Intelligent Technologies for Scientific Research and Engineering

When dealing with unstructured data, information extraction is crucial for identifying named entities and events, which is essential for natural language processing. The exponential growth of information in the agricultural sector has made data extraction increasingly challenging. Despite the widespread use of deep learningbased approaches in smart farming for tasks such as crop cultivation, identifying diseases, weed removal, and yield production, the persistent impacts of meteorological, soil, pest, and fertilizer data make it challenging to discover the semantic relationships among the extracted data. The paper is divided into two sections. First, we present a data preprocessing method for cleaning up input corpora of any ambiguity; Furthermore, we provide a novel method for agricultural named entity identification, events, and relations that is based on deep learning techniques. This method utilizes multilayer perceptrons and the Adam optimizer with corrective capabilities. Agricultural, meteorological, soil, along with insect and fertilizer corpora, were used to train and assess the proposed algorithm. On common measures such as accuracy, sensitivity, and specificity, the experimental findings show that the proposed algorithm beats current approaches like Weighted-SOM, LSTM+RAO, PLR-DBN, the KNN, as well as Naïve Bayes.

There has to be a better design of screening devices to meet the rising need for health screening. We develop health screening machine learning models and suggest screening elements to provide individualized health care. For each screening item, a machine learning algorithm is created while the system is offline using guidelines and clinical findings from institutions to construct a synthetic data set. A customer's health states, as well as machine learning algorithms, are used by the proposed server online to provide a real-time inventory of screening items. The performance investigation showed that for 1,000 users online at once, the server responded in under a second, and the learning model's accuracy was close to 100%. This work introduces a recommendation system that can automatically adjust to changes in the screening environment.

Academic and industry researchers have found sentiment categorization to be an intriguing subject due to the exponential expansion of online reviews. Collecting annotated training data is challenging, but the reviews help with several areas. Despite the development of several sentiment categorization algorithms, information retrieval remains inaccurate, inefficient, and sluggish when it comes to convergence speed. The authors of this article propose using the Spider Monkey Crow Optimization approach (SMCA) as a sentiment classification paradigm for training Deep Recurrent Artificial Neural Networks (DeepRNN). In order to remove inappropriate information and save the user's search time, the telecom assessment is used to eliminate stemming and stop words. In contrast, review sentiment is extracted using SentiWordNet for feature extraction. In addition to the extracted SentiWordNet features, DeepRNN also takes into account supplementary data like hashtags, numerical values, expanded words, and punctuation when classifying emotions. In order to get the required review, we use the distance metric Fuzzy K-Nearest Neighbor (FuzzyKNN). The proposed SMCA-based DeepRNN has been evaluated and tested extensively, and its results in terms of precision (95.5%), accuracy (97.7%), recall (94.6%), and F1-score (96.7%) are superior to those of its competitive models.

A lot of individuals have been devastated by the COVID-19 epidemic, which has caused a lot of worry, dread, and confusing emotions. A wide range of complicated emotions has surfaced after coronavirus immunizations were introduced. With the use of deep learning algorithms, we hope to decipher their feelings in this study. Right now, there's no better place to let your emotions and thoughts out than on social media. Twitter, in particular, can give you a good sense of what's popular and what others are thinking. To better comprehend the wide range of opinions on vaccination, we set out to conduct this study. From December 21 to July 21, the collected tweets were used in this study. The most popular vaccinations that have lately been accessible worldwide were detailed in the tweets. The opinions of individuals on different immunizations were assessed using the VADER tool, which is a Natural Language Processing (NLP) application. After dividing the responses into positive and negative categories, we had a clearer understanding of the whole issue. According to our numbers, 33.96 percent were in favor, 17.55 percent were against, and 48.49 percent were unsure. While doing our study, we also took into account the timestamps of the tweets because attitudes changed over time. To evaluate the prediction models, the RNN-oriented architecture—composed of Bi-LSTM and LSTM—was used. When it came to accuracy, LSTM achieved 90.59% and Bi-LSTM 90.83%. To further confirm our models and conclusions, we used other performance indicators, including accuracy and F1-score, as well as a confusion matrix. This research lends credence to the goal of global coronavirus eradication and sheds light on public sentiment about COVID-19 vaccinations.

