The styles of Japanese conversation change in a complex manner depending on various factors such as the attributes of the speaker and the relationship between the speaker and the listener. In this study, we focused on the variety of expressions of Japanese first personal pronouns and second personal pronouns, and investigated and classified the appeared expressions within conversation sentences based on the Balanced Corpus of Contemporary Written Japanese. In addition, the co-occurrence patterns of first personal pronouns and second personal pronouns were extracted by factor analysis.

　In character research, research on appearance and research focusing only on male characters have been conducted. However, research focusing on ttractiveness other than appearance and research focusing only on female characters have not been conducted much. In this research, we focus on the difference between the incentive attractiveness of a character and the continuous attractiveness of characters, and analyze the attractiveness of a female character. First, we investigated how attractive female characters are by questionnaire. Chi-square test and factor analysis were performed using the results of this questionnaire. Chi-square test revealed there was significant difference between incentive attractiveness and continuous attractiveness. In addition, factor analysis was performed to find out which combination of factors was attractive. Incentive attractiveness was an element that impressed the character, and continuous attractiveness was an element that could be added later to the character settings. As a future task, it is necessary to investigate whether similar results will be obtained with other mobile games.

　Manga are often provided on Web with tags or the keywords expressing subjects for the search based on contents. However, they are less structured and not useful for semantic search. This paper proposes a method for automatic linking manga and NDC-LD entities based on the subject by using two kinds of information, their subject tags and outline on the Web explaining manga and its content. The method using subject tags is to link by the similarity of tag words and correlative catchword of NDC-LD. On the other hand, The method using the outline is to link entities extracted as characteristic words of identified heading words of NDC in the outline. This paper also applied the proposed method to for 146 manga works, and compares the result with the classification by hands and the existing automatic estimation method of NDC by the machine learning. This comparison shows that the proposed method is more efficient for the reproduction of the manual classification. (160words)

　In this study, a questionnaire survey of 1,717 researchers was conducted to understand the impact of CORVID-19 infections on research activities in Japan from around January 2020 to September 2020. The results show that the degree of impact of CORVID-19 infections on research activities, the progress of research activities, and the use of digital tools varied greatly by the respondents’ regions. In addition, it became clear that the current concerns about research activities tended to differ by the respondents’ research fields.

　The spread of COVID-19 affects not only social changes but also personal lives and behaviors. Therefore, we focused on the content of Twitter with the aim of clarifying people's thoughts and social reactions. Twitter often posts various information and opinions in real time.By analyzing the contents of Twitter, it will be possible to utilize it for future correspondence and efforts for COVID-19.

　In this study, we analyzed the contents of Twitter of COVID-19 from text mining.

　At the end of 2016, Basic Act on the Advancement of Public and Private Sector Data Utilization came into effect. Even before the law came into effect, the electronic administration open data strategy was formulated in 2012, and the disclosure of open data has been promoted by the national and local governments. Will the enactment of data utilization legislation affect the promotion of open data measures in local governments? This study confirms the influence of the plan formulation by confirming the formulation status of the “Plan for the Advancement of Public and Private Sector Data Utilization” for local governments that have already started open data.

The progress of digitization and openness of regional and municipal materials, such as history books, states and public relations magazine published by municipalities in Kanagawa prefecture has been estimated based on survey of metadata and digital materials. The survey was carried out by using search services such as the National Diet Library Digital Collections (NDL Digital), the NDL Web Archiving Project (NDL WARP), the cross search of public libraries in Kanagawa prefecture, municipal homepages and OPAC of public libraries. A proposal on the progress of digitization and openness of local history materials has been made.

AI learning systems are now able to accomplish useful tasks such as image classification and spoken language comprehension. You can now perform AI calculations on your web system. It is important to teach AI systems how to recognize images. I asked the third grade students to do the test and training. First, provide a training example so that you can learn what a particular image looks like. The students tried the examples, became interested, and built their own AI system.

　Students with low vision make effective use of ICT by setting accessibility. In particular, tablet terminals such as iPads are regarded as one of the tools that dramatically enhance the learning environment for students with low vision. Therefore, in this study, based on the daily ICT utilization situation of students with low vision, we practiced lessons by digitizing notebooks using iPad. As a result, digitization of notebooks using the iPad was effective in promoting content understanding and home learning.

