Thanks to the popularization of information and communication technology, the nurses work using mobile devices to communicate with co-workers and record nursing care at hospital. In this paper, aiming to facilitate nursing care, we propose a method to recognize nursing work activities by using topic models from acceleration data stored in mobile devices and knowledge of the work. In contrast to simple tasks such as walking or running, working activities are more difficult to recognize because of their complexity and length. To address this difficulty, we define the system composed of two layers, simple task recognition layer and working activity recognition layer, based on the assumption that work activities consist of a probabilistic combination of various simple tasks. In the simple task recognition layer, the system first recognizes simple task by applying supervised learning techniques to time-domain features extracted from sensor data. Then it recognizes working activities by applying topic models to simple tasks and annotation with knowledge of nursing work. We conducted an experiment at a hospital and collected nursing activity data for 96 hours by 12 nurses as a result. Using those data, we show that our method surpasses the conventional methods in recognizing nursing activities.
This paper proposes a novel method of multimodal categorization for learning word-meaning. It is assumed that the mutlimodal observations do not necessarily capture matched object unlike previous work, but parts of them are still often matched with each other. Subjective consistency is introduced, which measures to what extent probability of object category in one modality is judged to be close to those in different modalities and expected to be utilized for finding the matched parts in multimodal observations. We apply this idea to extend the previous method called Multimodal Latent Diriclet Allocation for coping with the above assumption. Experimental results both with real and artificial data show the efficiency of the proposed method for multimodal categorization using multimodal data involving unmatched observations which are considered to be normal in more realistic situation of learning word-meaning through interaction with humans.
The Ministry of Economy, Trade and Industry and the corporate world in Japan have recently embraced the ‘Cool Japan’ policy. They have assisted Japanese content industries in exporting Japanese media contents, such as manga (comics), to foreign countries, especially in Asia. However, this overseas expansion has not been successful in producing profits to the expected degree. The main reason for this shortfall is that companies are unable to perceive local consumer consumption trends in those countries easily and at low cost. Consequently, they are unable to select media contents that might be popular in such areas in the future. Herein, we design a consumption trend calculating system that incorporates web mining. It is readily applicable to many countries and contents. Specifically, we use web mining of data elicited from search counts on search engines, tweet counts from Twitter, and article data from Wikipedia to calculate consumption trends based on weekly manga sales data in Japan. Results show that this model can predict consumption trends for six months with high accuracy and that it can be adapted to calculate consumption trends of other contents effectively, such as anime (animation) in Japan and manga in France. Moreover, we establish the ASIA TREND MAP web service to inform industries about these calculated consumption trends for Asian countries.
Webpage optimization is an experimental method to make continuous improvements on websites based on users' behavior. This method can be implemented easily but has a drawback that small websites take long time to gather enough data to evaluate the ideas. Although many optimization methods are proposed and conducted so far, there's no mathematical model of this problem. We propose Webpage Optimization Problem and organize existing webpage optimization methods. Combining these methods, we also propose a new webpage optimization method that performs well no matter how many people get to the website. We evaluate the proposal method by simulation experiments and introducing the optimization program to both large and small websites. The results show that our proposal method outperforms existing methods at any size of websites. Webpage Optimization Problem is a framework to create new webpage optimization methods.
In social networking service (SNS), popularity of an entity (e.g., person, company and place) roles an important criterion for people and organizations, and several studies pose to predict the popularity. Although recent papers which addressing the problem of predicting popularity use the attributes of entity itself, typically, the popularity of entities depends on the attributes of other semantically related entities. Hence, we take an approach exploiting the background semantic structure of the entities. Usually, many factors affect a person's popularity: the occupation, the parents, the birthplace, etc. All affect popularity. Predicting the popularity with the semantic structure is almost equivalent to solving the question: What type of relation most affects user preferences for an entity on a social medium? Our proposed method for popularity prediction is presented herein for predicting popularity, on a social medium of a given entity as a function of information of semantically related entities using DBpedia as a data source. DBpedia is a large semantic network produced by the semantic web community. The method has two techniques: (1) integrating accounts on SNS and DBpedia and (2) feature generation based on relations among entities. This is the first paper to propose an analysis method for SNS using semantic network.
The purpose of this paper is to test the Spiral of Silence theory in Internet society. Even today Noelle-Neumann's Spiral of Silence Theory is an important topic on the formation of public opinion. In the Spiral of Silence Theory up to now, the willingness to speak out has been handled as a dependent variable. However, there is significant bias in the question as to what extent the willingness to speak out actually influences the number of times a person speaks out. In addition, snowball sampling has been used, even in regard to the distribution of opinions of persons close to an individual. Accuracy increases because the attitudes of direct close users can be studied; however, only a small portion of close users can be studied. One defect of this approach is that it is actually quite costly. We use as a dependent variable the actual number of `tweets' on Twitter rather than willingness to speak out. In addition, for the attitude of close users, we used machine learning to estimate the attitudes of persons the users came in contact with, and we quantified homogeneity. We used and combined social investigations and behavior log analysis. With these, we were able to adopt simultaneously the following to a model: 1) individuals' internal situations, which can only be clarified by a questionnaire. 2) the actual quantity of behavior and the structure of communication networks, which can only be clarified through analysis of behavior logs. In the result, we found that a person's perception that their opinion in the majority and estimated homogeneity had a positive effect on the number of times a person spoke out. Our results suggest that the spiral of silence in regard to actual speaking out on Twitter.