The betweenness centrality is a widely used property to identify important nodes in social networks. Several algorithms have been studied to efficiently compute the top-k nodes with the highest betweenness centrality on a graph where all the data is available. However, all the graph data of real social networks are not typically available to third parties such as researchers or marketers, and hence, an estimation algorithm based on sampling the graph data is required. Accurately estimating the top-k nodes with the highest betweenness centrality from a small sample of a graph is a challenging task. First, the top-k nodes need to be included in the small sample. Second, nodes with the high betweenness centrality that is defined on the whole graph need to be accurately identified from the small sample. We propose a random walk-based algorithm to estimate the top-k nodes with the highest betweenness centrality by utilizing the ego betweenness centrality that has a high correlation with the betweenness centrality in social networks. The proposed algorithm firstly obtains a small sample that includes many of top-k nodes with the highest betweenness centrality via a random walk on a social network. Then, we obtain unbiased estimates of the ego betweenness centrality of sampled nodes and approximate the top-k nodes with the highest betweenness centrality as the top-k nodes with the highest estimated ego betweenness centrality. The proposed estimator efficiently estimates the ego betweenness centrality of each sample without additionally sampling the graph data by utilizing the neighbor data of the previous and the next samples. The experiments using real social network datasets show that the proposed algorithm estimates more accurately the top-k nodes with the highest betweenness centrality than existing algorithms when the sample size is small.

We consider the problem of community detection. Although network embedding and representation learning methods are recently getting popular, we claim that they fall into suboptimal solutions for community detection, because they are based on indirect approach, which requires to apply clustering methods such as k-means to the embedding/representation vectors. We present PPNMF, proximity preserving nonnegative matrix factorization for community detection. The idea of PPNMF is three-fold. 1) PPNMF is based on direct approach: it directly minimizes its loss function for community detection. 2) Users can control the importance of observed edges over unobserved edges. 3) PPNMF can precisely capture the effects of the first-order and second-order proximities of vertexes to communities. Also, PPNMF employs the Adamic Adar index as the second-order proximity. The experiments validate that PPNMF performs better or comparable to existing methods in various real datasets for the tasks of community detection.

Given a bi-type information network, which is an extended model of well-known bipartite graphs, how can clusters be efficiently found in graphs? Graph clustering is now a fundamental tool to understand overviews from graph-structured data. The RankClus framework accurately performs clustering for bi-type information networks using ranking-based graph clustering techniques. It integrates graph ranking algorithms such as PageRank or HITS into graph clustering procedures to improve the clustering quality. However, this integration incurs a high computational cost to handle large bi-type information networks since RankClus repeatedly computes the ranking algorithm for all nodes and edges until the clustering procedure converges. To overcome this runtime limitation, we present a novel RankClus algorithm that reduces the running time for large bi-type information networks. Our proposed method employs dynamic graph processing techniques into the iterative ranking procedures included in RankClus. By dynamically updating ranking results, our proposal reduces the number of computed nodes and edges during repeated ranking procedures. For further improving the efficiency, we also present a parallel implementation of our proposed algorithm by using thread-based parallelization on a modern manycore CPU. We experimentally verify using real-world datasets that our fast and parallel proposed methods successfully reduces the running time while maintaining the clustering quality of RankClus.

Elliptic curve cryptosystems (ECCs) are widely used because of their short key size. ECCs can ensure sufficient security with shorter keys, using less memory to reduce parameters. Hence, ECCs are typically used in IoT devices. The dominant computation of an ECC is scalar multiplication $Q = kP$ for $P \in E(\mathbb{F}_q)$. Thus, the security and efficiency of scalar multiplication are paramount. To render secure ECCs, complete addition (CA) formulae can be employed for secure scalar multiplication algorithms. However, this requires significant memory; thus, it is not suitable for compact devices. Several types of coordinates exist for elliptic curves such as affine, Jacobian, Projective and so on. The CA formulae are not based on affine coordinates and, thus, require considerable memory. In this study, we achieve a compact ECC by focusing on affine coordinates. In fact, affine coordinates are highly advantageous in terms of memory but require many if statements for scalar multiplication owing to exceptional points. We propose two scalar multiplication algorithms with the extended affine formulae to delete some exceptional inputs for scalar multiplication. Our two algorithms reduce memory cost up to 37% or 21%. In many cases such as NIST elliptic curves, our two algorithms are the most efficient if $\frac{I}{M}<12$, for the ratio of computational cost of inversion and multiplication. The experiment shows that our algorithms can compute the elliptic curve scalar multiplication correctly and efficiently.

The RAM encryption encrypts the data on memory to prevent data leakage from an adversary to eavesdrop the memory space of the target program. In this paper, we discuss the feasibility of software based RAM encryption and clarify that it is possible to be secure against so-called semi-honest adversaries under some additional and reasonable constraints. In addition, we tentatively embed our RAM encryption mechanism to SHA-256 hash function. The performance evaluation results are also reported in this paper.

