This paper focuses on multimedia positive computing, which is positive computing using multimedia data. Presenting a piece of multimedia data induces a person to have a mood corresponding to its impression. According to the impression transition of multimedia data, a person's mood is looking forward to being transitted. A transition method for the impression of multimedia data in the impression space has been proposed. This method decides intermediate impression points between the impression points corresponding to the initial and the destination moods for a gradual mood change. This paper revises the transition method. The revised method selects different impression points as many as possible. This enables a smoother transition in the impression space. The values of impression points and curvey transitions in the affective and impression spaces are also examined.

This paper clarifies the semantics of the specification of the impression evaluation method by space (IEMS), a paint-based graphical evaluation method. The IEMS uses a plane containing impression words. The impression of an object is specified by circling the areas matching the impression. The degree of the impression is expressed by the painting's color. The semantics are clarified based on the peaks of the painted area. It is shown that closed curves may not have meanings when peaks are just on impression words. The painted areas other than peaks are not considered by the peak-based semantics.

This paper applies a method of using a paint-based graphical user interface (PGUI) of the multimedia retrieval system based on the impression to multimedia data retrieval based on impression words. The PGUI is based on the impression evaluation method by space (IEMS). The IEMS uses a plane containing impression words as the impression space. The impression of an object is specified by circling the areas matching the impression. The degree of the impression is expressed by the painting's color. The weights of impression words specified by the PGUI are converted to text strings. A search engine is invoked by using these text strings. Execution examples of PGUI prototype system executions are demonstrated to show the feasibility of this prototype system.

In the National Institute for Fusion Science, high-temperature plasmas have been produced in the Large Helical Device (LHD). A large number of videos having plasma discharge are captured during nuclear fusion operations. An observation of these recorded videos can lead to a new discovery and guarantee the safe operation of the experiments. In order to optimize the operational parameter of the experiment, the future emission states of plasmas should be estimated. Thus, this paper proposes a method to make a prediction of sequential future frames containing plasma light emission using the Convolutional Long Short-Term Memory (ConvLSTM). We confirmed that the future frames of plasma discharge could be well predicted with the satisfactory results of Mean Squared Error as 2.78×10-5and Cosine Similarity as 0.9997.

This paper proposes a method of weighting an adjacency matrix for the prediction of association between lncRNAs and diseases based on experimental evidence. A lncRNA-disease adjacency matrix is weighted by experimentally validated times. As a result, we showed that the proposed method provides good predictive performance and improves adequacy compared with TPGLDA [1].

Facial expression intensity has been proposed to digitize the degree of facial expressions in order to retrieve impressive scenes from lifelog videos. However, correlation of facial features compared to each facial expression is not determined objectively. Therefore, we recognize facial expressions by using a neural network and calculate the contribution score of input toward output. We verify the score correctly by comparing the accuracy transitions depending on reducing useful and useless features, and process the score statistically. As a result, we extract useful facial features from the neural network.

This paper improves the parallel multi-dimensional indexing system which stores data into several multi-dimensional indexes and retrieves required data from them in parallel. This system treats a set of data as a rectangle, which covers the data. The rectangle is inserted into an index. We must consider two important points. One is that the area of each index covers as wide as possible so that many indexes could be used in the retrieval. The other is that the number of rectangles which each index has is the same as possible so that the time in finishing the retrieval on each index could be the same as possible. The conventional system selects an index to which a rectangle is moved by using the area that the index covers and the number of rectangles that the index has. The conventional system first evaluates the degree of the expansion of area and then evaluates the number of rectangles. This paper changes the evaluation order. The number of rectangles is firstly evaluated. It is experimentally shown that the new evaluation order gives us good retrieval performance.

This paper proposes the system realizing integrated use of Relational Databases and a NoSQL Database. A heterogeneous information source integration system has been proposed. It enables equi-join and projection of tables without converting data. With this system, it became possible to handle various databases at the same time by using JDBC, but the databases which can be handled are only relational databases. Therefore, we add functions to the system to handle the NoSQL database which has been increasingly used in recent years. The table join execution time and the memory consumption of the system are experimentally evaluated. It showed that the performance is tolerable in practical use as compared with integrated use of only relational database. We also showed that it is practical to change the join order or to register column families which are unique elements of HBase.

