Evolutionary processes are one of the most widely studied methods for market analysis. Genetic algorithm (GA) and genetic programming (GP) are considered to be two most used evolutionary processes to find solutions. Many were reported successful but some were not. Looking at those of GA or GP generated technical rules, they have some computational forms consisting of variables, constants, market data, operators, functions and so on. Most important ones are the functions (or indicators) such as moving averages. Those are essentially and often directly based on the indicators themselves. This suggests that developing good technical indicators is essential even with evolutionary processes. Most of the currently used market timing indicators were developed when fast computers were not conveniently available. There should be many undeveloped indicators which should be suitable with computers. In this paper, new indicators called numerical composite indicators are proposed and tested. And it is also shown that different ways to apply indicators make some significant differences.

View full abstract

The purpose of this study is to know company's consciousness on information literacy, need of information literacy of new recruits(=university graduates), and present condition of information literacy of new recruits in order to find the ideal information education for improving information literacy of university students in Japan. So we conducted a questionnaire survey of company's consciousness toward information literacy of new recruits in January, 2008. Main items of this survey are need and present condition of information literacy, condition of training for improving information literacy of new recruits, and present condition of company's informatization. From the results, it was shown the next things:majority of the companies put a point of information literacy on recruitment; acquirement of new recruits with high information literacy is not enough;companies didn't have high satisfaction of information literacy of new recruits; and new recruits didn't have enough information literacy except office applications & the internet applications.

View full abstract

We have a job with N tasks each of which is executed successively until it is completed. A double modular system of error detection for the processing of each task is adopted. Either type of checkpoints such as compare-checkpoint or compare-and-store-checkpoint can be placed at the end of tasks. Three schemes for the above process of a job are considered and the mean execution times of each scheme are obtained. Three schemes are compared and the best scheme is determined numerically. Finally, as an example, a job with four tasks is given and six types of schemes are compared numerically.

View full abstract

A high quality and low cost, the improvement of the appointed date of delivery are demanded from the medium size or small size Production enterprises. In the global environment, speed and communalization of the information are needed to offer a product quickly. For customer satisfaction and security of the company profit, a production schedule system is necessary. A computer-assisted system to decide production activity is essential to the medium and small-sized enterprises. However, the burdens of duties increase by system introduction and affect production schedule. However, the burdens of duties increase by system introduction, and there is a thing affecting production schedule. I investigate problems to occur by the production schedule management that used a computer-assisted system. In this study, I suggest the production schedule management that is appropriate than a thing of a medium and small-sized enterprises.

View full abstract

The Major companies have been starting to use the ICT in order to accelerate the development term and unify the management of information as well reduce the ecological problem. For example, they would use the ICT instead of the paper. In order to digitalize the technical information for the vocational Initiation the effective usage of the ICT must be the key technique. So we set some idea that the digitalized technical information (called digital contents) made by the companies would work effectively for the education in the school as the digital contents are always updated and useful for paperless activity. Thus we have investigated the operating education method by the assistance of the ICT. And also we will investigate the risk of the digital contents through the ICT.

View full abstract

This paper concerns the decentralized dynamic output feedback control for nonlinear interconnected systems with time delay in the discrete time case. We use Takagi-Sugeno (T-S) fuzzy model of nonlinear interconnected system. Based on T-S fuzzy interconnected system, the decentralized dynamic output feedback controller is design. For stabilization of interconnected systems, sufficient conditions are derived for control gains and expressed as the linear matrix inequalities (LMIs). A numerical example is given to show the verification of the proposed methodology.

View full abstract

In this paper, we focus on the guaranteed cost control for nonlinear neutral systems with time-varying delays represented by Takagi-Sugeno (T-S) fuzzy model. A linear quadratic cost function is considered as a performance index of the nonlinear neutral systems. The problem addressed that of designing a guaranteed cost control that guarantees the asymptotically stable of the nonlinear neutral systems. The sufficient stability conditions are derived via Lyapunov--Krasovskii functionals (LKF). When these conditions, which are given in terms of linear matrix inequalities (LMIs), are satisfied, the designed controller gain matrices can be obtained via a convex optimization. Furthermore, an illustrative example is provided to demonstrate the effectiveness of the proposed design procedure.

