While much research in recommender systems focused on improving the accuracy of recommendations, issues pertaining to their presentation have been under-explored. Considering the uptake of recommendations as one of their success indicators, we investigate the role of proponents in affecting user's decision to accept a recommendation. We refer to proponent as a person or avatar, advocating in favor of the recommended item. This paper reports on a user study that evaluated the impact of including several types of proponents in the recommender interface and their impact on the uptake of recommendations. We observe that out of the studied proponents, real-world contacts have the strongest impact on the uptake of recommendations, which can inform the design recommender system interfaces.

The goal of this research is developing a system that a rich facial animation can be used in communication is generated from only speech. Generally, a source of the generating facial animation is a camera. Using cameras as an input source, it causes limitations of the angle of view of the camera or problems that cannot be aware of the human face, depending on the orientation of the face. Therefore, it is reasonable for developing a system for generating a facial animation using only voice. In this study, we generate facial expressions from only speech using LSTM-RNN. Comparing 3 patterns of speech analysis data, we showed that the proposed method using A-weighting is effective for facial expression estimation.

Typical Head-Mounted Displays (HMDs) that provide a highly immersive Virtual Reality (VR) experience make any interaction between a user and real space difficult by occluding the user's entire field of view. Video see-through type HMDs can solve this problem by superimposing real-space information on the VR environment. The existing method of supporting interactions with the real space is superimposition of boundary lines of the real space on the virtual space in the HMD. However, overlaying the boundary lines on the entire field of view may reduce the user's immersive feeling. In this paper, we propose two methods to support interactions with the real world while playing immersive VR games without reducing the user's immersive feeling as much as possible, even when the user wanders. The first method is to superimpose a 3D point cloud of real space around the user on the virtual space in the HMD. The second method is to deploy familiar objects (e.g., furniture in his/her room) in the virtual space in the HMD. The user traces the familiar objects as subgoals to reach the goal. We implement the two methods and conduct a user study to compare interaction performance. As a result of the user study, we find that the second method provides better spatial information about the real space without reducing the user's immersive feeling, compared to the existing method.

User system trust is critical to the uptake of recommendations, and several factors of trust have been identified and compared. In this paper we present a cross-cultural, crowdsourced study examining user perceptions of nine factors of trust and link the observed differences to trust development processes and cultural dimensions. While some factors consistently instil trust, others are preferred only in certain countries. Our findings and the discovered links are important for design of trusted recommender systems.

When people anonymously use social network sites (SNSs) like Twitter, they may interact with not only real world friends but also strangers, who are not acquaintances in the real world. Therefore, both of real friendship (RF) and virtual friendship (VF) coexist in these SNSs. In this research, we investigated the differences in similarity of user pairs in Japan by their types of relationship, i.e. RF or VF. The primary results indicated that RF user pairs have more common follow users on SNSs than VF user pairs, and that contents posted by VF user pairs include more similar words than RF user pairs. It is implied that two users with RF have a similar interest in neighborhood users, and two users with VF are interested in similar topics. After that, we built the models to classify user pairs into RF or VF using the similarity measures. These models showed high performance to distinguish between RF and VF (their F-measures are larger than 0.80).

In this paper, we present results of investigation on the dynamics of group decision making - how people discuss and make a decision-with collaborative web search. Prior works proposed systems that support group decision making with web search but have not examined the influence of discussion behaviors especially on the satisfaction levels with the final conclusion. In this study, we conducted a set of experiments to observe discussion behaviors and the consequent satisfaction with the conclusion using our experimental system and a set of questionnaires. The task for each participant was to make a decision on a restaurant. Our primary results revealed (1) the similar activities across all groups at the beginning and the end of the group discussion, (2) a lack of correspondence between the satisfaction with the conclusion and the time spent to reach the conclusion, and (3) the presumption that a member who actively engaged in the activities that were visible for the other members was likely to be voted as a leader in the group discussion beyond the discussion. Finally, we discussed how to implement intelligent systems that aid group decision making.