The Internet of Things (IoT) malware keep evolving and utilize multiple vulnerabilities to infect IoT devices. Besides malware, human attackers also utilize various tools to access and collect variable information on the device. For instances, web UI of IP Cameras and routers are constantly searched and accessed if vulnerable. In order to observe and analyze such a variety of attacks in depth, there is an increasing need for bare-metal IoT devices as a honeypot, since it is costly to emulate device-specific vulnerabilities and complex functionalities from dedicated services. However, operating bare-metal IoT honeypots has unique technical challenges mostly coming from their low configurability as an embedded system. A bare-metal honeypot needs proper access control while it is allowing attackers to access its inside to some degree, such as filter out bricking commands and changes of critical configuration. From this observation, we propose ThingGate, a gateway for flexible operation of bare-metal IoT honeypot. ThingGate employs a man-in-the-middle proxy to control and manage inbound and outbound traffic of the bare-metal IoT honeypot. Moreover, it adds the functionality of web tracking, which is not provided by the web UI of the original devices. We evaluate ThingGate with seven bare-metal IoT devices and show that it successfully blocks unwanted incoming attacks, masks wireless access point information of the devices, and tracks attackers on the device web UI while showing high observability of various attacks exploiting different vulnerabilities.

Targeted email attacks are one of the main threats to organizations of all sizes and fields. In targeted email attacks, malicious VBA (Visual Basic for Applications) macros are often embedded into the attachment files to compromise the target computers. These malicious VBA macros are obfuscated in several ways to deceive anti-virus programs. Therefore there are limitations on applying pattern-based detection to detecting these unknown malicious VBA macros. To detect unknown malicious VBA macros, some methods with machine learning techniques are applicable. One method extracts words from the source code, and constructs a language model to represent VBA macros for machine learning techniques. This method constructs a language model from all the extracted words which include trivial words. Hence, there seems still room for improvement of this model. To construct an efficient language model, this paper focuses on LSI (Latent Semantic Indexing). LSI is a fundamental technique in topic modeling and calculates similarity of documents. Our method extracts words from the source code and converts them into feature vectors with several natural language processing techniques. Our method trains a classifier with benign and malicious VBA macros and detects unknown malicious VBA macros. Several thousands of samples for evaluation are obtained from Virus Total. The experimental results show that our method could detect unknown malicious VBA macros more efficiently, and reveal the advantages and disadvantages of each language model.

Existing research on information security for IP cameras has been primarily focused on issues with authentication or malware, but not on the peeping method itself. How cyber peeping is conducted in real world can further help in strengthening defenses accordingly and spread more awareness about dangers of IP camera. In this research, we observed peeps by setting up a honeypot using decoy cameras in two scenarios. First, where background information (handwritten URL and ID/password bait) can be read by humans. Second, simulating a living-room in a home environment. As a result, many examples of peeping into the decoy cameras were confirmed in reality. Also, a rapid increase in peeping (over 20, 000 times/day) was seen after a decoy camera's feed got posted on a well-known website, showing a large scale peeping danger also exists due to such websites. The results of this study were used in several TV programs to show the dangers of using IP cameras over a national broadcasting station and also were directly shared with IP camera vendors, resulting in the improvement of IP camera security. Therefore, we believe that this study can further help in improving the security and awareness on the dangers associated with IP cameras.

De-identification is a process to prevent revealing the identity of a person based on personal data of that individual including personal identification information. In conventional de-identification studies, re-identification is a process used to identify individuals from static data where there is one record specified for each individual. In contrast, in this paper, we employ dynamic data, for example, trajectory data and online payment records. In particular, we consider the open competition data from the 2016 Privacy Workshop Cup (PWS Cup 2016) held in Japan consisting of purchasing history data. Throughout the analysis, we find that attackers can re-identify individuals with a high degree of accuracy from their de-identified purchase history data based on a feature of the set of goods. To address this re-identification risk, we propose a new method to de-identify history data by adding dummy records under certain restrictions. In our method, we use the Jaccard coefficient and the TF-IDF to form user clusters. We evaluate the performance of our proposed method and compare it with the performance of the PWS Cup 2016 participants as an experiment in data privacy. Even in the best de-identified data in PWS Cup 2016, 22.25% of customers were re-identified by our re-identification algorithm based on the Jaccard coefficient. However, only about 12% of customers are re-identified by random re-identification method and about 17% of customers are re-identified by re-identification method based on the Jaccard coefficient in the data that are de-identified by our de-identification method.

Recently, use of cloud computing has expanded to include tasks such as data sharing among multiple enterprises for open innovations and IoT systems for device management, etc. Public-key searchable encryption is particularly useful for cases requiring protection of shared secret data. However, there is no public-key searchable encryption scheme that simultaneously achieves: (A) efficient search performance, (B) multi-user support and (C) implementation on database management systems, simultaneously. In this paper, we propose a new public-key searchable encryption scheme which (A) provides an index generation mechanism for public-key searchable encryption, and (B) generates different secret keys for multiple users. In our scheme, (A) is achieved by a portion of the deterministic values generated from a keyword, and (B) is achieved by hierarchical inner-product predicate encryption. Our scheme in particular, forms wildcards as the user identity from hierarchical inner-product predicate encryption in order to easily represent hierarchical identities. To achieve (C), we further propose an integration method which implements our scheme into database management systems through a user-defined function so that its search functionality can be easily used via SQL.