View full abstract

Photovoltaic systems (PVS) are among the fastest growing energy resources in the world. Most of the PVS had been installed in the distribution systems as distributed generation. Conventionally, distribution systems are designed and operated based on the assumption of centralized generation, with the corollary that the power flows from the distribution substation to the end-use customers. The problem for correct sizing and positioning of PVS distributed generation in distribution systems is of great importance. For that reason, the use of an optimization method capable of indicating the optimal sizing and position for a given distribution network can be very useful. In this work, particle swarm optimization (PSO) algorithm is presented to simultaneously determine the optimal size and location of PVS in a given distribution system network.

View full abstract

In this paper, we develop an intelligent diagnosis system of real-time fault monitoring system for transmission line. Based on the Time-Frequency Domain Reflectometry (TFDR) algorithm, the wire detecting system shows the condition of the wires. We analyze the reflected signal which is sent from the wire detecting system and classify the fault type of the wires by using intelligent diagnosis system. Generally, the conditions of the wires are classified into the three types - damage, open and short. For classifying the fault type efficiently, we use the fuzzy-Bayesian classifier which is represented as IF-THEN rules. We show the utility of the proposed algorithm by performing the simulation which is based on the data of the transmission line.

View full abstract

This paper presents a new type of clustering algorithm by using tolerance vector. The tolerance vector is considered from a new viewpoint that the vector shows a correlation between each data and cluster centers in the proposed algorithm. First, a new concept of tolerance is introduced into optimization problem. This optimization problem is based on conventional fuzzy c-means (FCM) by Bezdek. Second, the optimization problem with the tolerance is solved by using the Karush-Kuhn-Tucker conditions. Next, a new clustering algorithm is constructed based on the unique and explicit optimal solutions of the optimization problem. Finally, the effectiveness of the proposed algorithm is verified through some numerical examples.

View full abstract

A new fuzzy c-means algorithms for data with tolerance is proposed by introducing a penalty term in feature space. Its idea is derived from that support vector machine introducing a penalty term for "soft margin" in feature space. In the proposed method, the data is allowed to move for minimizing the corresponding objective function but this move-ness is controlled by the penalty term. First, an optimization problem is shown by introducing tolerance with conventional fuzzy c-means algorithm in feature space. Second, Karush-Kuhn-Tucker~(KKT) conditions of the optimization problem is considered. Third, an iterative algorithm is proposed by re-expressing the KKT conditions using kernel trick. Fourth, another iterative algorithm is proposed for fuzzy classification function, which shows how prototypical an arbitrary point in the data space is to the obtained each cluster by extending the membership to the whole spa ce. Last, some numerical examples are shown.

View full abstract

The design of various fuzzy architectures with mixed signal current mode techniques have been discussed here. The proposed work focuses on generation of various shaped membership functions like triangular, trapezoidal, guassian. It also describes the implementation of a membership function circuit using MOSFETs. It implements the simplified inference mechanism wherein MOS analog circuits are used to represent signals as currents. However, the whole system is externally communicated through voltages, thus enabling simple interface with conventional control circuitry. The design requires only one differential pair for each membership function & the output currents are automatically normalized. The normalization circuit is integrated within the fuzzifier. By choosing different reference voltages and different W/L ratios, various types of membership functions can be accomplished. The fuzzy rule evaluation or fuzzy inference is also discussed with reconfigurable inference engine.

View full abstract

Asynchronous circuits are a technique in order to solve some synchronous circuits problems such as clock skew and noise. We researched NCL circuits which is one of asynchronous circuits, and it can be synthesize only "Hysteresial Threshold Gate" which is a special gate. We have proposed Hysteresial Threshold Gate in Neuron MOS. We now propose "Multiple Threshold Gate with Hysteresis" which is general gate using of a characteristic of variable threshold value in Neuron MOS. Especially this paper show that can unify hysteresis and characteristic of variable as these pay attention to two common points. We designed Multiple Threshold Gate with Hysteresis which can be operate a function as threshold 1, 2, or 3, and simulated it with SPICE. As a result, we can synthesize Multiple Threshold Gate with Hysteresis which has 3 inputs and 2 control signals and then NCL Half Adder can be synthesized in this gate.

