21241 research outputs found
Sort by
Analysis of Turkish spam mails with sentiment analysis and machine learning methods
Get PDFÇevrimiçi platformların kullanımının artmasıyla birlikte metin verilerinin hacmi artmakta ve bu verilere erişim kolaylaşmaktadır. Bu durum metin sınıflandırma alanında yapılan çalışmaların sayısının artmasına neden olmuştur. Özellikle spam tespiti ve duygu analizi gibi alanlarda metin sınıflandırma teknikleri büyük önem taşımaktadır. Literatürde İngilizce metinler üzerine yapılan çalışmaların sayısı oldukça fazla olmasına karşın Türkçe veriler üzerine yapılan çalışmalar oldukça kısıtlıdır. Bu çalışmanın amacı, Türkçe maillerin duygu analizi ve makine öğrenmesi teknikleri ile morfolojik analizini gerçekleştirmek ve modellerin spam ve normal mailleri tespit etmedeki başarısını karşılaştırmaktır. Bu amaçla literatürde yer alan iki Türkçe mail veri seti kullanılmıştır. Bu veri setleri spam ve normal olarak etiketlenmiş maillerden oluşmaktadır. Çalışma kapsamında bu iki veri setinden bir veri seti elde edilmiştir. Bu veri kümesine üç işlem uygulanmıştır ve bu uygulanan işlemler sonucu üç adet veri seti elde edilmiştir. İlk veri seti, verilere temel veri ön işleme adımları uygulanarak oluşturulmuştur. Bu adımda sırasıyla veriler küçük harflere dönüştürülmüştür. Daha sonra web sitesi adları "website" ve mail adresleri "email" olarak yeniden adlandırılmıştır. Buna ek olarak ilk veri seti olarak noktalama işaretleri ve sayısal ifadelerin kaldırılması elde edilmiştir. İkinci veri seti, ilk veri setinden Türkçe kökenli olmayan kelimeler ve dört harften kısa sözcüklerle oluşurulmuştur. Üçüncü veri seti ise ikinci veri seti ile birinci veri setinin kesişim kümesinden elde edilmiştir. Bu çalışma ile literatürde yer alan çalışmalarda etkisi göz ardı edilen Türkçe veri setleri içerisindeki Türkçe kökenli olmayan kelimelerin de sonuçlar üzerindeki etkisi gözlemlenmiştir. Bu veri kümeleri K-means ve Isolation Forest yöntemleri ile kümelenmiş ve bu yöntemlerin performansı değerlendirilmiştir. Ayrıca bu veri kümeleri üzerinde duygu analizi yapılarak spam ve normal maillerin duygu durumları gözlemlenmiştir. Son olarak veriler Naive Bayes, Random Forest, Logistic Regression ve Support Vector Machine sınıflandırma algoritmaları ile sınıflandırılmış ve yöntemlerin sonuçları doğruluk, kesinlik, geri çağırma ve f1-skor kriterleri ile değerlendirilmiştir. Çalışma sonucunda en yüksek başarım puanlarına ilk veri seti ile ulaşılmıştır. Naive Bayes ve destek vektör makinesi 0,92 doğruluk değeri ile en başarılı sonucu verirken Lojistik Regresyon ile 0,90 ve Random Forest ile 0,89 doğruluk değerleri elde edilmiştir. K-means ve Isolation Forest, orijinal veri kümesindeki etiketlere kıyasla verileri etiketlemede yetersiz kalmıştır. Yapılan işlemler sonucunda maillerin kategorik morfolojisi çıkarılmıştır.The increasing use of online platforms has led to a growth in the volume of text data, and access to these data has become easier. This has resulted in a rise in the number of studies conducted in the field of text classification. Text classification techniques are particularly important in areas such as spam detection and sentiment analysis. While there is a significant number of studies on English texts in the literature, studies on Turkish data are quite limited. The aim of this study is to perform sentiment analysis and morphological analysis of Turkish emails using machine learning techniques, and to compare the success of the models in detecting spam and normal emails. For this purpose, two Turkish email datasets from the literature, which are labeled as spam and normal, were used. A single dataset was obtained from these two datasets. Three processing steps were applied to this dataset, resulting in three datasets. The first dataset was created by applying basic data preprocessing steps to the data, such as converting to lowercase, renaming website names to "website" and email addresses to "email", and removing punctuation marks and numerical expressions. The second dataset was created from the first dataset by removing non-Turkish words and words shorter than four characters. The third dataset was obtained from the intersection of the second and first datasets. This study observed the impact of non-Turkish words in Turkish datasets, which has been overlooked in previous studies. These