The format of the video evaluation questionnaire is varied. The user can search high-evaluation videos by referring to the video evaluation, and the video producer is motivated to produce the video by getting a good evaluation, and by analyzing the evaluation, they can get clues to make a better video. However, the evaluation can be affected by several elements which are not directly related to the video itself. In addition, not all of the users evaluate the video because the evaluation is voluntary. Hence, it is difficult to obtain accurate video evaluation. This research focuses on the importance of the fun of the video in evaluating the video. We calculate the degree of smile from the face of the user at the time of viewing the video by the expression intensity estimation, and use it to evaluate the video automatically. First, smiley face images and the expression intensity of the face image are obtained by subject experiments. Using this data, an expression intensity estimation model by a support vector machine is created. Next, the expression intensity is calculated for each frame from the face video of the person at the time of watching the video, and it is linked to the evaluation on the video. As the performance evaluation, the video evaluation obtained through the questionnaire is compared with one obtained by the proposed method. The comparison result indicates that the expression intensity is useful for the video evaluation.

This paper proposes a new method of coloring terms in presentation slides. The proposed method consists of two ranking methods. One is based on TF-IDF and the other is based on cosine similarity between a term and a presentation title. The proposed method tries to extract important terms from slides written in Japanese. After that, these terms are colored. It is shown the proposed method colors terms more properly than the previous one.

To eliminate mismatches between the intentions of questioners and respondents of Question and Answer (Q&A) sites, we have clarified that the impression of the statements could be captured by nine factors, and the factor scores could be estimated from the feature values of the statements. Objective scores of the statements could be estimated fairly good, and that those of subjective statements could be estimated well by taking the natural logarithm of the factor scores with the consideration of cross-validation. This paper tries to perform multiple regression analysis with the consideration of semantic similarity between Q&A. A statement is represented with an average word vector of the words appearing in the statement. The semantic similarity of two statements is calculated through the cosine of the two average word vectors of the two statements. As a result, there could be a possibility that semantic similarity would be effective in estimating objective scores.

This paper describes analysis of perception of external people of school about school mottos. Most of schools show their school motto on their web page to advertise their educational policy and philosophy. However, school mottos are sometimes created to make a guideline of behavior, attitude, and thought for their students to be used in education purpose. In this paper, we survey the school motto to various kinds of people and clarify the difference of perception of people. Particularly, we divided two groups who answered our survey. One is non-academic worker and other is academic worker. Since their career and experiences are difference with each other, the perception about some school mottos is different for them. Furthermore, when the school motto contains Cristian keywords, people have similar perception and make similar image to the school. The contributions of this paper include (1) what kind of school motto attracts and improves the image of the school, (2) Appropriate school motto to make better business to get a lot of applicants to enter the school, and (3) what kind of school motto should be invented and created to be understood easily by all people.

This paper tries to estimate the amounts of sewage flow, which is one of daily life data, for their efficient management. The non-rainy day having good data is called a normal day. Normal days are divided into typical days and irregular ones based on the existence of outliers. Typical days are classified into regular days and non-regular ones also based on the existence of outliers. The amounts of flow on regular days are almost those of the typical day. The amounts of flow of a typical day are tried to be adjusted to those of a non-regular day. The values for the adjustment are tried to be estimated by using the multiple regression analysis. It is shown that these values can sufferably be estimated based on temperature and the amount of rain fall.

This paper describes analysis of perception of people who do not concern in the mottos of schools. We can find school mottos on the web and see advertising their educational policy and philosophy. Students should act as “a student” because some schools make school mottos rules or disciplines of behavior, attitude, and thought for their students to be used in education purpose. On the other hands, school motto is one of a strong message which has role of advertisement for external people of the school. In this paper, we survey the school mottos to external people of the school and clarify the difference of perception of people. In our analyses, we compared perceptions in marriage and place of living. Respectively, we divided two groups, married and unmarried and living in metropolitan area and local area of Japan. Since their experiences are difference between each other, the perception about some school mottos is different for them. Furthermore, when the school motto contains Cristian keywords, people have similar perception and make similar image to the school. This paper includes these three contributions; (1) Analysis where external people of school are attracted by school motto, (2) Clarifying appropriate school motto depends on places of school to improve their business to get a lot of applicants to enter the school, and (3) Clarifying what school motto understood easily by several types of people.

This paper integrates the human database with the drosophila one, and uses the integrated database to acquire new terms about human diseases/traits automatically. This paper uses ortholog to show the relationship in the gene of two kinds, and compares phenotypic terms of the human and these of the drosophila to discover new terms about human diseases/traits. This paper shows the databases used in the study, the method using these databases, the data acquired from databases, and new phenotypic terms can be obtained.

This paper tries to clarify the retrieval and playback method considering temporal harmony. In this study, change points of the direction of object movement and the cycle of the motion are used as the accent points of a video clip. Beats are used as synchronization points of a music clip. The music measure is extracted in order to synchronize with the beat at the beginning of the measure. The synchronized playback method to make retrieved data impressive is also proposed.