View full abstract

In this paper, an effective adaptive real-parameter simulated annealing genetic algorithm (ARSAGA) that is applied to cope with mixed-integer nonlinear programming problems. The proposed method synthesized the merits of both genetic algorithm and simulated annealing. Adaptive mechanisms are also included to make evolutionary scheme active and result in improving the hill-climbing ability and the convergence speed. The performances of this proposed algorithm are demonstrated in several large parameter optimization functions. Due to their versatile characteristics, these examples are suitable to test the ability of the proposed algorithm. The results of this novel hybrid algorithm under different population sizes and frozen numbers were discussed and appropriate parametric combinations of both two parameters were also suggested in this paper. ARSAGA also shows excellent performances in large parameter mixed-integer optimization problems.

View full abstract

We propose a new type decision support system of modeling the illustration designs suited for personal preference, and extract the coloration pattern from personal designs. Many coloration patterns give various impressions and effects to people, and the color design is one of the most important elements that influence the deciding the image of products. However, people's preferences of colors are differed to individual personality. Nevertheless, modeling the color design is depended on the designer's sensibility. Therefore, it is needed that we extract individual coloration pattern suited for their preference. In addition, the technique of modeling the various color designs people satisfy is needed. In this paper, we propose automatic modeling system using interactive genetic algorithm (:IGA). The proposed system can be helpful for the modeling the illustration designs. We apply the proposed system to some subjects, and collect the design data. By using the design data, we extract each subject's coloration pattern.

View full abstract

In order to solve the wide range complex optimization problem such as maintenance strategies for existing infrastructures, in this study, proposed was to make maintenance strategies of bridges by using sharing-type GA. Through the numerical simulations, it found that the proposed GA model could successfully achieve the escape from the local minimum. Then, we confirmed that this model is effective for making maintenance strategies of bridges.

View full abstract

In this paper, we propose an adaptive method to construct a fuzzy rule-based systems for a point-to-point car racing game. In the car racing game, two car agents compete with each other for ordered waypoints. This fuzzy system makes high-level decision making for car agents. The proposed method updates the grade of certainty of fuzzy if-then rules through the course of games and makes the fuzzy rule-based system more adaptive to an opponent car agent. In computational experiments, we evaluate the performance of proposed fuzzy rulebased system.

View full abstract

This paper discusses a perceptual system for intelligent control of robot arms. Especially, our target of the robot arm system should perform the task of cleaning the table. First steps for execution, we proposed perceptual system using a 3D-range camera and a CCD camera for the intelligent control of robot arm. In this study, we proposed perceptual system based on retinal structure by using 3D-TOF(Time-Of-Flight) range camera that can measure distance directly in finite field. Moreover, we proposed a image feature matching that detect a target dish more robustness in changing environment. Therefore, we apply the proposed method for a task of clearing the table. As an experimental result, we show the directly perceive the necessary information for control of robot arms.

View full abstract

There are many researches on the communications between human and robot. The authors are working on controlling the robot with users' intuitive gestures. In this paper, the systems are enhanced to handle symbolic expressions for more suitable use of human side. Firstly, the robot learns how the users want to control it with intuitive gestures. Gestures are classified using growing self-organizing maps. The system learns the correspondence between the gesture and the command to be executed. To enhance the communication scheme more easily, the correspondences between the gesture and the symbolic expression through LED/sound have been developed. Some simple experiments show the feasibility of our proposal.

View full abstract

We developed an Online BARNNGA, which is social simulation game towards development of social agent.We observe human's social skills of leaning to implicit understanding. This game based on BARNGA whose purpose is experience to explore factors related to communication problems in intercultural situations with cards game: trick taking game. In BARNGA, differences of a game's rule represent cultural difference. Likewise, implicit rules can represent likely differences of a game's rule. Then we observe behavior of players on the game and analyze social skills.

