The Behavior Informatics 2010 (BI2010) will be Held in conjunction with
The 14th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD2010)
The BI2010 Proceedings will be published in the PAKDD2010 Workshop Proceedings by Springer-Verlag as a volume of LNCS/LNAI series.
Deep and quantitative behavior analysis such as in social network cannot be supported by methodologies and techniques in traditional behavioral sciences due to the behavior implication in normal transactional data. Therefore, it is imperative to develop new behavioral analytics technologies that can derive an accurate understanding of human behaviors beyond the demographic and historical tracking. This leads to the emergence of the inter-disciplinary Behavior Representation, Modeling, Analysis, Mining and Management (namely Behavior Informatics).
The BI2010 workshop provides a premier forum for sharing research and engineering results, as well as potential challenges and prospects encountered in Behavior Informatics, for instance:
- Behavior modeling: formalizing behaviors, relationships, impact and networks.
- Impact-oriented behavior mining: behaviors associated with high impacts are of particular importance, while impact-oriented behaviors are often sparse, rare and imbalanced isolated in business and data; identify impact-oriented behavior patterns involves different pattern types and computational challenges.
- Analysis of behavior social networks handling challenging issues such as convergence and divergence of behavior, and the evolution and emergence of hidden groups and communities.
- Extracting discriminative behavior patterns from high-dimensional, high-frequency, high-density, and huge amount of data.
- Large intra-class variance between behaviors: Due to the highly overlapped nature of behavior data, it is extremely difficult to build a robust behavior model which is tolerant for one behavior category while differentiate amongst other categories.
- Behavior data processing from transactional space to behavior feature space: Customer demographic and transactional data is generally privacy-oriented, distributed and not organized in terms of behavior but entity relationships. In such transactional entity spaces, behavioral elements are dispersed and hidden within complex business applications with weak or no direct linkages. As a result, current behavior analysis which focuses on exterior features in demographic and service usage data cannot effectively and explicitly scrutinize human behavior patterns and impacts on businesses. To support genuine behavior analysis on behavior interior, a challenging task is to extract and transform transactional behavior-related elements into explicit behavior features.