Building Fake Review Detection Model Based on Sentiment Intensity and PU Learning.

Fake review detection has the characteristics of huge stream data processing scale, unlimited data increment, dynamic change, and so on. However, the existing fake review detection methods mainly target limited and static review data. In addition, deceptive fake reviews have always been a difficult point in fake review detection due to their hidden and diverse characteristics. To solve the above problems, this article proposes a fake review detection model based on sentiment intensity and PU learning (SIPUL), which can continuously learn the prediction model from the constantly arriving streaming data. First, when the streaming data arrive, the sentiment intensity is introduced to divide the reviews into different subsets (i.e., strong sentiment set and weak sentiment set). Then, the initial positive and negative samples are extracted from the subset using the marking mechanism of selection completely at random (SCAR) and Spy technology. Second, building a semi-supervised positive-unlabeled (PU) learning detector based on the initial sample to detect fake reviews in the data stream iteratively. According to the detection results, the data of initial samples and the PU learning detector are continuously updated. Finally, the old data are continually deleted according to the historical record points, so that the training sample data are within a manageable size and prevent overfitting. Experimental results show that the model can effectively detect fake reviews, especially deceptive reviews.

An analysis on importance of drugs in traditional Chinese medicine for treatment of complications and symptoms of hypertension by QUEST algorithms / 中国中西医结合急救杂志

Objective To approach Professor Zhu Guangli's regular pattern of differentiation of symptoms and signs in patients with hypertension accompanied by complications and in accord with the individual manifestations to add or subtract the ingredients in the prescription.Methods The patients with essential hypertension selected were those who the first time came to the Affiliated Guangxing Hospital of Zhejiang Chinese Medical University or Hangzhou Hospital of Traditional Chinese Medicine (TCM) of Zhejiang Province from September 2012 to January 2015, and were continuously followed up in the hospitals having their clinical data comprehensive. The complications, combination diseases and symptoms of hypertension were conducted as dependent variables, while the applied Chinese medicines were acted as independent variables. QUEST algorithms were employed to analyze the different dependent variables to evaluate the importance contribution rate (ICR) of every kind of Chinese medicine.Results There were 9 types of complications or combination diseases [including hyperlipemia, coronary heart disease, cardiac arrhythmia, hyperglycemia, trace albuminuria, lacunar cerebral infarction, elevation of creatinine in blood (azotemia), liver dysfunction and cardiac dysfunction] and 39 types of symptoms obtained (including dizziness, palpitation, chest oppression, insomnia, lack of strength, soreness of the waist and knees, dry mouth, loose stool, fatigue, apontaneous sweating, headache, shortness of breath, coprostasis, fullness in the stomach, blurred vision, limb numbness, low spirit, chest pain, poor appetite, abdominal distension, excessive dreaming, tinnitus, flush of the face, neck rigidity, hectic fever, bitterness in the mouth, cough, ructus, nocturia, fremitus, stenagma, pharyngalgia, heaviness in head, nocturnal sweating, frequent urination, coughing of sputum, chillness and cold limbs, aphthous stomatitis and sialosis) being the QUEST decision models. All kinds of the TCM employed in all the patients paying visits for 755 times were summed up, and the individual drug frequency of presentation in the prescription was recorded. In the study, 171 sorts of TCM and 10 620 of cumulative frequency (freq.) were involved in 755 prescriptions. The average sorts of TCM was 14.07±1.53 per prescription. The top ten kinds of TCM the most commonly used (freq.) were as follows: tuckahoe (588), Chinese yam (551), wolfberry fruit (496), white peony root (442), abalone shell (405), gastrodia tuber (395), lilyturf root (325), Huai wheat (293), curcuma root (281), rush pith (277), accounting for 38.16% in the total frequency. There were 33 kinds of TCM with frequency > 100, accounting for 74.65% in the total frequency. By the analyses of QUEST algorithms directing to the complications and combination diseases to add or subtract the items of TCM, it was discovered that the drugs with ICR > 5% were as follows: in lacunar cerebral infarction, the drug application frequencies or ICRs of batryticated silkworm and bitter cardamom were 32.69% and 50.74% respectively; in coronary heart disease, the ICRs of pseudostellaria root, astragalus root, white sandal wood and lepidium seed were 23.05%, 17.70%, 15.31% and 5.06% respectively; in cardiac arrhythmia, the ICRs of astragalus root, flavescent sophora root, Huai wheat and dragon tooth and dragon bone were 46.65%, 11.56%, 7.56%, 7.13% and 6.28% respectively; in cardiac dysfunction, the ICRs of lepidium seed and white hyaciath bean were 92.97% and 7.03%; in hyperlipemia, the ICRs of cassia seed, white sandal wood, dried rehmannia root and sargentgloryvine stem were 24.26%, 12.47%, 10.51% and 5.81% respectively; in hyperglycemia, the ICRs of trichosanthes root and dried hawthorn fruit were 51.02% and 30.18%; in trace albuminuria, the ICR of snow in June herb was 83.33%; in elevation of creatinine in blood, the ICRs of Chinese actinidia root, herb of snow in June, glossy privet fruit and centella were 81.37%, 6.21%, 6.21% and 6.21% respectively; in liver dysfunction, the ICRs of coral ardisia root, purse grass and herb of stringy stonecrop were 86.37%, 6.82% and 6.82% respectively. When the symptoms were used as the dependent variables, for the top 3 commonly seen symptoms, in the prescription, the items of TCM used with ICR > 5% were as follows: in cases with elevation of blood pressure, parasitic loranthus (ICR 89.59%) was used; with dizziness, gastrodia tuber (ICR 46.96%) and/or abalone shell (ICR 33.61%) were added; headache, rhizome of Chinese ligusticum and/or rhizome of Sichuan lovage and/or large gentian with ICR at 85.77%, 7.11% and 7.11% respectively. Conclusion Professor Zhu's essential thought in differentiation of disease and expectant treatment of hypertension in accord with the individual manifestations to add and subtract drugs in TCM is under the guidance of TCM principle of differentiation of symptoms and signs for treatment, and based on the classical theories in TCM, the reliable drugs whose actions are demonstrated clinically and experimentally are preferentially selected.

