优酷：分享全媒体盛宴

致力于打造＂优酷出品＂的原创概念,古永锵和他的团队希望把传统屏幕前的观众带到视频网站上。     

Designing an Interactive Visualization to Explore Eye-movement Data

For many user applications large data sets may be collected passively and unobtrusively in the flow of their activity, and on scales ranging from the individual to increasingly larger communities. Large data sets, however, bring a concomitant need for tools to help understand what the data indicates. With the emergence of smart eyewear and the availability of sophisticated but affordable eye-tracking devices, eye movement data becomes a source of detailed information about a user’s focus and indirectly about their cognition and attention. Visualizing this usefully in terms meaningful for diagnosis however, remains a challenge. In this paper we report a new data representation from significant data sets generated by a gaze-controlled digital reading application for second language speakers. Current tools provide data sets aimed primarily towards statistical analysis of patterns: our focus is on end-user exploration of data sets in domain terms, so that practical implications can be readily identified. The visualization of horizontal eye movement data allows rapid diagnosis of problem areas in texts, informing educators immediately of individual or wider issues. The general applicability of this visualization to other applications is discussed.     

Graph-Based Entropy for Detecting Explanatory Signs of Changes in Market

Graph-based entropy, an index of the diversity of events in their distribution to parts of a co-occurrence graph, is proposed for detecting signs of structural changes in the data that are informative in explaining latent dynamics of consumers’ behavior. For obtaining graph-based entropy, connected sub-graphs are first obtained from the graph of co-occurrences of items in the data. Then, the distribution of items occurring in events in the data to these sub-graphs is reflected on the value of graph-based entropy. For the data on the position of sale, a change in this value is regarded as a sign of the appearance, the separation, the disappearance, or the uniting of consumers’ interests. These phenomena are regarded as the signs of dynamic changes in consumers’ behavior that may be the effects of external events and information. Experiments show that graph-based entropy outperforms baseline methods that can be used for change detection, in explaining substantial changes and their signs in consumers’ preference of items in supermarket stores.     

Simulation Analysis of a Dynamic Reallocation-based Routing Functionality for SAGE Application

A large amount of scientific data is generated through computation and analysis. Also, today’s scientific research necessitates an aggregation of diverse expertise from experts with different backgrounds. Therefore, visualizing and sharing these data among scientists in an intuitive and easy manner is of more importance in scientific collaborative research than ever. We have focused on SAGE, a middleware that realizes a large-scale remote visualization leveraging network streaming technique as a promising building block technology. However, no advanced network control function for efficient use of the network is equipped with SAGE although SAGE relies on a network streaming techniques. Our previous research integrated a functionality that detected network failures avoided such failure by rerouting network flows of SAGE into the functionality. Nonetheless, a network congestion avoidance functionality to adapt a collection of network flows generated by SAGE to the underlying network has not been developed yet. In this paper we explore the feasibility of a reallocation-based dynamic routing functionality for SAGE, with which all network flows on the network are rearranged. For feasibility research, we perform simulation experiments to observe the behavior of the functionality. The evaluation confirms that our proposed functionality works correctly and brings better throughput for each flow.     

The Boosting Effect of E-WOM on Macro-level Consumption: A Cross-Industry Empirical Analysis in Japan

In this study, we carry out an empirical analysis on how electronic word-of-mouth (hereinafter “e-WOM”) marketing on e-WOM websites and electronic-commerce websites on the Internet boosts consumption on a macro-level. In our analysis, we conduct a model analysis of consumer behavior using data composed of more than 30,000 questionnaire surveys and quantitatively find the elasticity coefficient of the boost to consumption by performing a two-step GMM (generalized method of moments), which uses instrumental variables. The results of the analysis show e-WOM significantly increased expenditures in six fields: computers, electrical appliances, etc.; music; hobbies; clothing, accessories, etc.; beauty products, etc.; and goods for everyday life, etc. Furthermore, there was no field that had a significantly negative value. These results showed that, in the majority of the target fields, e-WOM had not only the effect of winning customers from the competition, but also the effect of boosting consumption on a macro-level. In addition, even from people’s subjective evaluations, there were many in all the generational groups who said that e-WOM boosted expenditures.     

启迪创新研究院为中国城市创新创业环境“体检”——《中国城市创新创业环境排行榜》子榜发布

清华大学启迪创新研究院通过“政、产、学、研、金、介、贸、媒”等软硬环境指标对GDP过千亿、人均GDP排名前100的城市进行创新创业环境对比分析。以期为区域政府完善创新创业环境提供借鉴和参考。     

The Future of Socionetwork Strategies

In the first stage of creating technological improvements for a democratic society, the leaders of the Society of Socionetwork Strategies have conducted statistical research on the effect of information and communication technology on the global economy. In the second stage, younger professors, based on their Positive Studies, have leapt into a multi-agent simulation of Grid Computing. In the third stage, a terra-bite scale of social data-collecting and the mining of large-scale data in our real world will be next. The next generation of scientists will share knowledge and intellectual inquiry in the fields of statistics, mathematics, and the social sciences with the help of computers connecting such scholars from all over the world.     