View full abstract

In this paper, we propose a fully automatic real time one image face gesture simulation using image morphing. Given a single image of a subject, we create several facial expressions of the face by morphing the image based on prior information stored in a databank. The process involves the automatic detection of the control points both on the target image and the source data. The source data is a string of frames containing the desired facial expressions. A face detection neural network, lips detector and edge detectors are employed to detect the face position and features. Five control points and the lips contour, for both the source and target images, are then extracted bvased on the facial features. Triangulation method is then used to match and warp the source image to the target image using the control points.

View full abstract

Recent studies have been shown that 2DLDA loses the covariance information between different local geometric structures, such as the rows or the columns, which is useful for discriminant feature extraction, whereas 1DLDA could preserve such information. However, keeping all this information will result in a huge covariance matrix which it needs to a large number of training data for correctly evaluation. In this paper a new method called Extended two-dimensional LDA (E2DLDA) is proposed keeping more covariance information than 2DLDA. Furthermore, it adjusts the amount of preserved information according to the number of training samples. Moreover, it represents the face image with fewer coefficients than 2DLDA. Experimental results on ORL database shows that proposed method obtains better recognition accuracy than both 2DLDA and 1DLDA.

View full abstract

This paper presents a new incremental learning algorithm named Incremental Simple-FLDA. It is the algorithm that improved Simple-FLDA to update an eigenvector by using incremental data. Simpe-FLDA is an approximation algorithm of the linear discriminant analysis where an eigenvector can be calculated by a simple repeated calculation. Using this proposal algorithm, the eigenvector is approximately obtained whenever incremental data is added one by one. We carry out computer simulations on personal authentication that uses face images by incremental learning to verify the effectiveness of this algorithm. As a result, the effectiveness of proposal algorithm is shown.

View full abstract

In recent years, some of the developments have brought in severe competition that challenges the vitality of Taiwan's 160 universities and colleges. Not to lag behind in the competition, these educational institutions have to boost their educational quality and competitiveness. The concept of accreditation of engineering programs emerged out of this environment as a driving force to assure the standard of bachelor degree level of engineering programs in Taiwan. In preparing the IEET accreditation, one issue should be addressed: an evaluation system to measure students? learning. In order to solve these issues, this research proposes the use of multi-criteria and multi-connection comprehensive assessment (MCCA) model to quantify overall students? learning. Finally, the proposed approach will be applied to the Environmental Engineering Department of Da-Yeh University to demonstrate its use.

View full abstract

In this paper, we propose a system for assisting teachers with making effective comments from descriptive answers of a quiz. The system provides a summary of answers that helps teachers to identify the level of learners' comprehension. Teachers would make effective comments by the summary without reading all answers. Especially, we aim to shorten a time that learners wait for comments after their submissions. The longer the time is, the worse learners understand. We improve our previous system that classifies answers and visualizes the result based on SOM. Our new system provides an incremental SOM map that is updated with progress of submissions. Teachers can make a guess about major contents of answers even during submission. When all learners finish their answer, a teacher can make effective comments immediately.

View full abstract

Ad hoc wireless networks operate without any infrastructure where a node can be a source and a router at the same time. This indispensably requires high throughput and low delay performance throughout the wireless network coverage span, particularly under heavy traffic conditions. Recent research on using multiple antennas in beam-forming or multiplexing modes over a wireless channel has shown promising results in terms of high throughput and low delay. Directional antennas have shown to increase spatial reuse by allowing multiple transmitters and receivers to communicate using directional beams as long as they do not significantly interfere with each other. However, directional antenna performance asymptotically approaches the omni-directional performance in a high density ad hoc network. Simulation results in QualNet validate that average throughput and packet corruption ratio of directional antenna approach omni-directional performance. Moreover, we further highlight some important issues pertaining to the directional antenna performance in wireless networks.

View full abstract

Many sensor network applications require that each node's sensor data stream be annotated with its physical location in some coordinate system. Equipping GPS on every sensor node is often expensive and does not work in indoor deployments. Recently, cricket-based localization system is often used for indoor localization. It is very important to know the exact position of beacons in cricket-based localization system for identifying moving sensor node's position. In this paper, a new method, Mobile Listener Detect Algorithm (MLD) which can automatically calculate the unknown newly installed beacons is proposed. For the verification of the feasibility of the proposed scheme, we have conducted several experiments.