Basing on the Original Language and Ideas Would Make A Better Way to Understand Treatise on Febrile Diseases / 浙江中医药大学学报

Objective Our purpose was to find out a better way for modern learner to understand the ancient medical book named as Treatise on Febrile Diseases. [Methods] We adopted the way of document research to find out the key points for modern learners. [Results] There were many interchangeable words, ancient and modern words and Nanyang dialect speech in the book. Further more, some drugs had been evolved in a state of flux. [Conclusion] Our conclusion was: To make a better understanding of Treatise on Febrile Diseases, we must put our fingers on the original language and ideas, not just focus on the clauses. And we cannot use contemporary language to explain ancient words;otherwise we would misunderstand the truth.

A Retrospective Survey on Nosocomial Infection of Inpatients in Tianjin Nankai Hospital in 2004 / 中国慢性病预防与控制

Objective To analyze the prevalence of nosocomial infections among inpatients and its influential factors.Methods The retrospective survey was performed by reviewing and analyzing the clinical data of 7788 inpatients in 2004 in Nankai Hospital of Tianjin.Results Totally 250 nosocomial infection cases were identified,which gave a prevalence of 3.21%.The all-year highest prevalence was found in September.Among all the departments,the highest rate,5.20%,occurred in Department of Gynecology.The inpatients aged more than 60-year-old showed higher prevalence than those under 60 years old.The main infection position was lower respiratory tract(41.60%).Concludes The prevalence of nosocomial infections of Nankai Hospital were within the acceptable standard range defined by Ministry of Health,China.Nevertheless,the prevalence would be reduced if measurements for surveillance and control could be properly applied.

Phenotype and Genotype of Nosocomial Infection Strains of Klebsiella pneumoniae / 中华医院感染学杂志

OBJECTIVE To investigate the prevalence,antibiotic resistance and genotype of the extended-spectrum ?-lactamases(ESBLs)-producing clinical isolates of Klebsiella pneumoniae.METHODS A total of 104 isolates of K.pneumoniae were examined for the ESBLs production and the susceptibilities of the bacteria to 15 antimicrobial agents.PCR was performed to detect the genes encoding the ESBLs belonging to SHV and TEM families as well as CTX-M-1 and CTX-M-9 groups.RESULTS The ESBLs-producers of K.pneumoniae were 54.0% in the total of 104 isolates.Almost all of the ESBLs-producing isolates were resistant to the antibiotics commonly used,and only remained susceptible to carbapenems and the combination of cefoperazone with sulbactam.The genes of SHV,CTX-M-1 and TEM groups were detected in the ESBLs-producing isolates by 64.3%,46.4%,and 32.1%,respectively,and 35.7% and 8.9% of ESBLs-producing K.pneumoniae strains carried two and three genes.CONCLUSIONS The clinical isolates of K.pneumoniae in Tianjin Nankai Hospital are shown a high rate of ESBLs-producing and antibiotic resistance.SHV and CTX-M-1 groups of ESBLs are the dominant genotypes in the isolates of ESBLs-producing K.pneumoniae.