This paper proposes a method to convert a factor score matrix for the cooperation between multiple video retrieval systems based on impression. A matrix transforming the factor loading matrix is used. As a result, we showed that the results of two applications of the factor analysis can satisfactorily be transformed by using the transformation matrix.

Lifelog videos, which are the records of a person's daily life as video data, draw attention since various recording devices such as smartphone become widely used. Due to privacy reasons, lifelog videos are usually managed by the owners of the videos although the amount of video data is quite large. This makes it quite difficult to find interesting and impressive scenes from lifelog videos. In order to facilitate the retrieval of such scenes, we develop a video scene retrieval system for lifelog videos. We focus on emotional scenes as interesting scenes taking into consideration the strength of emotion derived from the intensity of facial expression. Our retrieval system thus displays emotional scenes of various intensity to users in a comprehensible way. We clarify the usability of the retrieval system through an evaluation experiment.

Similarity retrieval method of plasma emission videos using color and luminosity information has been proposed. This paper revises this method to introduce time series similarity. Plasma emission videos are clustered by using several distances proposed and revised in order to categorize them. The soundness of the proposed methods is clarified through the experiment.

This paper proposes a method for measuring human body movement similarity by considering human body motion as a set of 3-dimensional curves representing the changes of direction. This analysis method is used for obtaining similarity values of the body movement of various subjects. Firstly, the performing speed of movements are compared and adjusted. Next, two types of normalized methods, which are based on the Space Harmony Theory, normalize the human postures in motion data frames. Then, the direction sequences describing the motion paths are derived. After that, an improved method inspired by the dissimilarity measure of the 3D curves is used for measuring the similarity of human body movements.

A mutual retrieval system based on the impression for the multimedia data on the Web is realized. Although this system can retrieve pictures having the impression similar to that of a music piece, it cannot follow the impression change of the music piece. This paper proposes a method to detect change points of the impression of a music piece. The proposed method uses factor scores of the impression of audio materials. It is shown that the proposed method can detect the change points of the impression at the time when the impression subjectively changes.

This paper tried to find the cause of fear by horror games. We prepared two types of horror games whose reality of pictures are changed, and observed the change of autonomic nervous function and the amount of perspiration. As a result, the physiological response to the horror game which has real pictures was scarier. Also, people who are familiar with horror games are scared at several horror scenes in the horror game, whereas the others are not. Additionally, men are afraid of the horror games of real pictures, while women feel scared of horror games of simple pictures.

This paper elaborates the similarities and differences of impression of pictures between Japanese and Chinese. We conducted a questionnaire survey of 32 Chinese college students, asking them to evaluate 40 pictures from their feelings. By comparing the results of Chinese and Japanese, we found several interesting conclusions. First of all, both Chinese males and Japanese males have come up with 5 identical main factors for pictures, but these main factors have different degrees of influence. Second, Chinese females only have four main factors for pictures. Third, we compared Chinese males with Chinese females and then found their respective main factors. Lastly, by changing the way of the factor rotation, the correlation between the main factors was found.

The purpose of this paper is to compose music clips satisfying selected ambience on intended melody easily, and to get satisfaction of self-making. The targets are beginners of music theory. To achieve those purposes, we propose the automatic composing system that takes impression words and outlines, which represent melody's profile lines, as inputs. This system makes several arranges and modification to the initial melody based on the music theory. It is experimentally clarified that the music clips composed by the system are satisfactory, and their moods fit to the impression words specified.

Summary The cyber security threats from phishing emails have been growing buoyed by the capacity of their distributors to fine‐tune their trickery and defeat previously known filtering techniques. The detection of novel phishing emails that had not appeared previously, also known as zero‐day phishing emails, remains a particular challenge. This paper proposes a multilayer hybrid strategy (MHS) for zero‐day filtering of phishing emails that appear during a separate time span by using training data collected previously during another time span. This strategy creates a large ensemble of classifiers and then applies a novel method for pruning the ensemble. The majority of known pruning algorithms belong to the following three categories: ranking based, clustering based, and optimization‐based pruning. This paper introduces and investigates a multilayer hybrid pruning. Its application in MHS combines all three approaches in one scheme: ranking, clustering, and optimization. Furthermore, we carry out thorough empirical study of the performance of the MHS for the filtering of phishing emails. Our empirical study compares the performance of MHS strategy with other machine learning classifiers. The results of our empirical study demonstrate that MHS achieved the best outcomes and multilayer hybrid pruning performed better than other pruning techniques. Copyright © 2016 John Wiley & Sons, Ltd.