View full abstract

In genetic and evolutionary algorithms, the goal of linkage identification is to detect strong relationships among the decision variables of the objective function. If such relations can be identified, the crossover operator can accordingly mix and recombine the identified sub-solutions without destroying them. In our previous study, we have proposed a linkage identification technique, called inductive linkage identification (ILI), which integrates the mechanisms of perturbation and decision tree induction. With the proposed technique, linkage information of the objective function can be acquired via constructing an ID3 decision tree to model the mapping from solution strings to their corresponding fitness changes caused by perturbations and then inspecting the constructed decision tree for the decision variables exhibiting strong interdependencies with one another. In this study, we observe the behavior of ILI on decomposable problems of different subproblem complexities and make an attempt to understand the population requirement of the proposed linkage identification technique. The experimental results demonstrate that the population size required by ILI to correctly learn linkage grows sub-linearly with the problem size while grows exponentially with the complexity of constituting subproblems.

View full abstract

In this paper, we propose a new Estimation of Distribution Algorithms which can cope with Reinforcement Learning Problems. Basic procedure of the EDAs is that 1) select better individuals, 2) estimate probabilistic models, and 3) sample new individuals. In the proposed method, instead of use of individuals, input-output records in episodes are directory treated. Moreover the estimated probabilistic model is regarded as policy so that new input-output records are generated by the interaction between the policy and environments. Computer simulations confirm the effectiveness of the proposed method.

View full abstract

A lot of researches applying the N-Person Iterated Prisoner's Dilemma Game have been done for simulation in human social relations. In actual society, a difference from people's thinking capability and the quantity of information that they can possess would have enormous influence. In this research, simulations were performed paying attention to this point. Therefore, in order to attain differentiation of an agent's thinking capability, strategies of deciding an action that takes into account the history of past match-ups were used. We introduce the concept of entropy and complicate the strategy. Thus, we conducted a simulation after adding a restriction to the strategy that each agent can take. In the results, it turned out that the network where agents of high and low thinking capability can leave a high score in the N-Person Iterative Prisoner Dilemma Game.

View full abstract

In many manufacturing factories, an efficient inventory control has become important due to diversification, uncertainty and instability in consumers needs. This study is optimization of a production scheduling to improve control of inventory in an actual manufacturing company which produces various kinds of primitive parts. Here, we stabilize a fluctuation of warehouse utilization of each item in the current product line to make enough margin in the warehouse for new products. In addition, we minimize the number of times of a setup operation of a machine following switching of product for efficient running of machine. Based on the modeling, Genetic Algorithm (GA) is applied to generate a production schedule focusing on downhold of overproduction with holding stable product of supply against customer needs. The production schedules acquired by GA are compared with an actual production schedule obtained by experts. The effectiveness of GA is shown in the results and the discussions.

View full abstract

In networked control systems, time-varying delay of the transmitting signal is inevitable. If the transmission delay is longer than the fixed sampling time, the system will be unstable. To solve this problem, this paper proposes the method to predict the delay using logic-based fuzzy neural networks, and the predicted time delay will be used as a sampling time in the networked control systems. To verify the effectiveness of the proposed method, the delay data collected from the real system are used to train and test the logic-based fuzzy neural networks.

View full abstract

The surface curvatures extracted from the face contain the most important personal facial information. In this paper, we developed a range face recognition method that uses back-propagation(BP) that optimizes initial parameters such as bias values or weights, which is combined by face component including the curvature attributes. In the first step, the proposed approach calculates face curvatures which present the facial features for the normalized range face image using the singular value decomposition (SVD). In the second step, PCA and Fisherface method are applied to each component range face to generate the reduced image dimension. In the last step, the back-propagation's weight is trained using the produced low-dimensional vectors and individual classifiers. The experimental results show that the proposed approach has outstanding classification in comparing with other methods.

