No abstract

We examine whether stock price effects can be automatically predicted analyzing unstructured textual information in financial news. Accordingly, we enhance existing text mining methods to evaluate the information content of financial news as an instrument for investment decisions. The main contribution of this paper is the usage of more expressive features to represent text and the employment of market feedback as part of our word selection process. In our study, we show that a robust Feature Selection allows lifting classification accuracies significantly above previous approaches when combined with complex feature types. That is because our approach allows selecting semantically relevant features and thus, reduces the problem of over-fitting when applying a machine learning approach. The methodology can be transferred to any other application area providing textual information and corresponding effect data.

Linear aliphatic copolyesters consisting of the biobased diols ethylene glycol, 1,3‐propane diol, and 1,4‐butane diol, the biobased dicarboxylic acids succinic acid, adipic acid, and sebacic acid, and a 9,10‐dihydro‐9‐oxa‐10‐phospha‐phenanthren‐10‐oxide (DOPO)‐containing diol or dicarboxylic acid are presented. Structural characteristics of the new copolyesters are studied by 1H and 13C NMR spectroscopy. The influence of chemical composition on the property profile is examined evaluating glass transition temperature Tg, melting behavior, thermal degradation and combustion, mechanical and burning behavior. Incorporation of DOPO monomers suppresses crystallization and yields materials with reduced toughness except in case of sebacates. Tg rises with the content of DOPO monomer. Correlations between the number of methylene groups in the repeating unit and the thermogravimetric analysis degradation maximum, as well as the heat release capacity from pyrolysis combustion flow calorimeter reflect the systematic influence of chemical structure. Copolycondensation with DOPO monomers enhances the limiting oxygen index providing materials with improved overall value.

We have been developing computational approaches to increase our ability to analyze the growing body of three-dimensional structural data with applications centered about the serine proteases. The emphasis of these approaches is to compare and contrast macromolecules at the separate levels of secondary, tertiary, and quaternary structure. Our assumption is that in functionally related molecules, regions of structural and/or physicochemical similarity will exhibit functional similarity; regions that are different in structure and/or physicochemical properties will function differently and, therefore, be the source of specificity. Based on this assumption, the independent observations from studies of these enzymes in solution and in biological systems are combined with the structural observations from X-ray crystallographic analysis. A goal of the present research effort is to probe the biomolecular architecture of the serine proteases to evaluate the role of the three-dimensional structure beyond that of the active site in determining recognition and reactivity determinants for these enzymes, and to determine those principles that might be applied successfully to other enzyme systems as well. Of particular note has been our observation of a macromolecular recognition surface which occurs as a topographical feature outside of the active site and under evolutionary control to produce specificity towards macromolecular substrates and inhibitors. In additon we have established the important role of conformational changes that occur beyond the active site of an enzyme and differentiate between natural and synthetic inhibitor-enzyme interactions. This suggests that the specificity and reactivity determinants of a macromolecule are derived from its architecture and structural organization.