Unsupervised Learning of a Chinese Spontaneous and Colloquial Speech Lexicon with Content and Filler Phrase Classification

There is significant lexical difference—words and usage of words-between spontaneous/colloquial language and the written language. This difference affects the performance of spoken language recognition systems that use statistical language models or context-free-grammars because these models are based on the written language rather than the spoken form. There are many filler phrases and colloquial phrases that appear solely or more often in spontaneous and colloquial speech. Chinese languages perhaps exemplify such a difference as many colloquial forms of the language, such as Cantonese, exist strictly in spoken forms and are different from the written standard Chinese, which is based on Mandarin. A conventional way of dealing with this issue is to add colloquial terms manually to the lexicon. However, this is time-consuming and expensive. Meanwhile, supervised learning requires manual tagging of large corpuses, which is also time-consuming. We propose an unsupervised learning method to find colloquial terms and classify filler and content phrases in spontaneous and colloquial Chinese, including Cantonese. We propose using frequency strength, and spread measures of character pairs and groups to extract automatically frequent, out-of-vocabulary colloquial terms to add to a standard Chinese lexicon. An unsegmented, and unannotated corpus is segmented with the augmented lexicon. We then propose a Markov classifier to classify Chinese characters into either content or filler phrases in an iterative training method. This method is task-independent and can extract even mixed language terms. We show the effectiveness of our method by both a natural language query processing task and an adaptive Cantonese language-modeling task. The precision for content phrase extraction and classification is around 80%, with a recall of 99%, and the precision for filler phrase extraction and classification is around 99.5% with a recall of approximately 89%. The web search precision using these extracted content words is comparable to that of the search results with content phrases selected by humans. We adapt a language model trained from written texts with the Hong Kong Newsgroup corpus. It outperforms both the standard Chinese language model and also the Cantonese language model. It also performs better than the language model trained a simply by concatenating two sets of standard and colloquial texts.     

QUANTITATIVE ANALYSIS OF DIACETYL,PENTANEDIONE AND THEIR PRECURSORS DURING BEER FERMENTATION BY AN ACCURATE GC/MS METHOD

A GC/MS method previously described for diacetyl was developed for the quantification of 2,3-pentanedione, and the derivatization procedure was modified for the determination of α-acetohydroxy acid. The reaction of 2,3-pentanedione with 4,5-dichloro-1,2-diaminobenzene produced 6,7-dichloro-2-methyl-3-ethylquinoxaline (DCMEQ), which was extracted with toluene and determined by gas chromatography using a mass selective detector. The formation of DCMEQ was linearly correlated with the 2,3-pentanedione concentration. The method was very simple and sensitive. The detection limit of the 2,3-pentanedione derivative and diacetyl derivative was 0.0007 mg/litre and 0.0002 mg/litre of the toluene extract respectively, and the determination limit was 0.001 mg/litre and 0.0007 mg/litre, respectively. Cautious sample treatment led to a low (10%) and controlled conversion of α-acetohydroxy acids to vicinal diketones. This reproducible percentage of conversion made it possible to determine precisely free vicinal diketones and α-acetohydroxy acids . The method was applied to the determination of precursors and vicinal diketones concentrations during beer fermentation .     

Iron-dependent oxidation, ubiquitination, and degradation of iron regulatory protein 2: implications for degradation of oxidized proteins

The ability of iron to catalyze formation of reactive oxygen species significantly contributes to its toxicity in cells and animals. Iron uptake and distribution is regulated tightly in mammalian cells, in part by iron regulatory protein 2 (IRP2), a protein that is degraded efficiently by the proteasome in iron-replete cells. Here, we demonstrate that IRP2 is oxidized and ubiquitinated in cells before degradation. Moreover, iron-dependent oxidation converts IRP2 into a substrate for ubiquitination in vitro. A regulatory pathway is described in which excess iron is sensed by its ability to catalyze site-specific oxidations in IRP2, oxidized IRP2 is ubiquitinated, and ubiquitinated IRP2 subsequently is degraded by the proteasome. Selective targeting and removal of oxidatively modified proteins may contribute to the turnover of many proteins that are degraded by the proteasome.     

Comparison of measurements made with two different instruments of the same atmospheric vertical profile

The validation of atmospheric remote-sensing measurements involves the comparison of vertical profiles of atmospheric constituents obtained by different instruments. This operation is a complex one because it has to take into account the measurement errors that are described by the variance-covariance matrices and the different features of the two observing systems that are described by the averaging kernels. The procedure is discussed and a method of comparison that is rigorous and does not involve degradation of the available information is developed by use of the formalism of functional spaces. The functional spaces that can be used for representation of the two profiles are reviewed, and criteria are determined for the choice of the most convenient functional space to minimize degradation of the measurements. Once the functional spaces are chosen, the components of the profiles are compared in the intersection space of these two functional spaces. If the intersection space coincides with the null vector, a pseudointersection space with useful geometrical properties can be used instead. A test of the method is made with a realistic simulation. In the test the profiles retrieved by two real instruments are simulated and quantitatively compared.     

The effect of safety initiatives on safety performance: a longitudinal study

Construction industry is one of the most dangerous industries, not only in the USA, but worldwide. In this longitudinal study we examined the effects of safety initiatives on the safety performance of construction companies. One of the measures commonly used in the USA to track a company's safety performance is the experience modification rate (EMR). The EMR is based on the company's safety records (injury claims) from the past three full years and is used to calculate the workers' compensation insurance premiums. In a longitudinal study, we studied the effects of safety efforts and initiatives on the EMR. The results show that safety initiatives and money spent on safety do improve safety performance, but only over time.     

Magnetoresistivity of CeSi1.86

We present measurements of the magnetoresistance up to 20 teslas from 2 K to room temperature on a CeSi_1.86 single crystal. This alloy behaves as an anisotropic Kondo lattice where strong ferromagnetic correlations are dominant. We observe a clear relationship between the magnetotransport and the magnetic susceptibility of this compound.     

Numerical investigations in Rayleigh breakup of round liquid jets with VOF methods

Simulations of the growth of a capillary instability and of the breakup of a jet were carried out using a one-fluid model to describe the two-phase flow motion and a VOF approach to capture the interface. The model considered each phase as fictitious sub-domains and accounted implicitly for jump conditions at the interface through a unique set of equations for which a source term of surface tensions appeared in momentum equations. The predominance of capillary effects in the breakup mechanism required to accurately describe the surface tension contribution. The Brackbill surface model was chosen because of its simplicity to represent tension forces, although it was known to generate parasitic currents susceptible to limit its precision. The flow incompressibility was ensured with an augmented Lagrangian method in case of sequential calculations and by a predictor/corrector approach for 3D simulations that required parallel computations. As a first step, the numerical methods were validated by simulating the growth of a capillary instability and comparing results to those predicted by the Rayleigh theory for capillary instabilities. The consistency of the Brackbill surface tension model and the accuracy of the methods were evaluated via a convergence study. As a second step, the simulation of a jet breakup was carried out using water as injected liquid and compressed carbon dioxide as surrounding medium. It was shown that the simulation predicted accurately the breakup length and the droplet size evidenced experimentally in literature.