View full abstract

In this paper, we introduce the detail technical design and development of VLBI correlation subsystem (VCS) for Korea-Japan Joint VLBI Correlator (KJJVC). Now we are developing the VCS, which is being discussed mainly by Korea and Japanese correlator teams. The VCS is able to perform the maximum 8 Gbps data rate per station, 16 stations, 8,192 channels. Several new functions and ideas are applied to the VCS, and The prototype design and development for VCS are already determined and will be completed in mid of 2008. KJJVC will cover the Korean VLBI Network(KVN), Japanese VLBI Network(JVN) and Chinese VLBI Network(CVN) including next generation space VLBI(VSOP2). In this paper, we briefly show the detail specification, technical design of VCS, and the recent progress of KJJVC development.

View full abstract

Various computational intelligences are formulated based on max-plus algebra which is composed of max and standard addition operations. They are essentially suitable for discrete information processing compared with the ordinal algebra. This paper introduces various computational intelligences based on max-plus algebra, i.e., cellular automata, percolation, and multi-agent system and complex network. As their applications, the traffic flow simulation, forest fire, and epidemic simulation are shown.

View full abstract

This paper deals with a watermarking scheme and its application to multimodal biometric recognition system using face and fingerprint. To show the effectiveness, we evaluate the various performance in terms of recognition rate as well as PSNR before and after applying the watermarking scheme. Experimental results demonstrate that the proposed scheme makes it possible to effectively hide and extract the biometric features without deteriorating recognition rate.

View full abstract

In this paper, we review the network currently in use by intelligent service robot. After derive problems of corresponding network, we review at technical requisite as control network of intelligent service robot. We proposed a method of implementation of internal communication network of mobile robots based on the photonic EtherCAT,industrial Ethernet.

View full abstract

We enforced robot assisted activities and robot assisted therapy (RAA/RAT) using a remote controlled robotic pet in the rehabilitation ward and geriatric nursing home, and have proposed a field of robot assisted rehabilitation (RAR). However, there remain still many problems to be solved in order to be accepted as an alternative medicine. For example, a robot's operator is necessary in addition to the therapist or intervener. One of the solutions is to develop a simple operation console that can be used by the therapist or the interveners by themselves. Based on this concept a hand held controller was developed and used by the therapist in the nursing home. To the ultimate extent of simplicity and easiness, the controller with motion sensing is developed and tested, which can be used by geriatric person and bring out their positiveness to the rehabilitation program. An in-house-walking program with robotic pet was carried out as a preliminary experiment and shown promising results.

View full abstract

Robot therapy for elderly residents in a care house was conducted over one year. Two therapeutic seal robots were introduced and activated for over 9 hours every day to interact with the residents. This paper presents a report of this experiment. In order to investigate the psychological and social influences of the robots, each subject was interviewed by using the free pile sort method, and their social interaction was analysed. In addition, the follow-up investigation was conducted after the robot was taken away. The results showed that the robots encouraged them to communicate with each other and then strengthened their social ties over the year.

View full abstract

We suggest a cognitive rehabilitation robotics method by using a robot for cognitive rehabilitation. We produce a rehabilitation support assembly robot, and test it at a hospital. The difficulty of the training level is set by small steps in this robot, and the user can feedback a sense of accomplishment. Because the robot is 3-dimensional, users can train it in areas of space recognition, executive function and finger behavior. From September to December 2007, we tested it on 4 people aged from 20-70 years. The highest score was obtained when we performed subjective satisfaction evaluation on a face scale.