Trajectories consist of spatial information of moving objects. Over contious time spans, trajectory data form data streams constantly generated from diverse and geographically distributed sources. Discovery of traveling patterns on trajectory streams such as gathering and companies enables value domain applications. Such a discovery needs to process arrival records in various sources and correlate across records near real-time. Thus techniques for handling trajectory streams should scale on distributed cluster computing. The challenge is at three aspects, namely a data model to represent the continuous trajectory data, the parallelism of the discovery algorithm, and an end-to-end parallel framework. In this paper, we propose a parallel discovery method that consists of 1) a model of partitioning trajectory samples on various time intervals; 2) definition on distance measurements of trajectories; and 3) a parallel discovery algorithm. We build a stream processing workflow and investigate experiments on a public dataset to evaluate the system's performance, scalability, stability, and data intensity. Our method discovers trajectory gathering patterns with low latency and scales as the size of trajectory data grows.

Robustness against failure and attack is one of the essential properties of large-scale dynamical system such as power grids, transportation system, communication systems, and computer networks. Despite its popularity and intuitiveness, a major drawback of descriptive robustness metrics such as the size of the largest connected component and the network diameter is computational complexity. Spectral measures such as the spectral radius, the natural connectivity, and the algebraic connectivity are much easier to obtain than descriptive metrics, but the predictability of those measures against different levels and types of failures has not been well understood. In this paper, we therefore investigate how effectively spectral measures can estimate the robustness of a network against random and adversary node removal. Our finding includes that, among five types of spectral measures, the effective resistance is most suitable for predicting the largest cluster component size under low node removal ratio, and that the predictability of the effective resistance is stable among different types of networks.

Content Delivery Networks (CDN) are regarded as an important internet infrastructure that supports business on the Internet. Network latency has been identified as an important metric to improve the quality of service (QoS) of CDNs. Given the limitations to response times because of the geographic distance between servers and end-users, CDN providers have developed architectures such as Global Server Load Balancing (GSLB) and IP Anycast to realize fast responses. This study investigates how GSLB and IP Anycast contribute towards improving QoS from the users' perspective. User traffic in the campus network was analyzed and it was found that: 1) 93.8% of the traffic reaps the benefits of well operated internet services, i.e., they have a Round-Trip Time (RTT) of less than 0.1 second. 2) Although GSLB is still the primary architecture for realizing fast responses, IP Anycast supports 5.9% of the traffic.

In recent years, Information-Centric Networking (ICN) that mainly focuses on contents that are transferred and received instead of on end hosts that transmit and receive contents has been under the spotlight. A straightforward approach for content routing in ICNs is to utilize a class of shortest-path routing mechanisms. In past research papers, several cache-aware routing mechanisms for ICNs to take advantage of content caches at intermediate routers have been proposed. In past studies, however, only link-level performance metrics (e.g., cache hit ratio and server load) of content routings have been investigated. Hence, it has been still unclear how the shortest-path routing is effective (or ineffective) in terms of application-level performance metrics. In this paper, we try to answer research questions regarding the optimality of the shortest-path routing. We compare the application-level performances with the shortest-path routing and with the optimal routing obtained by searching all detour paths existing in the vicinity of the shortest-path routing (optimal k-hop detour routing). Our findings include that the shortest-path routing is suitable when the network is balanced and cache sizes at routers are homogeneous, and that the optimal k-hop detour routing is suitable when variation in cache sizes is large.

Providing routes that are passable allowing movement by transportation modes such as wheelchairs or strollers requires accessibility information including details about the type and location of barriers. Earlier research detected the types of barriers using sensor data from people in wheelchairs and able-bodied people, but the amount and range of travel of people in wheelchairs is limited. Also, small barriers are not noticed by able-bodied persons since they tend to be easily ignored during movement. In our research, the goal was to detect barrier details using sensor data from various transportation modes. However, there are issues with the selection of barrier detection models for each mode and with the cost of collecting data. To overcome this model-selection problem, we propose a model that detects transportation modes and barriers in two stages. We also propose a method for reducing the cost of collecting data, with which we prepare a course with a smooth surface, collect data, and simulate rough surfaces by adding noise. We conducted three experiments to verify the effectiveness of the proposed method. The proposed method achieved an accuracy of 91.5% in detecting six transportation modes, and showed that adding noise increased accuracy by 3.7 percentage points on rough surfaces. When detecting eight types of barriers, our method achieved an accuracy of 87.7% for walking with a stroller, and showed that adding noise increased accuracy by 6.8 percentage points on rough surfaces. Therefore, the proposed method is effective in detecting barrier details using multiple transportation modes.