This work gives a new theoretical approach to interpretative topic modelling. Instead of words or n-grams, this technique analyses complete phrases. Clustering phrase embeddings and probabilistic sentence evaluations in the text corpus is the suggested technique. The topic model estimates sentence frequency distributions within topics and text-wide topic frequency distributions. Our technique provides explicit topic interpretation since sentences are more significant than words and encompass full grammatical and semantic frameworks. The automated process is also provided. Sentence embeddings are produced using BERT-based context embeddings. Our technique also shows that topic modelling may integrate internal and external information sources for large data processing. Conventional topic modelling uses a single text corpus as the internal knowledge source. The external knowledge source is the BERT, a machine learning model trained on huge textual data, which generates context-dependent phrase embeddings.

Data mining is an application that searches for information in various databases in order to discover new facts and patterns. Conventional data processing technologies and applications are inadequate for handling the massive amounts of data available in SMEs today. Important for SMEs' decision-making is the adoption of efficient and effective data mining tools. This study describes and analyzes seven popular open-source data mining tools: KEEL, KNIME, Orange, RapidMiner, R Project, Tanagra, and WEKA.

Social networking and microblogging have grown, reflecting this generation's ideas. User reviews of products, companies, individuals, forums, films, etc., may gauge people's opinions. Analysts automated positive and negative judgment categorization using algorithms. Its automatic categorization is sentiment analysis. This study uses SVM to classify product reviews' feelings and phrases. The study will critically assess several datasets utilised for this aim. The Support Vector Machine (SVM) learning approach has been trained, verified, and used on many data sets to assess ambiguous sensory polarity. To improve product review categorization, this research employs SVM artificial intelligence to examine several datasets for sentiment and language classification. This work trains, evaluates, and simulates a support vector machine training system to recognise ambiguous emotions or review polarity on numerous datasets. SVM has the highest starting accuracy (89.98%) among classification methods. Adding sentence forms improves accuracy. It concludes that SVM is reliable. Application/Improvements: Support Vector Machine models are evaluated. Finally, the Support Vector Machine (SVM) classifies better and more accurately.

It is well recognised that data preparation is necessary to provide reliable data with which mining tools may derive useful insights. Preprocessing methods need modification when dealing with several sources of continuous data inflow. This study suggests using active rule-based systems, particularly complex event processing (CEP) systems and engines, to aid domain experts in defining and executing pretreatment tasks for data streams. Our method's key innovation is how it enables domain experts to easily manage temporal data by formulating preprocessing methods as event identification rules expressed in a SQL-like language. This concept is implemented in a freely accessible software package that combines a CEP processor with libraries for web-based data mining. In order to test the efficacy of our method, we provide three examples of real-world applications in which CEP rules are used to preprocess data streams by incorporating new temporal information, modifying features, and dealing with missing values. Preprocessing activities may be expressed in a flexible and highlevel fashion using CEP rules, as shown by the experiments, without incurring excessive memory and time overheads. The generated streams of data not only aid in enhancing classification algorithms' predicted accuracy, but also permit simplifying decision models and shortening the time required to learn.

Connected to the web of things, “Big data,” an increasingly popular phrase, describes the massive amounts of data frequently produced by modern healthcare systems. The sheer volume and complexity of the information make it difficult to extract actionable insights for forecasting and policymaking. Big data technologies provide significant infrastructure support to enable critical decision-making and better achieve the objectives of IoT systems. Privacy protection, data integrity, and verification of identity are three tenets that must be strictly upheld in big data for healthcare service administration. To solve these problems, this study suggests a big data ecosystem-enabled, scalable computer system that allows authenticated data access methods for health data analytics enabled by the Internet of Things. A largescale data analytics tracking system, as well as a data storage/access system generated from blockchain, make up the two primary sub-architectures of the architecture that is suggested. In order to analyze, store, and authorize verified use of information collected by IoT-enabled devices, this solution utilizes big data systems as well as blockchain architecture. Data linkability may be avoided, and unauthorized users can have no access to private data by using the zero-knowledge protocol. The results demonstrate the efficacy of our method in resolving healthcare privacy and big data analytics issues.