View full abstract

A real-time performance and strictness is important to apply myoelectric signal for control of a well-fare robot such as a prosthetic hand, rehabilitation and robot therapy. Past researches also has tried to discriminate a human motion from myoelectric signal in a moment for real-time performance, so that the signal can be apply to robot control. There is a paper to describe a method to discriminate six motions from myoelectric signals within 100ms. It's probability, however, of some motions are less than about 80 %. In order to apply myoelectric signal to robot control, it is serious to discriminate a human motion correctly. Moreover it should be 100 % as closer as possible. Additionally there is a paper to report that the accuracy of motion discrimination falls when myoelectric signal comes through couple neighboring muscles is analyzed by neural network method. Therefore, it might be said that to apply a neural network method for such a problem like past trials would have difficult to reach near 100 % accuracy of discrimination of human motions from myoelectric signal. The purpose of this paper is to develop a motion discrimination method from myoelectric signal within 80 ms data from start to measure, with almost 100 % probability for a real-time performance and strictness. In this paper, the myoelectric signal generated in muscles of a forearm of right hand is analyzed using combination of morlet wavelet analysis and three-layered neural network and the motions which are "Grasp", "Flexion of wrist" and "Pronation" of a right hand are discriminated. A bipolar method is used to measure three kinds of myoelectric signal generated in three muscles related with each motion. Through the tests of 87 times for a subject and 300 times for another subject, the validity of the proposed method is confirmed. As a result, for a subject, all of three motions are identified completely without mistake, for another subject, the result showed very high identified completely without mistake, for another subject, the result showed very high identification ability of the proposed method such as 94 % for "Grasp", 100 % for "Flexion" except "Pronation".

View full abstract

This paper proposes an online text mining system that is developed based on M2VSM (Meta keyword-based Modified VSM). When conventional vector space model (VSM) is applied to document clustering, it is difficult to adjust the granularity of cluster in terms of topic. In order to solve the problem, M2VSM is proposed as an extended VSM so that it can consider meta keywords such as adjectives and adverbs, as additional value of indexing terms. The similarity between documents is calculated by considering the matching of meta keywords for each index term, which makes it possible to cluster documents with various granularities in terms of topic. The online text mining system is developed MUSASHI, which is one of the most popular open source data mining tools. By using the system, users can perform a series of text mining process online, including preprocessing, feature selection, clustering, and visualization of results. Experimental results show that clustering results by M2VSM match the results by test subjects in both rough and detailed clustering. It is also shown that the system can process database containing 5,000 documents within 7 minutes.

View full abstract

In this year, 2008, an American presidential election will be held. In this study, we examined the speeches of three candidates, that is, Barack Obama, Hillary Clinton and John McCain, to educe their linguistic features statistically. In short, frequency characteristics of character-appearance and word-appearance of their speeches were investigated. These characteristics were approximated by an exponential function: [ y = c * exp(-bx) ]. Furthermore, we calculated the percentage of required vocabulary for Japanese junior high school students and American basic vocabulary by The American Heritage Picture Dictionary to obtain the difficulty-level of each material. Besides, we examined the K-characteristic for each material. This index suggested by statistician named Udny Yule can express the richness of vocabulary in writings by measuring the probability of any randomly selected pair of words being identical. As a result, it has been shown that the speeches of Mr. McCain have the same tendency as English literature in the character-appearance, and his speeches tend to be more difficult than other candidates. In addition, the dispersions of the K-characteristic are similar to those of the coefficients c and b for character-appearance.

View full abstract

We propose the system that offers only the artical that is the relation to topics to the user in this research. When the user wants to read the article that is the relation to topics, the user must click the link to the article. Therefore, it is difficult for the user to read only the article related to topics. Moreover,there is the article that is similar to each other content or article. Therefore, user must read the article that is similar to other article. We propose the algorithm to find similar articles. For the proposed system, we use the feature of reported articles. There is an outline of the entire article at the beginning of reported articles.

View full abstract

The nursing care quality improvement is very important in the medical field. Currently, nursing-care freestyle texts (nursing-care data) are collected from many hospitals in Japan by using Web applications. The collected nursing-care data are stored into the database. Some nursing-care experts evaluate the collected data to improve nursing care quality. In order to evaluate the nursing-care data, experts need to read all freestyle texts carefully. However, it is a hard task for each expert to evaluate the data because of huge number of nursing-care data in the database. For reducing workloads to evaluate nursingcare data, we have proposed a support vector machine (SVM) based classification system. In this paper, in order to improve the classification performance, we propose a genetic algorithm (GA) based feature selection method for generating numerical data from collected nursing-care texts. First, we extract nouns and verbs from nursing-care texts using the morphological analysis software "MeCab" and store the extracted terms into a "term list". Some combinations of terms in the term list are selected by GA with two objectives; (1) maximization of the number of correctly classified texts and (2) minimization of the number of selected terms. And then, we classify the nursing-care numerical data using the SVM. From computer simulation results, we show the effectiveness of our proposed method.