This paper tries to generate an appropriate and valid concurrency control program by using genetic programming (GP). In GP, a program is represented with a tree. By using shared control variables, an appropriate program could be generated. However, invalid concurrency control program has been generated. This paper improves the concurrency control program generation system in order that the program generated can detect deadlock and can execute any schedule. It is experimentally shown that an appropriate concurrency control program can successfully be generated by using the proposed method.

This paper proposes a music similarity retrieval method considering the shift of the pitch and the difference of tempos of music pieces. These are not considered in the current various services yet. The proposed method can retrieve music pieces arranged from the original ones as similar ones. The proposed method uses the pitch information and the length of the sound. The pitch and the tempo are normalized in calculating the similarity of music pieces. The effectiveness of the proposed method is experimentally shown.

This paper tries to obtain the speed of a video, where one or two objects and a background move with various speeds and in various directions. In this study, we measure their speeds by using the beats per minute (BPM). We find out the relationship of objects and a background through several experiments. In the case that only objects move, the average speed of objects becomes the speed of a video. When objects and the background move at the same time, the speed of the background influences the speed of the video. We find that the influence of the background speed decreases according to the increase of the object speed.

This paper proposes the methods for improving performance of multi-dimensional index system. In the previous system, data are inserted into a reception-node, and are stored in a reception-index. When a reception-index stores a certain number of data, the reception-index is moved to a normal-node, and becomes a partial-index. At this time, insertion is paused. In addition, if the numbers of data in a normal-node are skewed, the retrieval performance degrades. This paper proposes an architecture enabling data to be inserted without a pause. Moreover, when a reception-index is moved to a normal-node, all normal-nodes are tried to store partial-indexes equally. It is experimentally clarified that the insertion time decreases by the proposed architecture. It is also clarified that partial-indexes can be distributed equally based on the proposed method, and that the retrieval performance can be improved.

Facial expression recognition (FER) has been widely studied since it can be used for various applications. However, most of FER techniques focus on discriminating typical facial expressions such as six basic facial expressions. Spontaneous facial expressions are not limited to such typical ones because the intensity of a facial expression varies depending on the intensity of an emotion. In order to utilize FER for real-world applications, therefore, it is necessary to discriminate slight difference of facial expressions. In this paper, we propose an effective FER method to recognize spontaneous facial expressions using ensemble learning which combines a number of naive Bayes classifiers. In addition, a method to estimate the intensity of facial expression is also proposed by using the classification results of the classifiers. The effectiveness of these methods are evaluated through an FER experiment and an experiment to estimate the intensity of facial expressions using a data set including spontaneous facial expressions.

This paper proposes the colors of characters suitable to presentation slides. We propose five color combinations. A combination has three colors of base, main, and accent colors. The base color is for the background. The color white is used as the base color. The main color is for the part becoming the frame. The accent one is for the place where we want to emphasize. The lightness differences between the base color and the others are considered. The color difference between these two colors is also considered in order to attain good visibility. The slides colored by the proposed color combinations are presented to demonstrate their effectiveness.

In the nuclear fusion experiment, a lot of videos containing plasma emission phenomenon are obtained. It is required to develop the system retrieving similar videos. This paper proposes the color distance and the similarities between shots and videos. Those similarities between shots and videos are calculated by using hash values obtained from frames. The color distance is calculated by using color differences Cb and Cr. From the experimental result, it is clarified that the proposed methods calculating similarities are effective in retrieving similar plasma videos.

This paper proposes a method of similarity video retrieval based on impressions. As it has been clarified that the impression of videos can be expressed by using five factors, factor scores are tried to be estimated through feature values of videos. As for the dissimilarity, eigen values of the factors are introduced as weights. Similarity retrieval process is divided into two steps. At the first step, videos are roughly divided into two categories: similar and dissimilar videos by using the factors whose eigen values are large. At the second step, dissimilarity is calculated from the factor scores of the remaining factors. Moreover, the threshold values for the similarity retrieval are also clarified. It is experimentally shown that videos similar to a key video can successfully be obtained by using the proposed method.

In order to eliminate mismatches between the intentions of questioners and respondents of Question and Answer (Q&A) sites, nine factors of impressions for Japanese statements have experimentally been obtained. Nine factors have also been obtained from the impression of English Q&A statements. This paper estimates factor scores of English Q&A statements through multiple regression analysis. These are words and characters, syntactic information, and appearance percentages. It is shown that estimation accuracies of all of the nine factors are very good.

This paper proposes a search method of optimal movements in multi-player games with imperfect information in order to implement a Mahjong player exceeding human top players. The proposed method uses tabu search. It does not use the record of a game. Tabu search is applied to the states of discarding tiles and using the tile that other players discarded. The experiments were carried out in order to evaluate the proposed method. In the experiment of evaluating the rate of concordance of the usage of tiles, the maximum rate of concordance reached to 83%. This means effective winning hands could be found in the initial states. In the experiment of playing a game with benchmark players, it is shown that the proposed method is better than benchmark players. From these results, the possibility of the effective search of optimal solution by using tabu search was indicated.