An obstacle for sentiment-based recommendation systems is the abundance of information available online. Because of flaws in their data-cleaning and analysis processes, existing machine learning (ML) methods fail to provide trustworthy recommendations. To get over this issue, this research proposed a recommender system that uses Twitter reviews and is based on Lite deep learning (LightDL). The information retrieved from Twitter is cleaned using data-cleaning processes. After that, the LightDL algorithm is trained to use this preprocessed data to identify the relevant characteristics (hashtags, unigrams, bigrams, etc.) in each dataset. We have learned four types of information from this: semantic, grammatical, figurative, and aspects based on tweets. Finally, the data is rated as positive, negative, or neutral based on the learned attributes. A battery of tests is run in MATLAB to evaluate the model's performance on many metrics, including recall, accuracy, precision, f-measure, and error rate. Specifically, the proposed LightDL model outperforms all other models across all metrics and achieves 95% accuracy on the Twitter dataset.

Thanks to the rapid development of remote sensing technologies, we now have greater availability of remote sensing data than in the past. Big data has arrived. Hyperspectral features, high geographical resolution, and high temporal resolution are characteristics of Big Data that are present in data acquired by remote sensing. By combining geographical feature data with remote sensing data, this study compares changes in water cover and vegetation and provides a feature-supporting, useful, and efficient data cube for use in time-series analytics. This research provides a spatiallyfocused definition of the SRSDC, or remote sensing data cube. This information cube provides an efficient, adaptable, and scalable method for analyzing large volumes of RS data using spatial attributes. It provides an overview of the SRSDC's organizational structure. For query operations, the SRSDC offers feature translation, and for storing and managing vector feature data, it offers spatial feature repositories. The article details the process of creating a distributed execution engine and a feature data cube for the SRSDC. The manufacturing process and long-term remote sensing evaluation are used as examples to assess a feature data cube and a distributed execution engine. To the pinnacle of the information economy now stands big data, a fresh strategic asset for humanity. Methods for discovering new information rely on data analysis, supervised learning, unsupervised learning, and combinations of the two.

Here, we provide a novel approach to simulating and reconstructing annual surface mass balance (SMB) data across glaciers using a deep artificial neural network (ANN; i.e., deep learning). This method is currently included in the SMB component of an open-source regional glacier development model. Instead of using physical processes, which are becoming more common in traditional glacier models, we use data science to build a parameterized model. A annual glacier-wide SMB may be modelled using topoclimate data with deep learning or Lasso (least absolute shrinkage and selection operator; regularized multilinear regression), and the glacier's form can be updated using a parameterization specific to the glacier. Our nonlinear deep learning SMB approach is tested and validated on a dataset of 32 French Alpine glaciers, compared with other conventional linear statistical methods. Compared with linear methods, deep learning outperforms them with a root-mean-squared error (RMSE) of 0.51 m w.e. and an estimated R2 of 0.77. This is due to the fact that deep learning can explain up to 108% more variation and up to 47% more space. Nearly one-third of deep learning's observed nonlinear behavior is attributed to the temporal dimension. The original measurements of ice thickness are the source of significant uncertainties in the development of glacier geometry. These results provide evidence that deep learning may be a powerful nonlinear tool for glacier modeling, enabling us to reconstruct or simulate SMB time-series data for specific glaciers within a given area to predict past and future environments.

Present on virtually every online service today, recommender systems have quickly become an integral part of people's everyday digital lives. Modern deep learning-based models can only perform at their peak when fed massive amounts of data. Multiple domains, including Amazon items, restaurants, and breweries, have offered several datasets meeting this condition. The hotel industry has seen relatively few advancements and databases, with even the largest review dataset being in the hundreds of thousands rather than millions. Traditional collaborative filtering methods are also inapplicable to the hotel domain due to its higher data sparsity compared to standard recommendation datasets. In this research, we present HotelRec, a TripAdvisor-derived, massively scaled hotel recommendation dataset comprising 50 million reviews. To the best of our knowledge, HotelRec is the largest recommendation dataset in a single domain and includes textual reviews (50M vs 22M) in the hotel domain (50M versus 0.9M).

Many businesses, including those in the banking sector, have begun offering their services online in response to the meteoric rise in internet use. Worldwide, financial fraud is on the rise, leading to enormous losses. From now on, threats like unlawful transactions and unusual assaults may be actively detected by developing sophisticated financial fraud detection systems. Data mining and ML have become more important tools for addressing these problems in recent years. However, in order to speed up processing, analyze massive amounts of data, and detect new attack patterns, these systems need a lot of improvements. According to this study, a Deep Convolution Neural Network (DCNN) can be used to detect financial fraud utilizing smart algorithms. When dealing with large datasets, this approach enhances detection accuracy. The performance of the proposed model is assessed by comparing it to other models that use auto-encoder methods, deep learning, and current machine learning models on a revised credit card fraud database. The results of the experiments demonstrate that the proposed model attained a 99% detection accuracy in only 45 seconds.