data sets were clustered using K-means and Isolation Forest methods, and the performance of these methods was evaluated. Additionally, sentiment analysis was performed on these data sets to observe the sentiment states of spam and normal emails. Finally, the data was classified using Naive Bayes, Random Forest, Logistic Regression, and Support Vector Machine classification algorithms, and the results of the methods were evaluated using accuracy, precision, recall, and F1-score criteria. As a result of the study, the highest performance scores were achieved with the first dataset. Naive Bayes and Support Vector Machines achieved the most successful results with an accuracy of 0.92, while Logistic Regression achieved 0.90 and Random Forest achieved 0.89 accuracy. K-means and Isolation Forest were insufficient in labeling the data compared to the original dataset labels. As a result of the performed operations, the categorical morphology of the emails was extracted
İnfluence of thermal pretreatments on dimensional change and humidity sensitivity of densified spruce and poplar wood
Get PDFDensification modification is an effective method to improve many properties of wood. However, densified wood is sensitive to humidity and is not dimensionally stable. The effect of thermal pretreatments on the dimensional change and humidity sensitivity of densified Picea orientalis (spruce) and Populus nigra (poplar) wood were investigated. A thermal pre-treatment was applied on the wood specimens at 140 degrees C, 160 degrees C, 180 degrees C, and 200 degrees C for 7 h and 9 h. Wood specimens were then compressed at ratios of 20 % and 40 % at a temperature of 150 degrees C. The results showed that spring-back and thickness swelling increased in all specimens (thermally pre-treated and untreated) depending on the increase in compression ratio. However, set-recovery wasdetermined higher at20 % compression ratio.Theequilibrium moisturecontentvaluesof untreated specimens and thermally pre-treated specimens at low temperatures (140 degrees C and 160 degrees C) were found lower than uncompressed specimens. The impact of compression ratio on equilibrium moisture content was not clear. Thermal pretreatments significantly affected the dimensional stability and hygroscopicity of densified specimens (especially poplar wood). Depending on the increase in thermal pre-treatment temperature and duration, spring-back, set-recovery and thickness swelling in wood specimens decreased up to 31 %, 67 % and 62 %, respectively. In addition, equilibrium moisture content and water absorption decreased with the increase in thermal pre-treatment temperature and duration. Moreover, the thermal treatment temperature was more important than duration on the investigated properties.Research Fund of Duzce University [BAP-2018.07.01.673]The authors are grateful for the support of the Research Fund of Duzce University, Grant No. BAP-2018.07.01.673
Biosourced polymeric cryogels for future biomedical applications with remarkable antimicrobial activities and tribological properties
Get PDFCryogels, known as a subclass of hydrogels, are promising biomaterials to use in various biotechnological fields. In recent years, applications of antimicrobial hydrogels with improved antimicrobial activities, high biocompatibility, and physicochemical stability have attracted attention as an alternative to using antimicrobial drugs against microbial interactions that may threaten human health, which may even result in death. In this paper, we investigated in detail the biological activities and tribological performances of the previously characterized 2hydroxyethyl methacrylate (HEMA)-based amphiphilic cryogels (PHEMA-PLinaOH) (HC series) that contain hydroxylated polymeric linoleic acid (PLinaOH) as biosource. The biocompatibilities of these cryogels were examined against human embriyonic kidney (HEK293) cell line with MTT assay and acridine orange/ethidium bromide (AO/EB) dual staining. The antimicrobial activities of the materials were extensively investigated against Staphylococcus aureus ATCC 29213 and Pseudomonas aeruginosa PA01 besides four different strains of the yeast Saccharomyces cerevisiae BY4741 by using biofilms eradication, antibiofilm activity and colony forming unit assays. Additionally, the possible morphological changes