No abstract

We propose methods to classify lines of military chat, or posts, which contain items of interest. We evaluated several current text categorization and feature selection methodologies on chat posts. Our chat posts are examples of 'micro-text', or text that is generally very short in length, semistructured, and characterized by unstructured or informal grammar and language. Although this study focused specifically on tactical updates via chat, we believe the findings are applicable to content of a similar linguistic structure. Completion of this milestone is a significant first step in allowing for more complex categorization and information extraction.

We show that the motions of supergranules are consistent with a model in which they are simply advected by the axisymmetric flows in the Sun's surface shear layer. We produce a 10day series of simulated Doppler images at a 15-minute cadence that reproduces most spatial and temporal characteristics seen in the SOHO/MDI Doppler data. Our simulated data have a spectrum of cellular flows with just two components -a granule component that peaks at spherical wavenumbers of about 4000 and a supergranule component that peaks at wavenumbers of about 110. We include the advection of these cellular components by the axisymmetric flowsdifferential rotation and meridional flow -whose variations with latitude and depth (wavenumber) are consistent with observations. We mimic the evolution of the cellular pattern by introducing random variations to the phases of the spectral components at rates that reproduce the levels of cross-correlation as functions of time and latitude. Our simulated data do not include any wavelike characteristics for the supergranules yet can reproduce the rotation characteristics previously attributed to wave-like behavior. We find rotation rates which appear faster than the actual rotation rates and attribute this to the projection effects. We find that the measured meridional flow does accurately represent the actual flow and that the observations indicate poleward flow to 65 • − 70 • latitude with equatorward counter cells in the polar regions.

Abstract. Ranking is an important functionality in a diverse array of applications, including web search, similarity-based multimedia retrieval, nearest neighbor classification, and recommendation systems. In this paper we propose a new method, called Boosted Ranking Model(BRM), for learning how to rank from training data. An important feature of the proposed method is that it is domain-independent, and can thus be applied to a wide range of ranking domains. The main contribution of the new method is that it reduces the problem of learning how to rank to the much more simple, and well-studied, problem of constructing an optimized binary classifier from simple, weak classifiers. Using that reduction, our method constructs an optimized ranking model using multiple simple, easy-to-define ranking models as building blocks. The new method is a unifying framework that includes, as special cases, specific methods that we have proposed in earlier publications for specific ranking applications, such as nearest neighbor retrieval and classification. In this paper we reformulate those earlier methods as special cases of the proposed BRM method, and we also illustrate a novel application of BRM, on the problem of making movie recommendations to individual users.

No abstract