Researchers in the field of information systems often talk about human capital. Organizations need trained personnel to create and implement IT objects. This is particularly true when dealing with sophisticated technologies like AI. Natural language processing (NLP) and computer vision (CV) are two important subfields of AI. From a business perspective, this article compares and contrasts the skills required of CV and NLP experts. To achieve this goal, we used a text mining-based analytical pipeline to analyze AI job ads. Named entity recognition and word vectors were used to assess actual job ads from both sub-disciplines that were scraped from a big multinational online employment portal. We suggested an improved machine learning method, the Artificial Neural Network (ANN), for the last skill analysis. The requisite skill sets for the two job descriptions may be different. It calls for individualized thought as there is no one-size-fits-all profile of an AI expert.

The public's swift adoption of wearable sensors and their rapid worldwide spread, together with the rapid advancement of the Internet of Things (IoT), have resulted in most companies and organizations being overwhelmed with massive amounts of data these days. Finding ways to use the data deluge could provide companies with an edge in the market, as data can enhance several aspects of a company. There has been some monetization in the industry across verticals in terms of stacking connected devices with various SaaS choices, such as subscription plans or smart device insights. A significant advantage in today's cutthroat digital marketplace will go to those who can properly monetize data rather than just hoard it, says the “machine economy” sprouting up in this sector. Innovations in big data, analytics, and AI have created new opportunities for competitiveness, with data strategically employed as an asset that can generate new revenue streams due to its ever-changing nature. Because of this surge, a plethora of new resources—including systems, platforms, tools, and markets—have emerged to help organizations make effective use of data. Actually, the goal of new business models is to shift the balance of power away from data-harvesting companies and toward customers. The selling of user data to companies like data analytics corporations is being promoted by startups and NGOs. Data monetization encompasses more than just data sales. It is also possible to implement measures that enhance data. Companies may generate money from data in three ways: 1) by improving internal processes or making more informed business decisions; 2) by integrating data into their main offerings; or 3) by selling data to current or potential customers. This article will address the key aspects of Internet of Things data monetization and the challenges that accompany it, with a focus on the healthcare industry. The topics covered will include data management, scalability, regulations, interoperability, security, and privacy. Not only that, but a detailed reference architecture for the healthcare data economy and a case study on the detection and prediction of cardiac abnormalities using privacy-preserving machine learning (PPML) and multiparty computing (MPC) methods are also provided.

Providing suggestions for integrated, cutting-edge, and efficient tools, methodologies, and technologies for accessing and processing increasingly growing volumes of data in diverse forms is a major problem in clinical data analysis and knowledge discovery. Personalizing a patient's care is a challenging task that requires the doctor to sift through and make sense of massive volumes of data. The scientific community behind precision medicine might benefit greatly from a unified system that facilitates data discovery, integration, preprocessing, model construction, storage, analysis, and visualization. The software package provides researchers with a simple, quick, and adaptable method for processing data, with the ultimate goal of enabling intelligent management, analysis, and visualization of massive genomic data. Services, data sets, and databases are at their disposal, or they can supply their own information for processing.

Tourism destinations and their online and social media information have made choosing and visiting them difficult. Tourists find tourism suggestion systems attractive, but designers must be able to deliver personalised services. This study proposes a personalised tourist recommendation system that extracts user preferences. For this, tourist social network user reviews are a rich resource of preferences. To identify visitor preferences, remarks are preprocessed, semantically grouped, and sentimentally analysed. The characteristics of attractions are extracted from all user evaluations. Finally, the proposed suggestion system semantically matches user preferences with attraction attributes to suggest the most relevant attractions. The technology also filters undesirable goods and improves recommendations based on time, location, and weather. The Python-based recommendation algorithm is tested using TripAdvisor data. The suggested system improves the f-measure.

