This paper describes an application of SVMs (Support Vector Machines) to interactive document retrieval using active learning. We show that SVM-based retrieval have an association with conventional relevance feedback (Rocchio-based method) by comparative analysis of relevance evaluation. We propose a cosine kernel which has the meaning equal with cosine similarity suitable for SVM-based interactive document retrieval from the analysis. We confirm the effectiveness of this method and experimentally compared it with conventional system in the Boolean, TF and TFIDF as representations of document vectors. The experimental results using TREC data set shows that cosine kernel is effective against all document representations, especially TF representation.

View full abstract

In this paper, we conducted an experiment to examine the role of verb acquisition ability on human attitudes toward a robot. As the results of the experiment involved 48 participants, we found that the robot that acquires two verb concepts, “put-on” and “away-from,” was perceived by participants as being more familiar and satisfied than the robot that knows the verb concepts from the beginning. In addition to that, we found that verb acquisition ability made the robot perceived by human as being more human-like.

View full abstract

We propose a new authentication method based on actions and spatiotemporal information such as location, elapsed time, and travel distance of a user. To be authenticated, a user performs certain actions at certain points defined with spatiotemporal information. To apply spatiotemporal information to authentication, it is needed that suppressing probability of authentication failure of regular user, for it is not easy to retry authentication. So we propose partial matching authentication which allows partial mistake of user's authentication procedure. Also, we define combination of spatiotemporal information and action as spatiotemporal character, and formalize security evaluation of our method based on it. In addition, we show security effectiveness of our method with an experiment. The experiment showed that false rejection rate of our method is 0.233% and false acceptance rate is 0.010%.

View full abstract