Learning Recursive Bayesian Multinets for Data Clustering by Means of Constructive Induction

This paper introduces and evaluates a new class of knowledge model, the recursive Bayesian multinet (RBMN), which encodes the joint probability distribution of a given database. RBMNs extend Bayesian networks (BNs) as well as partitional clustering systems. Briefly, a RBMN is a decision tree with component BNs at the leaves. A RBMN is learnt using a greedy, heuristic approach akin to that used by many supervised decision tree learners, but where BNs are learnt at leaves using constructive induction. A key idea is to treat expected data as real data. This allows us to complete the database and to take advantage of a closed form for the marginal likelihood of the expected complete data that factorizes into separate marginal likelihoods for each family (a node and its parents). Our approach is evaluated on synthetic and real-world databases.     

Environmental Problems in Sao Paulo: The Challenge for Co-Responsibility and Innovative Crisis Management

This article describes and analyzes the main environmental problems that affect the city of Sao Paulo. The article views the impacts of the local dynamics of urbanization and its effects on daily life. The discussion is centered on the complexity of the environmental management processes and the implementation of conventional and non-conventional policies and programmes to prevent environmental degradation. Possible solutions to alleviate the effects of floods and climate factors on households are debated as well as the possibilities of formulating more participatory oriented policies to engage citizens in the knowledge of environmental risks. These issues represent a very complex challenge to the urban government of Sao Paulo.     

The risk of seroconversion in surgeons of the hepatitis B, hepatitis C and human immunodeficiency viruses (in a specific surgical population)

The prevalence of hepatitis B surface antigen (HBs Ag) and antibody to hepatitis C virus (HCV) and human immunodeficiency virus (HIV) was determined in the serum specimens of 288 patients treated surgically in the orthopaedic department of an urban public teaching hospital. The cumulative risk of HBV, HCV and HIV seroconversion for an orthopaedic surgeon during the surgical career span was calculated. We found that 1.4%, 3.1% and 1.7% of patients were seropositive for HBsAg, HCV antibody and HIV antibody, respectively. Seropositivity was neither associated with age nor with trauma, whereas male patients had a greater likelihood of seropositivity. Risk factor assessment did not prove to be discriminating in identifying which patients may pose a potential exposure risk. This study supports the concept of universal infection control precautions for orthopaedic surgeons regardless of the patients' risk factor or serologic status.     

Screening of supports for the immobilization of pectinmethylesterase from acerola (Malpighia glabra L)

A partially purified extract of pectinmethylesterase (PME) from acerola fruit was immobilized on various supports: glass, celite, chrysotile, agarose, concanavalin A Sepharose 4B, egg shell, polyacrylamide and gelatin. In addition, reticulation with glutaraldehyde was assessed, as well as the use of gelatin in the presence of celite, glass and silica. The highest immobilization yields were obtained when the pectinmethylesterase was immobilized in concanavalin A Sepharose 4B (81.7%) and in gelatin‐water (78.0%). Copyright © 2004 Society of Chemical Industry     

Electrical Conductivity Measurement Through the Loaded Q Factor of a Resonant Cavity

We present a method for measuring the electrical conductivity of metallic materials that relies on the ratio of two loaded Q factors, Q_R/Q_X, with Q_R corresponding to a TE₀₁₁-mode reference cavity made of aluminum, and Q_X the Q that results upon replacing the aluminum plate with the one fabricated from the material to be examined. Electrical conductivity is mathematically inferred from the ratio Q_R/Q_X where the loaded Q factors are measured by using the transmission-type method. Within a 3.0 percent accuracy, conductivities determined at 8.7 GHz for electrolytic copper (5.6 times 10⁷ S/m) and brass (1.6 times 10⁷ S/m) show to be in good agreement with those reported in the literature.     

MODELING WATER LOSS DURING FRYING OF POTATO STRIPS: EFFECT OF SOLUTE IMPREGNATION

The effect of solute impregnation on water loss and oil uptake during potato strip frying was studied. Blanched potato strips were impregnated at 25°C by soaking in a solution of sucrose-NaCl-water, 20-5-75% by weight. After rinsing and air drying, strips were deep fried in sunflower oil at 160, 170 and 180°C. A control treatment, consisting of potato strips blanched but not soaked and later air dried was also conducted. Solute impregnation provided a decrease of the oil uptake. Two models, based on Fick's law were used to describe water loss during frying. The first one is the classic model with an effective moisture diffusion coefficient assumed a constant value. The second model considers that diffusion coefficient varies during the frying process. For a given frying temperature, constant diffusion coefficient for control potatoes resulted in lower values than the impregnated ones. The variable diffusivity model showed a two-stage behavior: during the first stage of frying, diffusion coefficient increased with frying temperatures, but from a given time on an inverse behavior began. This last fact was found to be related to an increase of the measured peak force needed to penetrate the potato crust.     

Identification of aroma components of Spanish ‘Verdejo’ wine

Aroma components of ine resulting from fermentation of grape must from the Verdejo cultivar ere isolated by continuous liquid-liquid extraction and further fractionation on a silica gel column. One hundred and thirty-to substances, including alcohols, esters, carbonyl compounds, terpenes, acids and sulphur compounds, ere identified by using GC-MS; some had not been previously detected in ines. Data are given hich could lead to the characterisation of the Verdejo cultivar and the effect of the elaboration of young ines. The sensory contribution of some volatile compounds to the flavour of Verdejo ine is also studied.     

Modal Kalman filtering based impedance relaying

An algorithm is presented for computer distance relaying. It is based on modal Kalman filtering (MKF) to estimate fundamental phasors, MKF exploits all the measurement information available from a three-phase line and is founded on enhanced models of the electromagnetic transients noise. To support the modeling of noise for different faults, a study of nontrivial electrical networks with accurate electromagnetic transients modeling is made for 400 kV transmission lines. MKF fits the behavior of the different modal noise for the Clarke phasors, but it is invariant for each type of fault, assuring robustness and minimum error variance. To computer distance, Clarke αB phasors are transformed to +/- symmetrical components, and then a well-known expression is used to computer the positive impedance. Statistical analysis if presented to assess observation time versus estimation accuracy for the different types of line faults