Track-based flow management systems like transportation systems and traffic control systems play a crucial role in our daily life. Safety and performance are among the most important quality requirements for these systems. This paper presents AdaptiveFlow as a framework for modeling, safety checking and performance analysis of track-based flow management systems. AdaptiveFlow is based on the Hewitt actors computation model. In particular, tracks are modeled as actors and moving objects as messages. Timed Rebeca is used for modeling, and the model checking tool Afra is used for safety verification and performance evaluation in AdaptiveFlow. To react to dynamic changes in the environment, AdaptiveFlow provides support for three adaptive policies, which can be examined and compared in different scenarios. To demonstrate the applicability of AdaptiveFlow, we consider the Electric Site Research Project of Volvo Construction Equipment as a case study. In this project, a fleet of autonomous haulers is utilized to transport materials in a quarry site. Furthermore, to show the reusability of the framework for other flow management scenarios, an experiment on an urban garbage collection system is presented.

Massive dataflows are generated from IoT applications for controlling real-world environments. For maintaining the long-running IoT application costs, it becomes important to reduce network traffic and delay considering hierarchical network characteristics, e.g., leveraging edge resources. Research and development of such a dataflow platform are ongoing, and the platform enables application developers to build a new dataflow application by just combining reusable software components. However, the existing dataflow platform does not support component management, which includes application deployment and placement selection, assuming the hierarchical network environments. In this paper, to realize the dataflow platform in a hierarchical network, we proposed an approach to deploy components into heterogeneous environments using a major open-source product and to enable the large-scale deployment by considering the structure of network clusters. Then, a component deployment layer selection method is also proposed considering network and resource conditions in the proposed platform. During the definition of the component placement problem, we also extracted parameters, which are dataflow platform parameters, dataflow application parameters and dataflow application component parameters, required to estimate component deployment budgets. Furthermore, we implemented a simple dataflow application based on two use cases and confirmed the validity of the proposed method.

Nowadays, vehicles are equipped with multiple Electronic Control Units (ECUs) each of which communicates with one another using a specification called Controller Area Network (CAN). CAN provides its own share of benefits in modernizing automobiles, but it also brought along a security issue to the automotive industry. CAN bus does not have any mechanism for encrypting or authenticating CAN payloads. As a countermeasure against these drawbacks, we have experimented on identifying intrusions in the CAN bus using Long Short-Term Memory Networks (LSTM). LSTM networks are trained with features extracted from reverse engineered packets. In a specific range of time windows, we have extracted three parameters, the number of packets, the bit flip rate and the average time difference that are used to train LSTM. The trained LSTM is later then used to predict all the three features which will be combined to a single anomaly signal using a root mean squared error. Depending on which side of the threshold appears the anomaly signal value, we managed to identify anomalies in an acceptable performance rate, up to F1 score of 98%. We have tested our methods with a variety of attacks on the CAN bus and demonstrated how effective our detection methods is.

Given the sheer amount of digital texts publicly available on the Internet, it becomes more challenging for security analysts to identify cyber threat related content. In this research, we proposed to build an autonomous system to identify cyber threat information from publicly available information sources. We examined different language models to utilize as a cybersecurity-specific filter for the proposed system. Using the domain-specific training data, we trained Doc2Vec and BERT models and compared their performance. According to our evaluation, the BERT-based Natural Language Filter is able to identify and classify cybersecurity-specific natural language text with 90% accuracy.

The lack of inbuilt security protocols in cheap and resource-constrained Internet of Things (IoT) devices give privilege to an attacker to exploit these device's vulnerabilities and break into the target device. Attacks like Mirai, Wannacry, Stuxnet, etc. show that a cyber-attack often comprises of a series of exploitations of victim device's vulnerabilities. Timely detection and patching of these vulnerabilities can avoid future attacks. Penetration testing helps to identify such vulnerabilities. However, traditional penetration testing methods are not End-to-End, which fail to detect multi-hosts and multi-stages attacks. Even if an individual system is secure under some threat model, the attacker can use a kill-chain to reach the target system. In this paper, we introduced first-of-its-kind, IoT-PEN, a Penetration Testing Framework for IoT. The framework follows a client-server architecture wherein all IoT nodes act as clients and “a system with resources” as a server. IoT-PEN is an End-to-End, scalable, flexible and automatic penetration testing framework for discovering all possible ways an attacker can breach the target system using target-graphs. Finally, the paper recommends patch prioritization order by identifying critical nodes, critical paths for efficient patching. Our analysis shows that IoT-PEN is easily scalable to large and complex IoT networks.

This work proposes the development of a method to allow the use of smartwatch as a control for interactive movies using continuous recognition of gestures in smartwatches. We developed two prototypes to interact with interactive movies. The first one uses gesture recognition on the smartwatch screen. Thus, we created a set of gestures composed of straight lines that represent the actions in the interactive movie. In this prototype, the recognition of gestures is performed by the algorithm of continuous recognition of gestures. In this way, a gesture can be recognized before the user finalizes, and the action is sent quickly to the movie. The second prototype uses a touch of pressure in the smartwatch to control the film. As a study case, we have conducted our user study with the movie “Black Mirror: Bandersnatch”, with usability and experience evaluation. A result has shown that the gestures proposed in the method are intuitive and that it has the potential to be used pleasantly and satisfactorily by the users. For 90% of the participants, the use of the smartwatch is more practical and easy to control Netflix compared to the remote control.