This paper extends the semantic generalization for having the capability of the class-based one. The attributes common to subclasses are inherited upward to the superclass in the ordinary generalization. The attribute name of the superclass is the same as those of the subclasses. The semantic generalization makes it possible for a class to have attributes by considering the semantics of the attributes of subclasses even if there is no common attribute in the subclasses. The attributes generalized can be specified by users rather than automatically decided. This specification sometimes becomes cumbersome. On the other hand, the names of the attributes of the superclass are automatically changed according to the name of the superclass in the class-based generalization. In the extended semantic generalization, the names of the attributes generalized can automatically be decided according to the superclass name. Those are also possible to be specified by users as in the original semantic generalization. This improves usability of the semantic generalization considering a class-lattice.

This paper clarifies the characteristics of analysis methods of the evaluation results obtained through the Impression Evaluation Method by Space (IEMS). The IEMS uses a plane containing impression words as the Kansei space. The impression of an object is specified by circling the areas matching the impression. The degree of matching the impression is expressed by painting color. Three analysis methods have been proposed for the IEMS. These use the baseline Kansei space, the impression words, and the peaks of darkness of the evaluation results, respectively. These methods are compared from the viewpoints of spatial evaluation, relationships among impression words, darkness of evaluated areas, impression words added, and impression words moved. Moreover, it is clarified the information obtained through the three analysis methods. Each of them is suitable in obtaining specific information. They should be used according to the information required to be obtained.

In analyzing big data, data are tried to be categorized into several classes based on their similarity or dissimilarity. As the success of the analysis depends on dissimilarity, dissimilarity plays a very important role in analyzing data. Dissimilarity is also used in the similarity retrieval. This talk focuses on time-varying data; time-series data and video data. Several similarity retrieval methods of timevarying data are surveyed with consideration for big data analysis. These are categorized and explained according to the domain where dissimilarity is calculated: the time domain and the frequency domain. Characteristics of the methods are described. Dissimilarity based on the visual aspect of time-varying data is also referred. This could improve precision of the retrieval.

This paper introduces the concept of fields for the purpose of increasing the efficiency of the analysis of big data. We focus specifically on time series data. Data are treated as points in a space. A field is a named subspace within that space. A field may restrict the position of a point. The subspace of a field may change according to the points included in the field. It may also be nested. After formally defining the concept of a field, we describe an approach to processing big data that incorporates this notion. By assigning a field to a meaningful portion, we can treat only the portions that we are interested in. As this reduces the amount of data processed, it results in the efficient processing of big data.

This paper clarifies the effect of color on the impression of a character. In experiments, we used four letters (B, H, N, 0) as the characters. These characters are considered to represent groups categorized by shape. Black, blue, green, red, and yellow were the colors tested. The colored characters were evaluated through 24 impression word pairs. The results showed a significant difference between men's and women's impressions. It was also shown that the impression of characters can be expressed by three factors, Elegance, Liveliness, and Potency, for both men and women. The importance of the factors differed for the men's impression when compared with the women's impression. The first factor of the women's impression was Elegance, whereas the first factor of the men's impression was Liveliness. It was also shown that the color is more important than the shape of characters. The impression of black and yellow in Liveliness and green in Elegance varied for women, and the impression of blue in Elegance varied for men.

This paper proposes a method for selecting respondents who can give an appropriate answer to a question, in order to eliminate mismatches between the questioners and respondents at question-and-answer sites. The possibility of detecting respondents capable of appropriately answering a newly posted question is examined and confirmed. Based on this, the proposed method uses the number of appearances of each respondent and scores them based on the distance between the factor scores of a question and previously posted answers. Impressions of statements were obtained for nine factors. Factor scores were estimated using multiple regression on the feature values of the statements. This method was evaluated by experiments that compared its precision and recall with those of methods based on average scores and distances. It was shown that the proposed method outperforms these other methods. It is also shown that, for a given question, the proposed method can successfully select those respondents who are more appropriate than around average.