An Interactive Independent Topic Analysis for a Mass Document Review Service

In this paper, we propose an interactive constrained independent topic analysis in text data mining. Independent topic analysis (ITA) is a method for extracting independent topics from document data using independent component analysis. In this independent topic analysis, the most independent topics between each topic are extracted. By extracting the independent topic, managing documents with a large number of text data is easy with document access support systems and document management systems. However, the topics extracted by ITA are often different from the topics a user requests. For the system to be of service to users, an interactive system that reflects the user’s requests is necessary. Thus, we propose an interactive ITA that works for the user. For example, if there are three topics, i.e., topic A, topic B, and topic C, and a user choose the content from topics A and B, a user can merge those topics into one topic D. In addition, if a user wants to analyze topic A in more detail, a user could separate topic A into topics E and topic F. To that end, we define Merge Link constraints and Separate Link constraints as user requests. The Merge Link constraint is a constraint that merges two topics into one topic. The Separate Link constraint is a constraint that separates two topics from one topic. In this paper, we propose a method for extracting a highly independent topic that meets these constraints. We conducted evaluation experiments on our proposed methods, and obtained results to show the effectiveness of our approach.     

A Statistical Analysis Between Consumer Behavior and a Social Network Service: A Case Study of Used-Car Demand Following the Great East Japan Earthquake and Tsunami of 2011

When a large-scale disaster hits a community, especially a water-related disaster, there is a scarcity of automobiles and a sudden increase in the demand for used cars in the damaged areas. This paper conducts a case study of a recent massive natural disaster, the Great East Japan Earthquake and Tsunami of 2011 to understand those car scarcities and demand in the aftermath of the catastrophe. We analyze the reasons for the increase in demand for used cars and how social media can predict people’s demand for used automobiles. In other words, this paper explores whether social media data can be used as a sensor of socio-economic recovery status in damaged areas during large-scale water-related disaster-recovery phases. For this purpose, we use social media communication as a proxy for estimating indicators of people’s activities in the real world. This study conducts both qualitative analysis and quantitative analysis. For the qualitative research, we carry out semi-structured interviews with used-car dealers in the tsunami-stricken area and unveil why people in the area demanded used cars. For the quantitative analysis, we collected Facebook page communication data and used-car market data before and after the Great East Japan Earthquake and Tsunami of 2011. By combining and analyzing these two types of data, we find that social media communication correlates with people’s activities in the real world. Furthermore, this study suggests that different types of communication on social media have different types of correlations with people’s activities. More precisely, we find that social media communication related to people’s activities for rebuilding and for emotional support is positively correlated with the demand for used cars after the Great East Japan Earthquake and Tsunami. On the other hand, communication about anxiety and information seeking correlates negatively with the demand for used cars.     

Mining History of Gamification Towards Finding Expertise in Question and Answering Communities: Experience and Practice with Stack Exchange

Recently, an online Q&A tool has become an essential part of individual communities and organizations of experts on specific topics. Using the answers to questions about specific topics will help such communities work more efficiently in their fields. Currently, Q&A online communities are adopting gamification to engage users by granting awards to successful users. In this paper, we investigate how to mine award achievement histories to find expertise. We propose the use of sequence analysis and clustering techniques. Specifically, we study the history of Stack Exchange, a large Q&A community that employs gamification. To the best of our knowledge, this is the first study of using award achievement history to find expertise in Q&A communities.     

二次创业的安方高科

继大气、水、噪声环境后,电磁环境正悄然成为一个新的关注焦点,逐步影响着大众对电磁环境安全的消费观。顺应消费者对电磁环境安全的迫切需求,无辐射家电将大行其道。如果全国的微波炉都做到无辐射,仅此一项就将催生上百亿元的市场,如此巨大的市场前景,怎能不激发起安方高科整装待发的斗志？     

Correlation and Regression Analysis for Characterizations of a University Hospital

One of the most important issues in hospital management is how to determine the indicators for revenue of a large-scale hospital. Although it has been pointed out that the averaged value of length of stay in a hospital is an indicator of its revenue, quantitative evaluation has not been conducted. In this paper, we propose an analytic process based on methods for correlation and regression analysis to evaluate this indicator by using stored data in hospital information systems. We applied this method to data combined with discharge summaries and medical treatment fees, collected from 1997 to 2000 in Chiba University Hospital. The result shows that the length of stay can explain around 60-90% of the revenue during this period.     

The Contents-Based Website Classification for the Internet Advertising Planning: An Empirical Application of the Natural Language Analysis

This study proposes a model for website classification using website content, and discusses applications for the Internet advertising (ad) strategies. Internet ad agencies have a vast amount of ad-spaces embedded in websites and have to choose which advertisements are feasible for place. Therefore, ad agencies have to know the properties and topics of each website to optimize advertising submission strategy. However, since website content is in natural languages, they have to convert these qualitative sentences into quantitative data if they want to classify websites using statistical models. To address this issue, this study applies statistical analysis to website information written in natural languages. We apply a dictionary of neologisms to decompose website sentences into words and create a data set of indicator matrices to classify the websites. From the data set, we estimate the topics of each website using latent Dirichlet allocation, which is fast and robust method for sparse matrices. Finally, we discuss how to apply the results obtained to optimize ad strategies.     