Open source software (OSS) has become indispensable to our society. The success of OSS depends on the participation of a large number of developers or maintainers (contributors). Shedding light on the mechanisms of their participation has been an important academic and practical matter. One aspect to decide participation is the future prospects of a project. However, the causal mechanism behind participation has yet to be studied exhaustively and remains unclear. In this study, we used cryptocurrency projects, many of them were developed on GitHub, to better understand this mechanism. Both GitHub and cryptocurrencies are highly transparent, i.e., information is fully disclosed; we can analyze relevant information on a project, such as the contributors' activities, financial information, and development status. We adopted market capitalization as the substitution index of future prospects and the number of contributors and analyzed the relationship using time series analysis techniques, such as the Granger causality test and regression. We found that the number of contributors increases two months after market capitalization increases. This quantifies the impact of the future prospects of the project, i.e., of the market capitalization of a cryptocurrency, on the participation of contributors.

As an effort to realize a sustainable society, Tokai University Takanawa Campus has used a smart garbage collection service called BigBelly^TM since 2016 for recovering wastes such as cans, bottles, and plastic bottles for recycling. The objective of this demonstration experiment is to clarify how much waste has been correctly separated and collected. As a result of this demonstration about three years, it found that about 30% of the recycling wastes had not correctly sorted. To improve this situation, we propose an automatic sorting function using sound recognition. In many types of research for voice recognition, Mel Frequency Cepstral Coefficient (MFCC) has been used as an algorithm for extracting features used for machine learning Support Vector Machines (SVMs). One reason is that MFCC extracts valuable features that focus on the low dimension for the human voice. However, the sounds of recycling wastes have features of frequency components found in higher dimensions. Based on this characteristic, we propose an improved method of MFCC suitable for sounds, rather than voice recognition for identifying recycling wastes and show the results of the automatic sorting of recycling wastes.

With the development of information and communication technology (ICT), smart home technologies are expecting to be a potential solution for the population aging problem. Smart home simulators and testbeds have been introduced to provide materials for researchers in the field of ambient assisted living to develop and evaluate their solutions. Since simulators reduce costs in terms of money and time, researchers are utilizing simulators to expand services to enhance user comfort, save energy, and detect abnormal behaviors in daily living activities, etc. However, available smart home simulators seem to focus on the operational aspects of simulated devices and environments. The network communication aspects have not been fulfilled so far. Following the development of the Internet of Things, smart home networks are more complex, and users in the smart home are ordinary people without knowledge about network management. Therefore, intelligent operation, administration, and maintenance (OAM) services for smart homes are desired. This paper proposes a smart home network simulator to generate a network dataset of a home network, which is essential to develop machine learning technologies for providing intelligent OAM services in smart homes. The simulator is implemented based on ECHONET Lite, a leading smart home protocol in Japan.

With the increasing spread of smartphones and wearable devices equipped with various sensors, human activities, biometric information, and surrounding situations can be recognized. The process of human activity recognition must construct a model that has learned annotated sensor data, i.e., ground truth, labels, or answer activity, in advance. Therefore, a large and diverse set of annotated data is required to improve and evaluate model performance. It is difficult to judge a user's situation even after observing acceleration data; thus, it is necessary to annotate the collected acceleration data. In this paper, we propose a method to estimate user and device situations from the user's response to a notification generated by a device, e.g., a smartphone. The user and device situations are estimated from the user's response time to the notification and the device's acceleration values. An estimation result with high confidence is given to the sensor data as an annotation. Increasing the frequency of notifications, response to the notifications can be used as a sensor. We assume that acceleration values are affected by a user and device situation when the device notifications are taken instantly after its generation. The system pursues a high precision of estimation by selecting input acceleration data based on the interaction to the notification so that the estimations can be used as annotations. Through an evaluation experiment, for seven types of annotation classes, an average precision of 0.769 and 0.963 for user-independent experiments and user-dependent experiments were achieved, respectively. We also tested the proposed method in a natural environment, where 25 correct annotations were given for 45 responses to notifications, no annotations were given for 19 responses, and only one incorrect notification was observed.

This paper presents an unsupervised learning-based method for detecting energy depriving malicious nodes in an energy harvesting cooperative wireless sensor network (EHC-WSN). In EHC-WSNs, nodes wirelessly transfer a portion of their energy to their neighboring nodes if their neighboring nodes lack energy. A malicious energy depriving mode may falsibly show that it has little energy and in this way obtains energy from neighboring nodes, thus depriving them of energy. To detect these malicious nodes, we utilize a clustering method. In our method, each node first observes the energy of its neighboring nodes, then it utilizes this information to obtain data points for the clustering. After clusters are formed, each node judges on a cluster of data points from malicious nodes and makes a malicious node determination. We investigate the performance of our method and confirm that our method outperforms the baseline method in terms of detection accuracy and false detection rate.