in microbial cells were evaluated by taking FESEM images. The tribological performances of the cryogels were evaluated in terms of their applicability for future biomedical applications such as artificial articular cartilage or tissue scaffold. Our results showed that while the cryogels did not show significant inhibition on HEK293 cell viability and intensive live cell population was observed after AO/EB staining, they exerted remarkable antimicrobial activities against all studied bacterial and fungal strains. The morphological deformations including the decrease in EPS density and formation of holes were recorded for bacteria and yeast cells with FESEM images, respectively. Finally, it was determined that the increase in the fatty acid ratio contributes positively to tribological properties of the cryogels. All the results indicate that these polymeric cryogels might be considered potential biomaterials for future tissue-engineering studies
Learning Attitudes of Higher Education Students: A Comparison Based on a CHAID Analysis among Turkic Republics
Get PDFThe aim of this study is to determine the factors affecting university students' attitudes towards learning at the primary level in the Turkic Republics and to make a comparison between the Turkic Republics. The study used quantitative methodology and a cross-sectional survey model. A total of 1868 university students from five different Turkic Republics (Turkey, Kazakhstan, Uzbekistan, Azerbaijan, and Kyrgyzstan) participated in the study. Students in Turkmenistan were not included in the sample as they could not be contacted. Data were collected using the Attitude Towards Learning Scale (ATLS) and analysed using SPSS. According to the results, the students with the highest attitude towards learning among higher education students in five different Turkic republics are in Uzbekistan, while the lowest attitude level is in Kazakhstan. There are significant differences in attitudes towards learning between countries. According to the results of the CHAID analysis, the attitudes of higher education students in the Turkic Republics towards learning are determined by variables such as following academic developments related to their fields, striving to learn new things, time allocated for individual learning, regular reading of books, and performance level. In particular, providing support to improve students' attitudes towards learning and expanding learning areas in educational can be beneficial
An innovative composite elbow manufacturing method with 6-axis robotic additive manufacturing for fabrication of complex composite structures
Get PDFFilament winding method is the most commonly used method to produce profiles with different cross sections as composite product manufacturing. In this method, fiber material is wound with resin at different angles on a mold that has a suitable cross section shape. As a winding strategy, angled and helical winding can be done. Motion planning for this process is done with geodesic and nongeodesic theories. Requirement to use mold in the filament winding method increases the cost. Also, there is an obligation to helical windings. In winding of different layers, 90 degrees angle cannot be given between the layers. To overcome all these constraints, UV curing can be achieved using photopolymer resin and continuous fiber glass fiber with the help of robotic additive manufacturing technology. Toolpath strategies for production has a key role in this work. As a tool path strategy, nonplanar slicing can be done and manufactured composite elbow in angular layers without mold. Then, under favour of 6-axis mobility of the industrial robot arm, layers can be obtained at exactly 90 degrees angle. In addition, in this method, unlike other winding methods, internal voids, i.e. a filling rate, can be given within the cylindrical encircled layers. In order to verify whether the elbows produced with this method meet the requirements of the desired applications in the industry in terms of mechanical properties, at different filling rates (50%, 75%, 100%), winding turns (0 and 1/8), and different fiber densities (45%, 55% and 65%) 90 degrees curved composite elbows were produced and their internal pressure strength tests were tested. Afterwards, an optimization study was carried out with the Taguchi method for the production parameters that will maximize the internal pressure strength. According to the results of the optimization study, it is seen that it is appropriate to choose the printing parameters that will obtain the highest internal pressure strength values for production with this method, 100% fill rate, 65% fiber density and 0 degrees winding angle. The products made of this process have the advantage of easy-shaping, reasonable ratio of axial strength and encircled strength, specification easy-unifying, stable product quality