　“Learning programming by typing” is a way of learning programming where a learner types the sample source code verbatim. The author has adopted this style in his courses of the C programming language so that the university students could acquire the knowledge of the syntax continually. In this paper, we attempted to verify whether learning contents are utilized in later classes. In the evaluation, 59 source files written in C were segmentalized into tokens such as “printf” and “++”, and set operations emerged the notable tokens.

Today, the number of accesses to specialized Databases in the humanities is declining. One of the biggest reasons for this is the decrease in the number of major users due to the decline in the so-called educational population. In addition, databases designed for use by specialists are not suitable for everyone due to their high degree of specialization. Therefore, as the number of specialists decreases, the recognition of such databases will naturally decrease, and the frequency of their use will also decrease. What is needed to increase the frequency of use of specialized databases in the humanities and to reaffirm their importance? If it is difficult to increase the number of users on its own, is it possible to increase the frequency of use and raise awareness of its importance through mutual cooperation? We will try to verify what kind of database is highly public and can be used by everyone, using the "Database of Japanese Literature Theses"

　In recent years, SNS has become widespread mainly among young people, and along with that, companies are also taking the initiative in using it. However, there are cases in which a fierce debate called flaming occurs on SNS. In this research, we aim to develop a system to support the early resolution of flames. This paper report that we tried to classify automatically between pros and cons of opinions using multiple machine learning methods for a tweet thread that burned on Twitter, and obtained results of 86.6-88.1% accuracy.

　In this research attack, I liked the Text-to-Speech service, and since it was in PowerPoint, I returned Rona and started developing the code now. It's like splitting the space using Google TTS, there is no point in trying to get it with just this. Discover the point It is not possible to set the automatic play setting to play automatically by tapping, tapping, and tapping. Or, he said that he researched what was achieved and attached a box with a degree of appearance.

　Today's Japan is said to be an aging society, and there is a possibility that long-term care will have to be taken care of parents rather than by others. Therefore, it may be necessary for everyone to engage in long-term care without vocational training or preparation. Therefore, in this research, we will develop a danger detection system in order to increase the number of eyes watched by AI and prevent accidents. With the aim of reducing the burden on caregivers, we used deep learning to focus on falls, which is the most dangerous case in long-term care, clarified what kind of danger there is, and designed the system. In this paper, we report the results of investigating the dangers in long-term care and the design contents of the danger detection system.

　In order to automatically generate high-quality stories, it is necessary to clarify the narrative structure and stylistic characteristics of each story type. In this study, we attempted to clarify the stylistic characteristics of the romance genre through factor analysis of full-length romance novels. As a result, it was confirmed that the characteristics of the descriptions of the two emotions of ”like”and ”sad”differed depending on the gender of the characters, such as the descriptions in the beginning of the story for female characters and the descriptions in the end of the story for male characters, which characterized the romance genre.

　The data in which the speaker and the conversational sentence are linked in the story text can be applied in various ways, but it has been difficult to identify the speaker in the past. Therefore, in this study, we focused on the writing style as a useful feature for speaker estimation of conversational sentences,and manually classified the conversational sentences of Japanese story texts based on the writing style. We also used an algorithm to estimate the writing style using functional words. developed. In the method of the previous research, the correct answer rate of the classification was about 50

In recent years, content in various languages has been posted, and many analyzes have been tried on texts in multiple languages. Here, there are not many sentiment dictionaries that can be used for cross-linguistic analysis in sentiment analysis. It would be useful if dictionaries intended for analysis in a single language could also be used. In this study, we examined whether the emotional polarity values contained in dictionaries were valid for other languages using novel texts translated into multiple languages. As a result, we were able to confirm a certain degree of effectiveness.

　In Japan, it is ongoing that promote work style reform now for successful women by Cabinet Office. So, this study proposed the quantitative analysis approach for reveal some differences and common points each region. We verified the effects of this approach in evaluation experiment using e-stats data (Labor Statistics Data, Social Security Statistics Data).

　In this study, the definition of otaku is “people who are deeply absorbed in a particular hobby,”with a particular focus on people who are absorbed in anime and manga. One of the aspects of otaku culture is the act of “Fukyo”. However, “Fukyo”itself has not yet been clarified, and basic research is needed. The purpose of this study is to clarify the attitudes and awareness of otaku in the process of “Fukyo”as a stepping stone to clarifying the reality of “Fukyo”. PAC analysis was used as the method, and a survey was conducted on six participants from September 2020 to March 2021. In this paper, we report the results of the analysis of three of them. The results of the analysis show that they were aware of the existence of their Fukyo partner in their “Fukyo”. The direction of their awareness is different among the three, although there are some common aspects such as “preference of the partner”.