As many psychologists have suggested, evoking memories of gratitude and then writing them down is recognized as an efficient daily activity to improve people's subjective happiness and well-being. This paper studies how using a smart voice agent, such as a smartphone and smart speaker, affects this activity. Instead of traditional handwriting, expressing gratitude through the voice agent has several benefits in terms of human-computer interaction aspects. As a first step for this study, we conducted short-term experiments using a voice agent with Japanese university students, including native Japanese and foreigners, and evaluated changes in their emotional state by using two psychological measurements, the Positive and Negative Affect Schedule (PANAS) and the Subjective Happiness Scale (SHS). The evaluation results revealed three important findings: (1) expressing and receiving gratitude via voice agent can potentially enhance positive affect and subjective happiness similar to traditional handwriting, (2) generating natural gratitude messages (e.g., from close friends or family) significantly improves positive affect and happiness, and (3) some people experience stress over tasks or methods of expressing gratitude, decreasing positive affect.

Distance cartograms are deformed maps in which the distance of each of the preselected point pairs in the geographic map is changed in step with a specified value such as a travel time. In distance cartogram construction, the preselected points are fixed in the first step whereas the locations of other points are converted in the second step. This paper proposes a new point location conversion method for the second step. The conversion process is modeled as a phenomenon in a three-dimensional vector field. Each point in the geographic map is connected with the corresponding point in the cartogram by a streamline of the field. The connection relationship becomes a smooth homeomorphism required in distance cartogram construction. The experimental results demonstrate its effectiveness.

A reverse dictionary maps a description to the word specified by the description. The neural reverse dictionary (NRD) learns to map word embeddings for an input definition into an embedding of the word defined by the definition using neural networks. Such a function encodes phrasal semantics and bridges the gap between them and lexical semantics. However, previous NRD has a limitation in accuracy due to its insufficient capacity. To solve this problem, we used novel combinations of neural networks, which are effective in neural machine translation and image processing, with sufficient capacities. We found that, an LSTM output adjustment by using a multi-layer fully connected network with bypass structures (CFNN) was more effective for reverse dictionary tasks than using more complicated LSTM. BiLSTM+CFNN was comparable to the commercial system OneLook Reverse Dictionary in some metrics, and noised biLSTM+CFNN which we tuned by a noising data augmentation outperformed OneLook Reverse Dictionary in almost all metrics. We also examined the reasons for the success of biLSTM+CFNN and revealed that a bypass structure of the CFNN and balance in the capacity of LSTM and the CFNN contribute to the improved performance of the NRD.

It is important to handle large-scale data in text formats such as XML, JSON, and CSV because these data very often appear in data exchange. For these data, instead of data ingestion to databases, ad hoc data extraction is highly desirable. The main issue of ad hoc data extraction is to serve both the programmability to allow handling various types of data intuitively and the performance for large-scale data. To pursue it, we develop CENTAURUS, a dynamic parser generator library for parallel ad hoc data extraction. This paper presents the design and implementation of CENTAURUS. The experimental results on ad hoc data extraction have demonstrated that CENTAURUS outperformed fast dedicated parser libraries in C++ for XML and JSON, and achieved excellent scalability with actions implemented in Python.

There is a great need to evaluate and/or test programming performance. For this purpose, two schemes have been used. Constructed response (CR) tests let the examinee write programs on a blank sheet (or with a computer keyboard). This scheme can evaluate the programming performance. However, it is difficult to apply in a large volume because skilled human graders are required (automatic evaluation is attempted but not widely used yet). Multiple choice (MC) tests let the examinee choose the correct answer from a list (often corresponding to the “hidden” portion of a complete program). This scheme can be used in a large volume with computer-based testing or mark-sense cards. However, many teachers and researchers are suspicious in that a good score does not necessarily mean the ability to write programs from scratch. We propose a third method, split-paper (SP) testing. Our scheme splits a correct program into each of its lines, shuffles the lines, adds “wrong answer” lines, and prepends them with choice symbols. The examinee answers by using a list of choice symbols corresponding to the correct program, which can be easily graded automatically by using computers. In particular, we propose the use of edit distance (Levenshtein distance) in the scoring scheme, which seems to have affinity with the SP scheme. The research question is whether SP tests scored by using an edit-distance-based scoring scheme measure programming performance as do CR tests. Therefore, we conducted an experiment by using college programming classes with 60 students to compare SP tests against CR tests. As a result, SP and CR test scores are correlated for multiple settings, and the results were statistically significant. Therefore, we might conclude that SP tests with automatic scoring using edit distance are useful tools for evaluating the programming performance.