This paper proposes an analysis method of the evaluation results obtained through the Impression Evaluation Method by Space (IEMS). The IEMS uses a plane containing impression words as the Kansei space. The impression of an object is specified by circling the areas matching the impression. The degree of matching the impression is expressed by painting color. As the impression words can be moved and/or added in the IEMS, it is difficult to analyze the evaluation results obtained from many subjects. The proposed analysis method focuses on the peaks of darkness. It is called the analysis method focusing on the peaks of darkness (abbr. AM_PD). By mapping the peaks of the darkness in each evaluation result to the same Kansei space, this method can analyze characteristic impressions. In this paper, the algorithm of extracting obvious peaks automatically is proposed toward realization of the AM_PD. The parameters of the AM_PD required to extract the obvious peaks are experimentally determined. This paper shows that the obvious peaks in the evaluation results can be extracted by using this algorithm.

It has been shown that pauses, which are considered to be important for transmitting traditional skills, can easily be obtained from the movement of a worker or a player by adjusting several parameters of the similarity test. This paper quantitatively confirms the validity of the values of the parameters. By using these parameter values, pauses in tea ceremony are tried to be derived. It is shown that the movements of a body around pauses are quite similar, and may form a typical and/or fundamental movement. The fluctuation of tempo is also examined from the viewpoint of timing of beats.

To eliminate mismatches between the intentions of questioners and respondents of Question and Answer (Q&A) sites, we have clarified that the impression of the statements could be captured by nine factors, and the factor scores could be estimated from the feature values of the statements. Objective scores of the statements could be estimated fairly good, and that those of subjective statements could be estimated well by taking the natural logarithm of the factor scores. This result, however, is likely to estimate inaccurate values due to overfitting to the explanatory variables. This could result from not having considering cross-validation through the attempts. This paper tries to perform multiple regression analysis with the consideration of cross-validation. For statements of Yahoo! Auction, PC and love counseling & human relationships, respective five trials have been attempted. As a result, most of the attempts show good or fairly good estimation accuracy. It is suggested that estimation errors should be reduced.

For the purpose of an efficient retrieval of impressive scenes from lifelog videos, we propose an emotional scene detection method based on facial expression recognition. Most of conventional facial expression recognition methods focus on dis- criminating typical facial expressions such as happiness, sadness and surprise, while lifelog videos contain various facial expressions. In addition, many training examples, which are quite troublesome to prepare, are required to construct the facial expression recognition models. The proposed method tries to solve these problems by constructing a facial expression recogni- tion model using an unsupervised learning based on Gaussian mixture model. Since our model is unsupervised, there is no need for preparing learning examples and predefining the types of facial expressions. The detection performance of the proposed method is evaluated in terms of detection accuracy and efficiency through several emotional scene detection experiments.

The semantic generalization and the semantic specialization have been proposed. The semantic generalization makes it possible for a shape graph, which corresponds to a relation schema in the relational data model, to have elements by considering the semantics of the elements of the original shape graphs even if there is no common element in the original ones. The element is the lowest common ancestor of the elements of original shape graphs in a hierarchy of concepts rather than their common element. The semantic specialization enables a shape graph to have elements which are different from those of the original shape graphs, but are semantically related to them. This paper extends the semantic generalization in order that the generalized shape graph can be created according to the user's specification. This enables users to create the generalized shape graphs that users want as in the sematic specialization. This extension also brings us unification of the definitions of the semantic generalization and the semantic specialization.

This paper tries to clarify various pauses, which are called "Ma" in Japanese, in performances in order to employ them in transmitting traditional skills from the current generation to the next one. Various pauses are tried to be identified from speeches, music pieces, and actions. A tea ceremony, the knitting of wire netting and the Mie of Kabuki performance are treated as actions. Pauses are classified to several groups from the viewpoints of the presence of sound, that of movement, that of a pose, the time duration, and the aim of a pause. It is also pointed out that there is a negative pause. From the point of view of time duration, there are three types of pause: short, normal, and long pauses. It is shown that short pauses should precisely be analyzed, and be effectively used in transmitting traditional skills because they often appear in the movements of traditional works.

In order to retrieve impressive scenes from life log videos, we propose an emotional scene detection method based on facial expression recognition. Many researches about facial expression recognition focus on discriminating typical facial expressions such as happiness, sadness and surprise. But they are not suitable for life log videos because more complicated or subtle facial expressions are frequently observed in them. The proposed method tries to solve this problem by constructing a facial expression recognition model based on unsupervised learning. It discriminates facial expressions and detects emotional scenes by a clustering approach using facial features based on the positional relationships of several facial feature points. This approach is fully flexible to detect various facial expressions from life log videos because it does not need to predefine the facial expressions. The detection performance of the proposed method is evaluated in terms of detection accuracy and efficiency through the emotional scene detection experiments.