　In recent years, most of the analyses of story's have been conducted on primary works, and only a few have been conducted on Fan Fiction. In this study, as a first step in the study of fan fiction, we exploratively surveyed the attributes such as genre and characters that are considered to represent the characteristics of each work, and created data of the works based on multiple attributes. As a result of quantitative analysis of the obtained data, three typical types of story patterns of Fan Fiction were revealed.

　Mythology is considered to be the original form of storytelling, and mythological stories are of great value in automatic story generation. However, unlike myths, which have a simple and basic narrative structure, modern narratives are considered to have a complex structure. In this paper, we compare the structure of mythological stories with that of modern stories based on myths, and discuss how to generate stories from the original. As a result of the analysis, it is clear that there is a different structure between the behavior of the protagonist of the mythical story and that of the protagonist of the modern story.

　Recently, it was clarified that structure extraction and automatic generation of the story of the role playing game were possible． However, only the continuous structure in the story was adopted in the analysis． The story structure such as nesting structure and parallel structure should be adopted for the automatic generation of the more complicated and free story development． The story contents also need to be included in the analysis． In this paper, I propose a method to describe and analyze basic narrative structure, such as continuous structure, nested structure, and parallel structure between quests in role-playing games． Using the proposed method, we describe and extract the basic narrative structure in a total of 331 quests, and the narrative contents are also described and extracted in relation to them．The effectiveness of the proposed method is verified by comparing the extracted results with actual stories． As a future problem, it is necessary to verify the objectivity of the proposed method．

　Browsing history is the data that records the information gathering behavior through the Internet in many situations in our lives, and is a part of the lifelog that records and stores personal experiences. This study aims to utilize the browsing history as a record of daily life, focusing on self-reflection, which is the process of analyzing past actions to gain self-understanding and behavior improvement. The proposed system supports the recall of contents of browsed web pages by analyzing the topics in the browsing history, and the recall of the background of browsing by visualizing the browsing history based on the topics and the browsing time series. In order to investigate the effectiveness of the proposed system, we conducted an evaluation using actual browsing histories as input. The results showed that the system provided sufficient support for content recall when the feature words of the web pages were successfully extracted. As for the background recall, the visualization of the browsing history using the combination of the topics and the time series helped the subject recall the thoughts and events of the time. Also the proposed system provided useful reflections on the information gathering behavior and life at that time.

　In information seeking process, a user often submits multiple queries and repeats the search to achieve the purpose. Therefore, in order to discover the user's search intention more accurately, predicting the query level satisfaction from user’s search behavior is one of the important issues currently. In this research, we arrange the queries which submitted in a session by time sequence, and propose a method for predicting user’s query level satisfaction. The proposed method uses click data, interval time and query reformulation as indicators, which are often used to predict query satisfaction. In addition, we examine the effect of similarity between reformulated query and the search results of the previous query on query satisfaction. We predict query satisfaction by combining these four features. From the evaluation experiment, it shows that comparing with query reformulation only, the estimation effect has been improved 6 points approximately by using the similarity between the query and the document, but when using all of four features, the accuracy is 71.27% which was not the best result. In this paper, we discuss results with a failure analysis, and further improvement measures.

　This study focused on the importance of a diary. In order to realize automatic generation of diary sentences, we analyzed the sentence structure. Assuming the influence of the author's attributes, we collected diary sentences from blogs in four categories: old men, old women, young men, and young women. The functions of each sentence in the collected blog articles were manually classified and symbolized. The time series structure of sentences was analyzed using n-gram. As a result, it was found that the basic sentence structure of the diary sentence is not affected by the attributes of the authors.

　There have been few attempts to quantitatively analyze the rhetoric and stylistic characteristics of Japanese essays other than the metaphors used. In this study, we collected data on the rhetoric used in two essays written by different authors in order to clarify the rhetoric used in essays on average. Based on the data obtained, the characteristics of the rhetoric used in the essays were quantitatively extracted using the χ-square test and factor analysis.