Every orthogonal graph, extruded orthogonally from a rectangle, can be folded from a rectangle of paper a constant factor larger. This computational origami result was proved a decade ago, and has since enabled the design of a mathematical/puzzle font and a variety of art prints. Here we survey the maze-folding art prints we have designed.

Every net of an isotetrahedron (I) or a rectangle dihedron (RD) is a Conway tile. Reversely, it is shown by using Alexandrov's theorem that every Conway tile can be folded into either I or RD. However, it was not known how to fold a given Conway tile into I or RD. The purpose of this paper is to give an algorithm for folding a Conway tile into I or RD. Moreover, for a given Conway tile we present a method to identify the exact shape of I or RD into which it can be folded.

We show that every minimum area isosceles triangle containing a given triangle T shares a side and an angle with T. This proves a conjecture of Nandakumar motivated by a computational problem. We use our result to deduce that for every triangle T, (1) there are at most 3 minimum area isosceles triangles that contain T, and (2) there exists an isosceles triangle containing T whose area is smaller than √2 times the area of T. Both bounds are best possible.

A data structure is presented that explicitly maintains the graph of a Voronoi diagram of N point sites in the plane or the dual graph of a convex hull of points in three dimensions while allowing insertions of new sites/points. Our structure supports insertions in Õ(N^3/4) expected amortized time, where Õ suppresses polylogarithmic terms. This is the first result to achieve sublinear time insertions; previously it was shown by Allen et al. that Θ(√N) amortized combinatorial changes per insertion could occur in the Voronoi diagram but a sublinear-time algorithm was only presented for the special case of points in convex position.

We investigate the phase transition of a dynamical system generating a possibly infinite orbit of points. The points of the orbit are generated according to the following basic operation. Given a positive real number $a$, called the expansion factor, and two points $p, q$ at Euclidean distance $|pq|$ we determine the unique point $p'$ on the straight line passing through $p$ and $q$ which is antipodal to the point $p$ with respect to $q$ and at a Euclidean distance $a |pq|$ from $q$. The operation on points previously defined is denoted by $p \Rightarrow_{a,q} p'$. Let ${\bf a} := a_0,a_1, \ldots, a_{n-1} $ be arbitrary but fixed positive real numbers and ${\bf q} := q_0, q_1, \ldots, q_{n-1}$, be $n$ (anchor) points. An orbit consisting of an infinite sequence $p_0, p_1, \ldots , p_m, \ldots$ of points in the plane is generated by using the anchor points as follows. The orbit is initiated with an arbitrary point $p_0:=p$ and for all integers $m \geq 0$, satisfies $p_m \Rightarrow_{a_{m \bmod n},q_{m \bmod n}} p_{m+1}$ so that $p_{m+1} := (p_m)'$. The resulting sequence of points is called the $({\bf a}, {\bf q})$-orbit of $p$. For any starting point $p$ and any pair $({\bf a}, {\bf q})$ we characterize the boundedness of $({\bf a}, {\bf q})$-orbits. Namely, we show that there is a phase transition concerning the boundedness of the resulting $({\bf a}, {\bf q})$-orbit which depends on whether the product $a_0 a_1 \cdots a_{n-1}$ of the expansion factors is less or larger than one. We also characterize the behaviour of the orbits when $a_0 a_1 \cdots a_{n-1} = 1$. The “boundedness” phase transition phenomenon described above is shown to be valid for any dimension $d=1,2,3$ in Euclidean space. In addition, we propose variants of this approach for generating orbits on convex polygons, and propose several open problems corresponding to phase transition phenomena.

Motivated by advances in microscale applications and simplistic robot agents, we look at problems based on using a global signal to move all agents when given a limited number of directional signals and immovable geometry. We study a model where unit square particles move within a 2D grid based on uniform external forces. Movement is based on a sequence of uniform commands which cause all particles to move 1 step in a specific direction. The 2D grid board additionally contains “blocked” spaces which prevent particles from entry. Within this model, we investigate the complexity of deciding 1) whether a target location on the board can be occupied (by any) particle (Occupancy problem), 2) whether a specific particle can be relocated to another specific position in the board (Relocation problem), and 3) whether a board configuration can be transformed into another configuration (Reconfiguration problem). We prove that while occupancy is solvable in polynomial time, the relocation and reconfiguration problems are both NP-Complete even when restricted to only 2 or 3 movement directions. We further define a hierarchy of board geometries and show that this hardness holds for even very restricted classes of board geometry.

We give a complete description of all convex polyhedra whose surface can be constructed from several congruent regular pentagons by folding and gluing them edge to edge. Our method of determining the graph structure of the polyhedra from a gluing is of independent interest and can be used in other similar settings.

We propose an algorithm for finding a minimum forcing set of a given flat-foldable single-vertex crease pattern (SVCP). SVCP consists of straight lines called creases that can be labeled as mountains or valleys, and the creases are incident to the center of a disk of paper. A forcing set is a subset of given creases that forces all other creases to fold according to the given labels. Our algorithm is a modification of an existing algorithm for 1D origami. We show that the size of a minimum forcing set of an SVCP is n/2 or n/2+1 where n is the number of creases in the SVCP.