Personal Authentication Using a Kinect Sensor

This article proposes a new approach to personal authentication by exploring the features of a person’s face and voice. Microsoft’s Kinect sensor is used for facial and voice recognition. Parts of the face including the eyes, nose, and mouth, etc., are analyzed as position vectors. For voice recognition, a Kinect microphone array is adopted to record personal voices. Mel-frequency cepstrum coefficients, logarithmic power, and related values involved in the analysis of personal voice are also estimated from the voices. Neural networks,support vector machines and principal components analysis are employed and compared for personal authentication. To achieve accurate results, 20 examinees were selected for face and voice data used for training the authentication models. The experimental results show that the best accuracy is achieved when the model is trained by a support vector machine using both facial and voice features.     

A Pricing Strategy with Consumption Externality when the Reference Group is Large

This paper provides an analysis of firms’ optimal prices when individual utility exhibits both material and other consumer effects. We construct a model and conduct an analysis in the following four steps: (1) By using a conjoint analysis of data, we set up a hierarchical Bayesian model and estimate its parameters, which are composed of individual utility functions. (2) We calculate the number of sales in each time period and derive the total sales throughout the given period. (3) We estimate the demand functions in the presence and absence of consumption externalities. (4) Finally, we calculate the profit functions and derive the optimal prices taking into account various product attributes. By applying our analysis to the book market of Japan, we find that the optimal price is lower when an externality is present than when absent. The intuition behind this result is that pricing low and selling a large number of books from an early stage increases the externality effect, yielding a higher profit for firms.     

The Role of Cross-Border Networks of Skilled Labor in Offshore Outsourcing: Empirical Evidence Based on the Network Theory

This study investigates the determinant role of the cross-border movement of skilled labor in the expansion of service trade between the US, and both developed and developing countries. For this purpose, we employ the key concepts of network theory as an analytical framework and conduct panel data analysis and graphical modeling analysis for 31 countries from 1999 to 2008. In this decade, offshore outsourcing in the service trade took off worldwide. We use data for each country’s service exports to the US, number of H-1B visas issued, GNI per-capita, network readiness index, and an English dummy for the official language. We illustrate the trajectory and interactions between these factors. These analyses yield three observations. First, service trade with the US is more intensive among higher income countries. Second, the number of H-1B visas issued has a positive effect on service exports to the US. Third, individuals in lower income countries tend to desire H-1B visas and create intensive skilled labor networks with the US, the path through which developing countries such as India expanded their service exports to the US.     

Data Mining Approach for Direct Marketing of Banking Products with Profit/Cost Analysis

Nowadays, many businesses, such as banks, use direct marketing methods to reach customers to minimize the campaigning cost and maximize the return rate. To achieve this, huge customer data should be analyzed to determine the most appropriate product offer for each customer and the most effective channel to reach her/him. However, since only a very small amount of responses collected from the customers are positive to the offers, the dataset is very imbalanced. This decreases sensitivity ratio of prediction results and makes it difficult to make a successful product and channel selection for the offer. In this paper, we propose a hybrid system, which first classifies the customers to decide if s/he is interested in the offered product, and then clusters them for product and channel suggestions. Experiments with real life banking data show very promising accuracy results for predicting the proper product and channel for the customers. Moreover, cost-profit analysis is also added to this problem. Our experiment results show that the proposed method decreases a fraction of the total profit, but since the decrease in the total cost is very large, there is a huge increase in the overall profit/cost ratio.     

A New Type of Computer Premium and Its Correlation to Individual Wages

This paper examines statistical correlations between workers?? income levels or wages, and Information and Communication Technology (ICT). The analyzed data set is micro data collected through the Web in 2008 and 2010 by the Research Institute for Socionetwork Strategies. This micro data estimates the rank correlation of income classes with respect to three types of ICT labor service. Next, multiple linear regression functions are estimated for the wage levels with respect to age, sex, years of continuous employment, and three types of ICT labor service over all industries and in the manufacturing industry or service industry. Finally, a multiple regression analysis is conducted for individual utility levels with respect to income, leisure, and three types of ICT labor service. The results show that, first, the coefficients of rank correlation for all pooled data are positive and statistically significant. Second, individual ICT labor services have positive effects on workers?? wage levels, and the partial regression coefficients are statistically significant. Third, the partial regression coefficient of personal computer operation time decreased but the partial regression coefficient of mobile computer operation time related to work increased drastically from 2008 to 2010. Fourth, the use of the personal and mobile computer shifts the utility function upward with respect to income and leisure.     

大唐电信：赢在平台

在金融与安全产业、移动互联网产业方面,大唐电信将在现有基础上,取得行业领先地位,并在细分领域成为主流供应商;在物联网和三网融合产业方面,大唐电信的业务将遍地开花,用＂大终端＋大服务＂的整体解决方案全面迎接智慧城市、智能商业中心等各种大规模行业应用时代的到来。