To eliminate mismatches between the intentions of questioners and respondents of Question and Answer (Q&A) sites, we have clarified the characteristics of the question and answer statements. It has been shown that the impression of the statements could be captured by nine factors, and the factor scores could be estimated from the feature values of the statements. Here, the objective scores of answer statements are provided. This paper tries to estimate the objective scores of answer statements through multiple regression analysis. They are estimated from the factor scores estimated by using multiple regression formulas already obtained. Absolute values of the differences between the factor scores of answer statements and those of question ones, those of the answer statements having the highest scores, and those of the ones having the lowest ones, are used as the explanatory variables. It is shown that objective scores of the objective statements could be estimated fairly good, and that those of subjective statements could be estimated well by taking the natural logarithm of the factor scores.

This paper tries to generate an appropriate concurrency control program by using genetic programming (GP). In GP, a program is represented with a tree. Nodes of a tree are selected from a symbol set. This paper tries two symbol sets: the high-level symbol set and the reduced one. The high-level one includes high-level symbols created by combining conventional ones. In the reduced symbol set, symbols are drastically decreased by changing the method of implementing the concurrency control program. Automatic defined functions (ADFs) are also used. Introducing high-level symbols caused the increase of the number of symbols. This made the program generation difficult. On the other hand, an appropriate program could be generated with the reduced symbol set. An ADF is also used in the program generated.

This paper proposes a method for evaluating impressions using a space containing impression words. The impression of an object is specified by circling impression words that match the impression. The degree of matching of the impression is expressed by the darkness of the color used to paint the circled area. This proposed method poses few restrictions to users evaluating an impression. Conversely, the semantic differential (SD) method evaluates an impression in a predefined range to enable statistical processing. A prototype system is implemented to evaluate the proposed method, and experimentally compare it with conventional methods, such as the SD method. Finally, it is demonstrated that the proposed method can evaluate aspects that other methods cannot.

This paper uses Postgre SQL, a database management system, in the data management of the similarity retrieval of subsequences of waveforms. By using Postgre SQL, the management of data could become easy, and a multi-dimensional index structure named the R tree could be used. This paper examines the parallel processing of the similarity retrieval by dividing a table into several tables. It is shown that high retrieval performance can easily be obtained by using a high performance computer. It is shown that using Postgre SQL could make the division of a table easy because R tree indexes are automatically constructed.

For the purpose of retrieving impressive scenes from lifelog videos based on facial expression recognition, an effective equipment to take facial videos is developed in this paper. The important advantage of this equipment is that it is able to take facial videos without excessive fatigue. In addition, we implemented an emotional scene management system. It provides the functions to store facial videos to a video database and to retrieve and display emotional scenes through Web browsers. In order to measure the retrieval performance of emotional scenes, a retrieval experiment was conducted. It was experimentally shown that about 78.6% of the emotional scenes were retrieved successfully.

This paper discusses the stimulation of the subject's strong emotions in presenting audiovisual materials. The mental states aroused are charity, nostalgia, nobility, and passion. Audiovisual materials are presented to subjects. The degrees of the mental states with strong emotion and the kinds of mental states are experimentally obtained. The experimental results show that different types of emotion were evoked by presenting audiovisual materials. The tendency of arousing the mental states with strong emotion is examined from the tendency of audiovisual materials used in the experiment.

This paper proposes the semantic generalization. This generalization makes it possible for a shape graph, which corresponds to a relation schema in the relational data model, to have elements by considering the semantics of the elements of the original shape graphs even if there is no common element in the original shape graphs. For this purpose, viewpoints and relationship trees are introduced. A viewpoint is a sub graph of a schema graph, and plays a role of a kind of hierarchy of concepts. A relationship tree is a hierarchy of relationships. The semantic generalization is informally and formally described. It is shown that the conventional generalization could be realized by using the semantic generalization.

In order to avoid the problem of mismatch between the questioner and the respondent, we have conducted impression evaluation experiment and nine factors are obtained as a result. Then factor scores of any other statements have been tried to be estimated by using multiple regression analysis@from feature values of statements. By adopting syntactic information of statements, word image ability and closing sentence expressions as the feature values, all the factor scores were well estimated. This paper tries to explain the estimation result with major feature values. The explanation is confirmed by comparing the features of the statements having high and low factor scores with the major features.

The effect of the harmony of the audio clip and the video one of an audiovisual clip on the estimation of its score has experimentally been clarified for the cross-media retrieval. Two estimation formulas have been proposed. One is of the estimation of the scores of the impression of audiovisual clips, and the other is of the estimation of the score of the total harmony based on the scores of static harmony and dynamic one. This paper confirms the first formula through another experiment. It is shown that the formula estimates the scores well. This paper also proposes the formula estimating the scores of the static harmony of audiovisual ones by using the scores of the impressive harmony and the semantic harmony.