In this paper, we study a variation of the map folding problem. The input is a 2 × n map with a box-pleated crease pattern of size 2 × n. Precisely, viewing the crease pattern as a planar graph, its vertices and edges respectively form the subsets of the vertices set and edges set of the planar graph of the square and diagonal grid. The question is whether the map can be flat-folded or not. If the answer is yes, then what is the time complexity to make the decision? Our conclusion is that any locally flat-foldable 2 × n map with such a box-pleated crease pattern is globally flat-foldable. We present linear-time algorithms for both deciding the flat foldability and finding a feasible way of folding.

In this work, we consider a decision problem on whether an overlapping order of all the squares of an m × n map is valid under a particular context called simple folding. This problem belongs to the field of map folding. It is a variation of both the decision problem of valid orders of a 1 × n map under the context of simple folds and the decision problem of valid orders of an m × n map under the context of general folds. We provide a method and its corresponding linear-time algorithm to solve this problem, which is based on the construction of a directed graph representing the adjacency relations among squares.

Deciding flat foldability of a given mountain-valley pattern is known to be NP-complete. One special case known to be solvable in linear time is when the creases are parallel to each other and perpendicular to two sides of a rectangular piece of paper; this case reduces to a purely one-dimensional folding problem. In this paper, we give linear-time algorithms for flat foldability in two more-general special cases: (1) all creases are parallel to each other and to two sides of a parallelogram of paper, but possibly oblique to the other two sides of the parallelogram; and (2) creases form a regular zigzag whose two directions (zig and zag, again possibly oblique to the two sides of the piece of paper) form nonacute angles to each other. In the latter zigzag case, we in fact prove that every crease pattern can be folded flat, even if each crease is specified as mountain, valley, or unfolded.

Several methods described in the literature have proved that any convex pyramid can be continuously flattened. Recently, the problem of continuous flattening of polyhedra having divisions, i.e., polyhedra in which some of the edges are incident to three or more faces, has been proposed. However, for such multi-layered structures, continuous flattening motions are unknown. In this study, under the assumption that every radial edge is rigid, we prove that a continuous flattening motion exists for a pyramid with a convex base. Moreover, in a similar manner, we demonstrate that a continuous flattening motion exists for a multi-layered pyramid having a common convex base, with each apex having a common perpendicular foot. Finally, we illustrate an example of a multi-layered pyramid with a non-convex base that cannot be continuously flattened while maintaining the rigidity of the radial edges.

In this paper, we investigate the problem that asks if there exists a net of a polycube that is exactly a rectangle with slits. For this nontrivial question, we show affirmative solutions. First, we show some concrete examples: (1) no rectangle with slits with fewer than 24 squares can fold to any polycube, (2) a 4 × 7 rectangle with slits can fold to a heptacube (nonmanifold), (3) both of a 3 × 8 rectangle and a 4 × 6 rectangle can fold to a hexacube (nonmanifold), and (4) a 5 × 6 rectangle can fold to a heptacube (manifold). Second, we show a construction of an infinite family of polycubes folded from a rectangle with slits. The smallest one given by this construction is a 6 × 20 rectangle with slits that can fold to a polycube of genus 5. This construction gives us a polycube for any positive genus. Moreover, by this construction, we can show that there exists a rectangle with slits that can fold to k different polycubes for any given positive integer k.

Let G be a 3-connected triangulated disc such that the boundary cycle C of the outer face is an induced cycle of G and G-C is a tree. In this paper we prove that $\gamma(G) \leq \frac{n+2}{4}$, which gives a partial solution for the conjecture that the same inequality holds for any 3-conneced triangulated disc. We also show related conjectures.

In this study, we consider a problem, for a given graph G = (V, E), of finding the minimum number of 3-cliques (K₃s) that cover all edges of G. Multiple covering or covering one edge by more than one 3-clique is allowed. Moreover, in this problem, we allow “spilling-out, ” i.e., a set of three vertices {x, y, z} can be covered by a 3-clique even if the induced subgraph by them is not a clique. We call this problem K₃ edge cover problem in a wide sense. This problem is a kind of extension of the schoolgirl problem and finite projective planes, and it has applications on experimental designs. Allowing spilling-out is useful for some applications: E.g., when we want to compare n items through some tries of experiments, in which at most three items can be compared simultaneously, and pairs of items that must be compared are given by a graph, finding the minimum number of tries is formalized as this problem. In the known literature, there are many results that considered problems for covering vertices or edges by the minimum number of cliques. However, there is no theoretical result that considers spilling-out. We obtain the following results: (1) The problem is NP-hard even if graphs are restricted to planar, cubic, and C4, C5-free in a sense of subgraphs (i.e., not restricted to induced ones). (2) For the problem with a parameter k, which is the number of 3-cliques in G, there is an O(mn + 2^km)-time algorithm. (3) If a tree-decomposition of tree-width t is given, there is an O(2^2(t+1)(t+2)t²n)-time algorithm.