View full abstract

Text mining has been growing; mainly due to the need to extract useful information from vast amounts of textual data. The target textual data of this paper is a collection of free descriptive data taken from questionnaires. Unlike research papers, newspaper articles, call center logs and web pages, which are usually the targets in text mining analysis, the free description data contained in the questionnaire responses have specific characteristics, including a small number of short sentences in each piece of data, while the varieties of contents preclude the applications of clustering algorithms used to classify the same. In this paper, we suggest the way to extract the opinions which are delivered by multiple respondents, based on the modification relationships included in each sentence in the free description data. Certain applications of our method are also presented after the introduction of our methodology.

View full abstract

Through recent developments in information and communication technologies, dynamic monitoring and control of transport systems are technically possible. These technologies enable constructing more flexible and cost-effective transport services which may vary based on demand. Since passenger demand for public transports has been declining in rural regions, DRT (Demand Responsible Transport), which provides transport service in response to the requests, is expected to have an important role as new public transport system to fulfill the mobility gap between taxi and bus. The route and departure time of DRT changes according to each reservation. Therefore passenger's mode choice influences on the service level. On the other hand, the mode choice is also influenced by the experienced service level. One might not want to use the DRT service any more if travel time is too long to accept in one day. After all, one's decision making influences on others through the DRT system. To explore these phenomenon, this study attempts to develop a multi-agent social simulation considering consider passengers? mode and departure choice learning to evaluate the DRT system.

View full abstract

Travel time of public transport such as railway and Light Rail Transit (LRT) is more reliable than road traffic. When assessing the effect of railway transits, the reliability should be considered. Otherwise, the effect may be underestimated. Then, we should not only treat both transit and road users simultaneously and consistently, but also travel time uncertainty of cars and the reliability (or punctuality) of railway for exact evaluation. A network equilibrium model with combined mode and route choice considering road travel time uncertainty does not necessarily have a unique solution, but may have many local solutions, because of the interaction between modes. The simulated annealing is one of the useful techniques to such a problem. to the calculation of the network equilibrium model with combined mode and route choice. In this paper, we apply the network equilibrium model proposed to the Kanazawa urban area using the simulated annealing technique.

View full abstract

Many complex aspects can be observed in urban transport planning. The study aims at showing the application of the artificial society approach to the problem of urban transport planning. The specific distribution can be observed as an emergence through the interaction of agents. The agents are defined to move in the society according to the action rules depending on the acquired information and knowledge of experience. The decision of each agent for day to day travel is formulated in the agent model corresponding to the urban transport planning. Each agent should have different preference to the alternatives of trips. On the other hand, traffic flow on urban network should be estimated summarizing the individual behaviours of commuter in the integrated model. Therefore, the simulator of artificial society consists of two interactive procedures of travel behaviour estimation and agent daily decision. Furthermore, the daily transport condition can be monitored by the illustration of GIS software. It is helpful to discuss the effective transport planning in artificial society.

View full abstract

The roadside parking of vehicles severely affects traffic flow conditions by increasing congestion, travel times etc. on urban roads as it reduces road capacity and need to be carefully examined to propose appropriate solution. Many microscopic traffic simulation (MTS) models have not analyzed driver behaviour around roadside parking areas deeply. In the present study, a separate parking area movement model has been proposed for MTS considering partial lane change behaviour which has been identified to describe driver behaviour especially at roadside parking areas. A suitable technique namely fuzzy logic has also been proposed to represent realistic driver behaviour around roadside parking areas considering uncertainties and perceptions because of human element and other complexities in modelling. The comparison between estimated and observed data on Gifu city network indicated that fuzzy reasoning approach has produced realistic driver behaviour than standard model. By applying proposed fuzzy logic based parking movement model in MTS, relevant transport policy namely on-street parking restriction has been evaluated.

View full abstract