A two-stage graph mining scheme for novel item recommendation is proposed. The first stage extracts preference attribute as a topic-rating vector using a graph mining technique given a data set of item ratings, and the second stage predicts one's preferred items out of novel domains using another graph mining procedure based on the extracted preference. We compared the prediction performance of four possible implementations using a data set of ratings of TV programs. The data set contains 368,400 ratings, ranging from -2 to +2 (negatives to positives). Each rating was provided by 1,228 subjects against 300 TV programs from the NHK On-Demand service. To describe semantic relationships between the items, we appended 1,000 metadata-nodes to each graph, which are extracted from the outline field of the Electronic Program Guide contents.
This report proposes an interactive public information system called "PushPullWall" which allows users to post and get various kinds of information using their own smartphones in a manner like throwing information at the wall or catching information from the wall. The preliminary experimental results using a prototype system built for the proof of concept are shown. They encourage us to envisage the promising future of intuitive interactions between public information systems and smartphone-equipped users.
There are quantitative analysis methods and qualitative analysis methods in the technique of evaluating small and medium-sized manufacturers. Although the accuracy of quantitative analysis is improved by development of financial engineering, etc. In recent years, qualitative analysis is difficult to evaluate objectively, and the establishment of a technique and improvements in accuracy have been a subject of debate. This paper has focused on the technology assets of small and medium-sized manufacturers, and also considers the synthetic quantitative analysis method.
This report proposes a multi-person interactive system called "Mobement Space" (Mob + Movement Space) which we hope to induce people to act harmoniously by presenting aware triggering graphics using the information observed from the people's behaviors in group. Experimental results using a prototype system, which comprises big screens and web cameras, show that each participant voluntarily acted by finding some rules how the graphics change while being aware of other persons' behavior, and harmonized interactions were eventually presented as a whole. We hope that this kind of found-and-aware-based multi-person harmonized interactions will be widely seen in public spaces in the future.
We describe our ongoing project to establish collaborative safety mechanism based on mutual sharing of driver's information. Most conventional driving safety approaches aims at assisting individual drivers by providing various safety information to the driver on the vehicle. However, the effect of such system highly depends on drivers' response. In our approach, we try to make the driver's present performance to be shared with other drivers and pedestrians. Using this sharing mechanism, people in traffic will be able to make better prediction of other driver's behavior and it leads to their proactive actions to prevent accidents.
Shape From Focus/Defocus (SFF/SFD) has been known as a depth estimation technique by using focus information. The technique estimates the depth for each point of a target object based on a focus measure extracted for each block on a set of multi-focus images. The authors previously proposed an SFF/SFD technique that uses the data from multiple viewpoints. However, the technique has not taken into account the shape of the square block from another viewpoint. In this paper, we propose a new version of the technique in which such block shape is adaptively modified depending on each viewpoint. The algorithm for finding the optimal depth-estimate is also simplified in comparison to that of the old version.
The authors have proposed a video coding technique that maximizes an estimated spatio-temporal Mean Opinion Score (EMOS), and have applied the technique to a spatio-temporal scalable video coding. However, the advantages of the scalable video coding technique have not been sufficiently investigated, especially from the viewpoint of an overall video quality in terms of reducing the frame rate by the temporal scalability. This paper thus examines the advantage of the scalable video coding technique by applying it to several test sequences.