This paper proposes a generation system of concurrency control program by using genetic programming (GP). This system generates concurrency control program according to the features of transactions, which are collections of database operations. Functions and terminals of trees representing program in GP, and the fitness measure function used in GP are proposed. The functions and the terminals include those changing and testing variables attached to data items and transactions as well as those checking the kind of operation etc. These will bring us general concurrency control program, which is beyond the combination of the parts of traditional concurrency control program. As the granularity of the functions and the terminals is small, the sub-trees, which are used for the popular concurrency control protocol, and are prepared in advance, are used. The fitness measure function considers the goodness of concurrency control program. The experiments show that a concurrency control program using locks could be generated under the concurrent environment, while a concurrency control program better than the two-phase locking protocol could be generated under the not-so-concurrent environment.

This paper proposes a framework for receiving the services on the Internet with keeping the information private. This paper treats the services handling the aggregate information rather than the information itself. In this type of service, three-layer information structure is commonly observed. This structure consists of whole, unit, and target layers. Client computers send the information corresponding to these layers and their meta-data if required. In this paper, the interface between client and server computers is described in the form of XML. The functions that should be provided by the client and the server are also clarified.

The effect of the harmony of the audio clip and the video clip in an audiovisual material on the estimation of its scores is experimentally clarified for the cross-media retrieval. In the experiment, four patterns of audiovisual materials are used. These are the statically matched or mismatched, and the dynamically matched or mismatched audiovisual clips. This experiment clarifies the degree of the contribution of the harmony of audiovisual clips to their scores. It also clarifies that the degree of total harmony of the audiovisual material could be estimated by using the degree of the static harmony and that of the dynamic harmony.

The paper clarifies that the main factors of sensitivity work well for accessing multimedia data. The main factors of sensitivity are the fundamental factors of human sensitivity. These are obtained by applying the factor analysis of the scores put to multimedia data for impression words. Five main factors are obtained for pictures, and four for sounds. Three main factors of sensitivity are in common for pictures and sounds. Substituting the feature values of multimedia data, which are called sensitivity measures, for the factor scores enables a lot of multimedia data to be retrieved through the impression words such as beautiful and fresh. Moreover, the sounds corresponding to a picture can be obtained by evaluating whether the sensitivity measures of sounds are similar to those of the picture. This function is called mixed-media navigation. A prototype system is constructed and evaluated. The evaluation results show that the prototype system can retrieve suitable multimedia data.

The structure of semistructured data is needed in order to query a large amount of semistructured data, although structure cannot be defined a priori. DataGuides and shape have been proposed to represent the structure of semistructured data. As these can behave as schema from the point of view of representing the structure of data, and are changed according to the existence of objects, these are called dynamic schema. The paper proposes a semi-dynamic construction method for the dynamic schema. The proposed method uses versions of the dynamic schema. Several versions are kept in the system. Each version of the dynamic schema keeps the objects required to construct the next version of it in a list. The correct dynamic schema can easily be constructed by using the versions of dynamic schema and the lists of objects. The performances in deriving the shape, and inserting and deleting objects are experimentally evaluated. Experimental results show that the proposed method is effective in the case where the retrieval occurs more frequently than the deletion.

This paper proposes a storage structure for graph-oriented databases called the flattened separable directory method. In this method, a data representation graph, which is a unit of the represention graph, is primarily represented with an array of edge or node types. As every node or edge can be accessed without navigation, the values of nodes and/or edges can be quickly evaluated. Experimental evaluations support these characteristics, and clarify that the data insertion performance is high, and that less storage overhead is needed for graphs consisting of many node and edge types.

This paper proposes the kernel data model called a named set model for bottom-up databases. This model represents an element with a triplet of its identifier, its name, and a set of data values and/or elements. This simple structure leads us to be able to represent both of the elements in the DREAM model, which is a data model for a bottom-up database, and the shape, which is the information on the structure of a database. The operators and operations for a set with a name rather than only a set are introduced. The representation of relationships in a bottom-up database is also proposed. The proposed method is to describe the information of relationships as objects. Users can use them in querying and manipulating relationships.

The paper proposes a data model for a scientific database. Scientific applications require the DBMS support of the discovery process. Unstructured or semi-structured scientific data is required for storage in databases, and manipulation as structured data. To meet these requirements, the proposed data model has the following features. It is based on the concept of sets in order that a database has flexible structure, a perspective of an object is represented with the subset of the object, and a collection of objects are naturally represented. A data value is the principal element for expressing polymorphism and sharing data flexibly. A preliminary database is adopted in order to store unstructured or semi-structured data without any definition. The attribute of data called shape, which varies according to the creation or updating of data, is introduced to query a database. The authors present the application of the model to the archeological data in order to demonstrate its advantages.