Fuzzy multisets are appropriate as a model of web information retrieval system, because the situations where the same element appears with different memberships several times can be described using fuzzy multisets. This paper aims at proposing fuzzy clustering algorithms of information elements in an information retrieval model based on fuzzy multisets. Furthermore, a kernel function used for the support vector machines is applied to the clustering. An example of the clustering is shown and effect of the kernel function is discussed.

View full abstract

In this paper, we propose a new method of transformation of retrieval conditions from conceptual queries. This method enables efficient retrieval by users' conceptual queries such as "reasonable" or "for youth". In this method, conceptual queries are defined by fuzzy reasoning rules which are constructed by attributes in the target databases. Next, the range of input values for transformation of retrieval conditions is calculated. Inputs within the range output the value, which exceeds a certain threshold based on the defined fuzzy reasoning rules. When the given database is constructed by a simple table, the SQL type queries are easily defined according to the retrieval conditions. Those queries can be applied to conventional DBMS. As a result, the main features of DBMS become available even in user's conceptual query retrieval.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

Structured discussion showing complicated flow of topics greatly helps us discover and understand the main points. In this paper, we propose a new system which visualizes discussion structure by linking the topic segments identified from a minutes of meeting. Based on the structured discussion, we also aim at discovering influential topic segments triggering fruitful discussions by applying Influence Diffusion Model (IBM) The results of questionnaire surveys show that (1) the structured discussion can support user's understanding of discussions instinctively (2) the precision of influential topic segments by IDM is higher than that by in-degree and out-degree measurements.

View full abstract

Acceleration plethysmogram (APG) is applied noninvasively and is used mainly for estimation of arteriosclerosis and blood vessel age. However, physicians of the Chinese school, from ancient times, have applied sphygmopalpation for health examination. This is because APG sensitively reflects the degree of cardiac output, distensibility of the peripheral blood vessels, arteriosclerosis, and the autonomic system. From our previous research employing the recurrence plots method as a chaotic analysis approach on APG for quantitative health examination, it was possible to evaluate healthy and disease states with an accuracy rate of over 86%. But the APG with small amplitude and normal topological structure is evaluated as disease. Based on this result, we carried out new research on classification of disease state; not disease name, and appropriate evaluation for APG with small amplitude and normal topological structure. As a result of our new research, all examinees were estimated quantitatively from the viewpoint of homeodynamical and hemorheological state.

View full abstract

To develop a "computer that can laugh," it is necessary to model the quality of being funny (referred to herein as "laughableness"), which is an internal human emotional response. I sought the basis for such a quality in "social value of an object," as these are comparatively easy to convert into computer models, while incorporating aspects of several well-known theories of laughter. As an example of the software, I used ranks (like those employed in competitive situations) as the variables to express value differences and then devised a formula to represent laughableness.

View full abstract