Improvement of worsened diesel and waste biodiesel fuelled-engine characteristics with hydrogen enrichment: A deep discussion on combustion, performance, and emission analyses
Get PDFDue to the strict emission policies, fuel researchers are dedicated to mitigating the tailpipe emissions from internal combustion engines (ICEs). Therefore, researchers have considered biodiesel as the best alternative to conventional diesel fuel (D) for a while. However, many scientific papers experimentally announced that the use of biodiesel significantly worsens engine behaviors. In this framework, hydrogen enrichment has become a very reasonable option in order to minimize the reverse influences of biodiesel-fuelled engine characteristics. In this direction, waste cooking of 25% (B25) was volumetrically blended to D and reference data was collected. Then, 15 and 30 Lpm hydrogen was introduced from the intake manifold by mixing with air along with B25 test fuel to observe the changes from the hydrogen effect. Tests were performed on a three-cylinder, water-cooled diesel engine at constant engine speed (2000 rpm) and variable engine loads (15, 30, 45 and 60 Nm). In the results, it is witnessed that BSFC (brake specific fuel consumption) for B25 fuel increased by 8.23% as compared D fuel. However, along with the introduction of 15 and 30 Lpm hydrogen to B25 fuel, the BSFC value dropped by 17.58%, and 30.75%, respectively. In a similar way, B25 test fuel reduces BTE (brake thermal efficiency) by 7.54% as compared to D fuel. However, the hydrogen introduction of 15 and 30 Lpm (Litre per minute) along with B25 fuel improves the BTE value by 10.19%, and 17%, respectively. On the other hand, the inclusion of 15 Lpm and 30 Lpm H2 to B25 fuel provided a reduction of 23.75% and 45.59% for HC (Hydrocarbon) emissions, and 53.1% and 62.6% for NOx (Nitrogen oxide) emissions, respectively. In conclusion, it is seen that deteriorations in combustion, performance, and emission characteristics resulting from the use of biodiesel can be minimized by using hydrogen for ICEs.Duzce University [2021.06.05.1200]The authors would like to thank Duzce University for its financial support (Project number: 2021.06.05.1200)
Improving the prediction of biochar production from various biomass sources through the implementation of eXplainable machine learning approaches
Get PDFExamining the game-changing possibilities of explainable machine learning techniques, this study explores the fast-growing area of biochar production prediction. The paper demonstrates how recent advances in sensitivity analysis methodology, optimization of training hyperparameters, and state-of-the-art ensemble techniques have greatly simplified and enhanced the forecasting of biochar output and composition from various biomass sources. The study argues that white-box models, which are more open and comprehensible, are crucial for biochar prediction in light of the increasing suspicion of black-box models. Accurate forecasts are guaranteed by these explainable AI systems, which also give detailed explanations of the mechanisms generating the outcomes. For prediction models to gain confidence and for biochar production processes to enable informed decision-making, there must be an emphasis on interpretability and openness. The paper comprehensively synthesizes the most critical features of biochar prediction by a rigorous assessment of current literature and relies on the authors' own experience. Explainable machine learning techniques encourage ecologically responsible decision-making by improving forecast accuracy and transparency. Biochar is positioned as a crucial participant in solving global concerns connected to soil health and climate change, and this ultimately contributes to the wider aims of environmental sustainability and renewable energy consumption