One noticeable development of the expanding internet age is text mining. Accurate exploration of its potential will enable various real-time applications to leverage text mining's unique characteristics. Conversely, social media platforms like Facebook and Twitter provide vast amounts of data that must be processed to derive actionable insights. Therefore, the purpose of this work is to explore the feasibility of using text-mining algorithms to analyze Twitter data. We want to use this knowledge to improve disaster management by applying the suggested technique to datasets pertaining to natural disasters. The three steps that make up the suggested approach are as follows. The subject of the database is first determined using a Latent Dirichlet Allocation (LDA) model. An SA is then suggested as a second step. The themes found in the sample are categorized into three feelings using this SA, which is applied to the LDA findings. We suggested using an ANN model for sentiment analysis. The third step is to apply data-mining algorithms to the themes in each sentiment using textual analysis (TA). At last, we find out how the supplied tweet feels. Anybody looking to enhance business intelligence analysis (BIA) procedures across industries may benefit from the suggested methodology's significant improvements in data text mining, including correctness, reliability, and the discovery of new insights.

The primary use of forecasting short-term loads is in control centres, where it is used to investigate shifting consumer load patterns and estimate the value of the load at a future time. It's a crucial piece of equipment for building a smart grid. Several dimensions of influence impact the load parameters. In this research, we present a Residual Neural Network (ResNet)/Long Short-Term Memory (LSTM) hybrid model for load forecasting, which can better take advantage of the time series properties of load data and lead to more reliable predictions. Before feeding the data into the ResNet network for the extraction of features, it is first rebuilt using numerous feature parameters. The second step in short-term load forecasting using LSTM is to feed the extractedfeaturevector into the network. Finally, the technique is compared to other models using a real example, demonstrating that the proposed combination method has better prediction accuracy and confirming the practicality and superiority of input featureextractionparameters. Additionally, this study conducts studies in weather prediction based on a variety of elements and characteristics

Thanks to advancements in educational technology, new approaches to controlling and understanding the teaching-learning processes have arisen, such as instructional analytics (TA). Instructional analytics is a new yet exciting approach to learning that has the potential to dramatically improve teaching quality and student outcomes by helping educational institutions better use the massive volumes of information generated from student data. This study offered a framework for educational process data mining with artificial intelligence (EPDM+ML) that uses student evaluation of teaching (SET) data to contextually evaluate teachers' performances and provide recommendations. The EPDM+ML model was created and implemented using a mix of text mining and artificial intelligence (AI) technology. It is based on descriptive decision theory, which examines why learners tend to make choices through quantitative analysis of textual data. To this end, the study considers instructors' gender while analyzing the pedagogical elements that influence students' suggestions for their professors and the effect of students' subjective experiences on their assessments of instructors. Predicting a student's potential recommendation for a teacher is conceivable with SET data such as students' gender, average mood, and emotional valence. In actuality, we mined text for the different sentiments and emotions (comment intensities) that students conveyed in the SET. Then, we used a Kruskal-Wallis test to determine which elements were the most important by analyzing the covariance using the quantitative data (average emotions and emotional valence). Furthermore, we examined the gender conceptions to see whether there were any differences in the suggestions students made for instructors. Taking into account both the mood and the sentiments expressed in the comments, we find that most (n=85,378) were positive in tone, while a sizeable portion was neutral. Female students were more likely than male students to propose that teachers take into account students' emotions in statements with an emotional valence (11.8%) and opinions (23.6% positive and negative) (p=.000). Men, on the other hand, exhibit borderline behavior in terms of sentiment (p=.077) and emotions (p=.056). The EPDM+ML model was also validated using the k-fold cross-validation method, with 63.6% of the optimal k-values observed, and it proved to be an efficient and effective predictor of students' recommendation scores for teachers, with high and acceptable values of precision (1.00), recall (1.00), specifcity (1.00), accuracy (1.00), F1-score (1.00), and zero error-rate (0.00). Theoretically, we find that the suggested approach (EPDM+ML) is helpful for effective examination of SET and its consequences in the field of education. There is a growing need to improve teaching and learning processes and students' learning experiences in an ever-evolving educational ecosystem. However, this can be done by identifying the most important factors that students use to evaluate and recommend teachers.