The plot (content or storyline of a story) may disappoint users who read review comments associated with items containing a story such as comics, novels, and movies. This paper proposes a new method for identifying sentences that include descriptions of the plot of the story. Conventional methods only use information based on the words contained in a target sentence, however, our new method uses contextual information in addition to word information. We identify contextual information by using the sentence location and the plot probability of surrounding sentences. An experiment showed that this method improved the accuracy with which plot-related information can be identified.

A picture lifelog is a type of lifelog that consists of pictures, mainly taken by the user. Recently, users have been able to easily create picture lifelogs because many portable devices such as smart phones have a camera. When a user sees a picture in their picture lifelog, it is sometimes difficult to recall the events related to the picture. Therefore, we proposed to combine search queries on a picture lifelog in order to support memory retrieval. Search queries are input into a web search engine to satisfy a user's need for information. Recently, because of the prevalence of smart phones, the opportunity to input search queries has increased to anytime and anywhere. Search queries are stored in a cloud user database such as Google search history. In addition, those search queries imply what the user was thinking at the time. We investigated whether search queries enable a user to recall their thoughts regarding picture lifelogs. Thus, we conducted an experiment to ascertain whether search queries reminded a user of past events. As a result, we reveal that displaying a picture with search queries performed around the time it was taken tends to improve users' memories better than its time, location, or emails sent during that time.

In many shopping sites such as Amazon.com it is possible to view and write reviews of items (products and content). Reviews of items including stories, such as novels, movies, and comics, include reviewers' opinions. Often, these reviews also include descriptions of the story. In some cases, these descriptions may spoil later reader's or viewer's enjoyment and excitement. Hereinafter, we call these descriptions spoilers. Spoilers may be related to the position in the story line. In this study we use story documents. Story documents are documents that record all of the details of the given story. Using the story documents, we investigate the location to which the content of the spoilers correspond in the story documents. Based on the result of the investigation, we consider how to detect spoilers in reviewers' comments.

Many researchers have studied complex networks such as the World Wide Web, social networks, and the protein interaction network. They have found scale-free characteristics, the small-world effect, the property of high-clustering coefficient, and so on. One hot topic in this area is community detection. For example, the community shows a set of web pages about a certain topic in the WWW. The community structure is unquestionably a key characteristic of complex networks. In this paper, we propose a new method for finding communities in complex networks. Our proposed method considers the overlaps between communities using the concept of the intersection graph. Additionally, we address the problem of edge in homogeneity by weighting edges using the degree of overlaps and the similarity of content information between sets. Finally, we conduct clustering based on modularity. And then, we evaluate our method on a real SNS network.

Kleinberg's Hypertext-Induced Topic Selection (HITS) algorithm is a popular and effective algorithm to rank web pages. One of its problems is the topic drift problem. Previous researches have tried to solve this problem using anchor-related text. In

Many recommender systems employed in commercial web sites use collaborative filtering. The main goal of traditional collaborative filtering techniques is improvement of the accuracy of recommendation. Nevertheless, such techniques present the problem that they include many items that the user already knows. These recommendations appear to be good when we consider accuracy alone. On the other hand, when we consider users' satisfaction, they are not necessarily good because of the lack of discovery. In our work, we infer items that a user does not know by calculating the similarity of users or items based on information about what items users already know. We seek to recommend items that the user would probably like and does not know by combining the above method and the most popular method of collaborative filtering.

Users cannot search information by mathematical formulas as queries in existing search engines. This is because mathematical formulas are not expressed as a sequence of characters. Some formulas are expressed in a complex structure like fractional numbers and index numbers. We present a search engine for MathML objects using the structure of mathematical formulas. The system makes the inverted indices by using the DOM structure of the MathML object. We proposed two types of index. One type is constructed from some paths of the DOM structure and expressed in XPath. The other type is constructed by encoding the nodes in the same level in DOM structure. This paper describes the experiment conducted to study the effectiveness of those indices by using the mathematical contents which are open to the public on the Internet.

The technologies for telecommunication, GPS, and mobile GIS have rapidly progressed in the last few years. They will allow us to access to the Internet at any time and any place. A user needs to acquire information easily and correctly through a web browser on a cell-phone or a PDA. Our purpose is to build a system providing blog pages written about a facility for a user who is standing in front of the facility. The user can see these blog pages through the web browser on his cell-phone. The system has two technical features. One is extracting users' goals of using facilities from a common sense database. The other is filtering blog pages based on the goal selected by the user.

