ايناس فاضل عبدالله التركي

The goal in this research was to transform unstructured textual data for drug datasets into structured texts while integrating sentiment analysis and recom-mendation systems to deviate from conventional recommendation systems once the sentiment feature was extracted from the rating. The procedure used in the research consists of multiple stages: the first involves using natural language processing tech-niques to prepare the data, and the second involves using classification models logistic regression findings to make predictions with a 90% prediction accuracy. The last major problem is creating drug recommendation suggestion lists utilizing two different scenarios: the first one uses the K-nearest neighbor, while the second one uses the cosine similarity.

Data mining is an effective method that uses sophisticated tools and strategies to sift through massive databases in search of useful patterns. Its usefulness extends to many fields, including medicine. We used AdaBoost, Logistic Regression (LR), K-Nearest Neighbors (KNN), and Random Forest (RF) as predictive models in our investigation (ADaB). Evaluating utilizing methods like cross-validation and random sampling allowed us to concentrate on improving accuracy. Medical datasets were used to construct the following: Heart Disease (HD) dataset, Breast Cancer Wisconsin (BCW) dataset, and Covid-19 dataset. We set out to improve the accuracy of disease predictions and push the boundaries of medical data analysis. After completing the assessment procedure with the training data, the performance measurements showed that the highest accuracy was achieved with LR 83% for HD, KNN 97%, and LR 98% for Covid-19.

NLP refers to the computer-based study of language when referring to textual data such as texts or reviews. Natural language processing frequently seeks to give otherwise unstructured natural language a representation of the text that gives it structure. Marketing, social media, and customer relationship management are just a few of the businesses that heavily rely on sentiment analysis (SA), one of the methods used in opinion mining, typically assesses a textual review's structure to see if it conveys a positive or negative impression .In this paper, the challenge was to convert unstructured texts into structured texts for drug datasets, and interlocking between sentiment analysis, the mechanism that was adopted in the paper starts with several stages. The first stage is data preprocessing with natural language processing techniques, and the next stage is the prediction step with classification models are logistic regression(LR), random forest(RF), Naïve Bays(NB), and Support Vector Machine (SVM), the best Result of prediction accuracy 92%,for random forest .

Overthelastseveralyears,sentimentanalysishasemergedasoneofthemostpopularapplicationsofmachinelearning. Itenablestheidentificationofauser’sattitudefromaremark,document,orreview.Asaresultof thedevelopment ofBigData, recommendersystems(RS)arealsofindingmoreuse inmanyaspectsofday-to-dayliving. Thereare threebasickindsofRS: collaborativefiltering, content-based, andhybrid. Thisarticlepresentsaquickdescription of therecommendersystemssupplementedwithasentimentanalysismodule. SentimentAnalysissystemsmayhelp recommendersystemsimprovebyassessingWeb-basedreviews.

Medicaldatamininghasbeenawidespreaddataminingareaof late. Mainly, diagnosingcancers isoneof themost important topics thatmanyresearchers studiedtodevelop intelligentdecision supportsystemstohelpdoctors. Inthisresearch, threedifferentclassifiersareusedtoimprovethe performance interms of accuracy. The classifiers areSupportVectorMachine (SVM),Adaptive Boosting(AdaBoost),andRandomforests(RF).Twomachinelearningrepositorydatasetsareused toevaluateandverifythe classificationmethods. Classifiers are trainedusing the10-foldcrossvalidationstrategy,whichsplitstheoriginalsampleintotrainingandtestingsets. Inordertoassess classifier efficiency, accuracy(AC), precision, recall, specificity,F1, andareaunder thecurveare used(AUC). TheExperimentsshowedthattheAdaBoostclassifier’sachievedanaccuracyof100% whichissuperiorinbothdatasetsincomparisonwithSVMandRFwithACof97%.Theaccuracy isalsocomparedwithanotherstudyfromthepreviousworkthatusesthesamedatasets, andthe resultsdemonstratedthatthecurrentresearchhasbetteraccuracythantheotherstudy.

An ad hoc mobile network can be defined as a set of communication hosts that are linked together with no predetermined topology of wireless links and also without the need for a fixed central point. One of the most challenges that occur in this type of networks is the routing. This paper aims to simulate, analyze, evaluate, and make a comparison to the performance of MANET routing protocols AODV and DSDV. Reactive protocols AODV, proactive protocol DSDV with their characteristics are analyzed based on number of parameters metrics such as throughput packets received, delay, jitter-delay, using NS2