for modeling people's movements. MAR-HMM is equivalent to a special case of an autoregressive hidden Markov model (AR-HMM), which takes into account changes of people's internal properties. The number of parameters is thus reduced in the case of MAR-HMM. A dataset is applied to evaluate MAR-HMM in this study. The prediction rate of MAR-HMM is 56.8% and that of AR-HMM is 51.5%. It is therefore concluded that MAR-HMM is applicable to trajectory analysis of pedestrians.

A method for predicting pedestrian movement on the basis of a mixed Markov-chain model (MMM) is proposed. MMM takes into account a pedestrian's personality as an unobservable parameter. It also takes into account the effects of the pedestrian's previous status. A promotional experiment in a major shopping mall demonstrated that the highest prediction accuracy of the MMM method is 74.4%. In comparison with methods based on a Markov-chain model (MM) and a hidden-Markov model (HMM) (i.e., prediction rates of about 45% and 2%, respectively), the proposed MMM-based prediction method is substantially more accurate. This pedestrian-movement prediction based on MMM using tracking data will make it possible to provide so-called "adaptive mobile services" with proactive functions.

In this paper, we discuss statistical analysis of human trajectories measured by GPS-like positioning devices. Our goal is to develop a system of trajectory analysis that distributes information optimized for each user. For such a system, we need a method to estimate a user's status from his/her trajectories. First, a trajectory needs to be divided into short temporal segments, which will be matched to action model patterns, to estimate a user's status. Second, we tried dividing actual human trajectories using a conventional trajectory-clustering method. Moreover, we adjusted parameters of the trajectory clustering by using information criteria experimentally. After the experiment, we confirmed that only a criterion in which noise data are counted worked well. However, we also confirmed that the number of clusters generated by the method is too small. Therefore, we conclude that an improvement in deciding which data are noise in trajectory clustering is necessary for estimating the status of users.

The need for traffic information service systems that warn about traffic congestion and accidents through mobile devices, such as cellular phones, is increasing greatly now. To display clearly these informations for these services, a well-formed map is more useful than a detailed one in many cases. Although methods to generate well-formed map automatically are known as map summarization methods, summarization of road maps that cover wide areas has not been studied sufficiently. Therefore, we developed a new algorithm that summarizes the network structure of a detailed road network into a simple macroscopic structure. Further, we evaluated this algorithm’s effectiveness by experiments with 40 samples of road routes from the main road networks in Japan. The proposed method reduced the link number by 41.3% and reduced the fractal indices by 62.6%, and this proved simplification. And the proposed method improved the well-formed map by user evaluation. From these results obtained by our evaluation, we found that our method is effective for constructing a system for providing practical mobile traffic information services.