A large number of semistructured documents exist on the Web. We can find pages that contain keywords by using a search engine. But when we want to obtain information about an object like a notebook computer with 1 GB memory, a method is needed that automatically extracts attribute name (in this example, ldquomemoryrdquo) and attribute value (in this example, ldquo1 GBrdquo). In the past, many researchers examined extracting attribute values corresponding to each attribute name. This paper discribes a method that extracts schemas (sets of attribute names) using bootstrapping algorithm.

One of the subjects in knowledge management is knowledge externalization and combination in which people express their tacit knowledge and formulate it. Communication among people is important for this process. This paper proposes a model that helps two users to have a discussion for knowledge externalization and combination. In this model, the computer verifies the consistency between the knowledge generated by inductive learning from cases and the two user's knowledge sets expressed by themselves. From the result of this verification, it provides the point at issue for the discussion to improve the knowledge. We conducted a user experiment in the domain of gure-fishing where real experts had participated in. We confirmed the effectiveness in activating people's communication and also in improving the quality of their knowledge.

Buyers in online auctions write feedback comments to the sellers from whom the buyers have bought the items. Other bidders read them to determine which item to bid for. In this research, we aim at helping bidders by summarizing the feedback comments. Firstly, we examine feedback comments in online auctions. From the results of the examination, we propose a method called social summarization method, which uses social relationships in online auctions for summarizing feedback comments. We implement a system based on our method and evaluate its effectiveness. Finally, we propose an interactive presentation method of the summaries based on the result of the evaluation.

In the area of information retrieval and information filtering, relevance feedback is a popular technique which searches similar documents based on the documents browsed by the user. If the user wants to conduct relevance feedback on demand, which means the user wants to see similar documents while reading a document, the existing user profiling techniques cannot acquire keywords in high precision that the user is interested in at such a short time. This paper proposes a method for extracting text parts which the user might be interested in from the whole text of the Web page based on the user's mouse operation in the Web browser. The objective of this research is to (1) find what kind of mouse operation represent users' interests, (2) see the effectiveness of the found mouse operation in selecting keywords, and (3) compare our method with tf-idf, which is the most fundamental method used in many user profiling systems. From the user experiment, the precision to select keywords of our method is about 1.4 times compared with that of tf-idf.

Net auctions have been widely utilized with the recent development of the Internet. However it has a problem that there are too many items for bidders to select the most suitable one. We aim at supporting bidders on net auctions by automatically extracting the information of the item's features from Web pages in net auctions and generating a table containing the features of some items for comparison. But because descriptions are not uniform in net auctions, there are two problems in extracting the features. The first problem is that there are some formats. The second problem is that the keywords of features are sometimes omitted. We proposed the solutions to the problems. The solution to the first problem is to distinguish the format type from tables, items and sentences, and extract the feature values in the most suitable way. The solution to the second problem is to learn the keywords in extracting from the descriptions with the keywords. And after that, the keywords are used in extracting from the descriptions without keywords. And we constructed the system which collects the information of items, extracts their features from their text information by text mining methods and generates the table containing extracted features.

In case of emergency, we need to grasp the situation and make correct assessment quickly. The video data taken from the monitoring camera is important information in the emergent situation. In our research, we attempt to build a support system for identifying the video data of the monitoring camera by dealing with video data as spatio-temporal data. We proposed a method, which displays those video data on one virtual large wall. For this presentation, the search key in our system is the location and the size of this large screen and from which direction to look. First we developed a general algorithm to index spatio-temporal data to get an idea for developing the algorithm for this search key.

Hypermedia systems have become increasingly popular as tools for user-driven learning and information retrieval. This paper describes a dynamic linkage method that allows adaptive modification of links in hypermedia dependent on users, with the aim of providing satisfactory navigation for each user. A design support tool for constructing network data in accordance with the method is also presented, and a prototype system and an associated design support tool are evaluated.