Okul Müdürlerinin Baş Etmekte Zorlandıkları Öğretmen, Öğrenci ve Veli Kaynaklı Sorunlar
Get PDFBu çalışmanın amacı, okul müdürlerinin öğretmen, öğrenci, veli kaynaklı karşılaştıkları sorunları belirlemek, deneyimlerini ve çözüm önerilerini incelemektir. Çalışmada nitel araştırma desenlerinden biri olan durum çalışması kullanılmıştır. Araştırma, 2022 yılının Temmuz-Ağustos aylarında, Düzce ilinde bulunan yedi ilkokul, beş ortaokul, üç ilkokul-ortaokul müdürü olmak üzere toplam 15 okul müdürü ile yürütülmüştür. Çalışma grubunun seçiminde amaçlı örnekleme yöntemlerinden biri olan maksimum çeşitlilik örneklem yöntemi kullanılmıştır. Araştırmada nitel veri toplama yöntemi olarak yarı yapılandırılmış görüşme soruları kullanılmıştır. Görüşmeler sonucu ile elde edilen veriler nitel veri analizi türlerinden olan içerik analizi ile çözümlenmiştir. Araştırma sonunda, okul müdürlerinin baş etmekte zorlandıkları öğretmen tutum ve davranışları temasında, iletişim sorunları ve iş etiği alt temaları; veli tutum ve davranışları temasında, iletişim sorunları, güven sorgulamaları ve ilgisiz veli tutumları alt temaları; öğrenci tutum ve davranışları temasında, iletişim sorunları, davranış sorunları, disipline edilemeyen öğrenciler ve ilgisiz öğrenciler alt temalarına ulaşılmıştır. Genel olarak tüm sorunların çözümü için, öncelikle okul müdürü, problemi doğru tespit edebilmeli, problemleri çözmeye niyeti olmalı ve çözüm odaklı bir yaklaşım sergileyebilmelidir
Cognitivist Presumptions of Moral Realism in Justification of Moral Truths
Get PDFThis study critically examines the foundational principles of impartiality and value independence advocated by moral realist epistemologies in the pursuit of objectivity. Central to moral realists is the cognitivist presupposition necessitating a clear distinction between cognitive and emotional components inherent in moral judgments. The investigation focuses on the cognitive-emotional dichotomy underlying the moral realist perspectives of David Enoch and Thomas Nagel. The research findings unveil that the interplay between cognition and emotion, as evidenced by experimental data, poses a formidable challenge to the traditional understanding of impartiality and value independence. The article's initial section delves into the ontological nature of moral judgments, followed by an exploration of the cognitive assumptions shaping Nagel and Enoch's conceptualizations of objectivity. The final section elucidates the cognitive-emotional interdependence that disrupts the conditions of impartiality and value independence, conventionally posited as prerequisites for objectivity
Antibacterial Activity of Boron Compounds Against Biofilm-Forming Pathogens
Get PDFThis study aimed to evaluate the antibacterial activity of nine boron derivatives against biofilm-forming pathogenic bacteria. The effect of boron derivatives (CMB, calcium metaborate; SMTB, sodium metaborate tetrahydrate; ZB, zinc borate; STFB, sodium tetra fluorine borate; STB, sodium tetraborate; PTFB, potassium tetra fluor borate; APTB, ammonium pentabo-rate tetrahydrate; SPM, sodium perborate monohydrate; Borax, ATFB, ammonium tetra fluorine borate) on bacteria isolated from blood culture was determined by the minimum inhibitory concentration (MIC) method. Then, biofilm formation potentials on microplates, tubes, and Congo red agar were examined. The cytotoxicity of boron derivatives was determined by using WST-1-based methods. The interaction between the biofilm-forming bacteria, fibroblast cells, and boron derivatives was determined with the infection model. We found that the sodium metaborate tetrahydrate molecule was effective against all pathogens. According to the optical density values detected at 630 nm in microplates, meticillin-resistant Staphylococcus aureus was observed to have the most substantial biofilm ability at 0.257 nm. As a result of cytotoxicity studies, it has been determined that a 1 & mu;g/L concentration of boron derivatives is not toxic to fibroblast L929 cells. In cell culture experiments, these boron derivatives have very serious inhibitory activity against biofilm-forming pathogens in a short treatment period, such as 2-4 h. Furthermore, using these molecules on inanimate surfaces affected by biofilms would be appropriate instead of living cells