Adaptive user profiling: experimenting on multiple interests and changing interests

Abstract

Self-adaptation is an important capability of applications deployed in dynamically changing environments, such as adaptive information filtering (AIF). Immunological inspiration is typically applied to the computational perspective to promote software with characteristics such as self-organization, adaptation and learning. In this paper, we experiment the features of dynamic clonal selection (DCS) focusing on the gene libraries to produce reasonable coverage of detectors to detect changes of multi interest of topics. We experiment how the capability of DCS through diversity of detector in gene libraries classify multi topics of user's interests. Moreover, we further investigate the ability of the DCS to forget a previous topic when it starts to process a new topic. The results shows that the antibody-antigen interaction of B cells and the introduction of the diverse coverage of detectors helps further improve the capability for identifying new and potentially uninteresting topics thus have a positive effect on the adaptability of the profile to changes in user interests.