Data mining has become an increasingly significant method for conducting data analysis as a result of the rapid increase in databases used by a large number of contemporary businesses. The community of people who study operations research has made major contributions to this discipline, particularly by formulating and solving a large number of data mining problems as optimization problems. Additionally, data mining techniques may be used to address a number of applications in operations research. The purpose of this study is to offer an overview of the relationship between operations research and data mining. The basic objectives of the study are to highlight the spectrum of interactions between the two areas, present specific instances of significant research effort, and provide extensive references to additional significant work in the area. The purpose of this study is to examine not only the many optimization techniques that may be used for data mining, but also the process of data mining itself, as well as the ways in which operations research techniques can be utilized at almost every stage of this process. The report also identifies many potentially fruitful avenues for further investigation throughout its body. In the last part of the study, many applications connected to the administration of electronic services, including customer relationship management and customization, are investigated.

While data mining finds extensive usage in the scientific and technical communities, it also finds beneficial uses in the financial and marketing spheres, where it helps address specific problems. By analyzing the actual situation, a data miningbased decision support system improves the organization's performance. Due to factors such as competition, costs, tax pressures, and other factors, every firm experiences economic volatility. A more competitive environment is dragged into the organization via privatization, globalization, and liberalization. To stay ahead of the competition and accomplish your goals, you need a well-planned and implemented marketing strategy. By using an effective data mining strategy, marketing decision support systems help organizations reduce the responsibilities associated with analysis and strategic planning. This study offered a data mining-based decision support system for estimating a company's marketing plans by combining decision trees and artificial neural networks: data mining, decision trees, artificial neural networks, and marketing decision support systems.

Over the last several years, research into recommender systems (RS) has grown to include a wide variety of artificial intelligence (AI) techniques, from more basic ones like Matrix Factorization (MF) to more complex ones such as Deep Neural Networks (DNN). Since they only examine a linear combination of consumer and item vectors, traditional CF recommendation algorithms like MF have little room for improvement in learning. Neuronal collaborative filtering (NCF) is a hybrid of deep neural networks (DNNs) and collaborative filtering (CF), enabling it to learn non-linear connections. But CF methods still have issues with cold beginnings and sparse data. This research presents a new hybrid RS, Neural Matrix Factorizing++ (NeuMF++), that can handle cold starts, boost recommendation accuracy, and address data sparsity. To improve NeuMF, we introduce NeuMF++, which uses Stacked Denoising Autoencoders (SDAEs) to better represent latent features. Merging GMF++ and MLP++ might yield NeuMF++. Mixed with Multilayer Perceptrons (MLP) and Generalized Matrix Factorization (GMF), NeuMF is a robust NCF architecture. By combining the linearity of GMF with the nonlinearity of MLP, NeuMF achieves stateof- the-art results. Simultaneously, the original GMF as well as MLP were enhanced with GMF++ and MLP++, respectively, by successfully including latent representations. Obtaining a latent representation from the SDAEs' hidden space significantly enhances NeuMF++'s learning power. This representation allows it to learn item and user attributes correctly. If feature extractions are shared across GMF++ and MLP++, however, NeuMF++'s performance could take a hit. As a result, its speed and flexibility are greatly enhanced when GMF++ and MLP++ are allowed to learn features separately. Using a real-world dataset, NeuMF++ achieved an experimental root-mean-square error of 0.8681, demonstrating its excellent performance. Additional data, such as text or photos, can be added to NeuMF++ in later development. NeuMF++ allows incorporating several neural network building blocks to create a more robust recommendation model.

The goal of this research is to examine how e-commerce and web-based businesses might benefit from the use of Big Data Analytics in managerial decisionmaking. The data used in this analysis comes from a single e-commerce website's database. User interactions with the website, such as page views, product additions, and online purchases, would be recorded. Association Rule Mining Algorithm (APRIORI), K-Means Clustering, and Pearson's Correlation Coefficient are only a few of the algorithms that will be utilized to evaluate the dataset. The information will be analyzed and used to provide insights into users' interactions with the website, enabling the identification of patterns that may inform future actions. For instance, which product receives the most attention and sales, how many pages are viewed before a purchase is made, what percentage of customers buy the product again, and so on. This study would also determine whether the company's current Big Data implementation can be enhanced and whether doing so would be a clever use of resources.

In recent years, the quantity of textual data carried via online digital files, emails, databases, and other formats has increased at an exponential rate due to the fast growth of Internet technology. Ontology and other contemporary models of knowledge representation rely heavily on attributes, which are formal descriptions of any realworld thing. Attributes may be either qualitative or quantitative. There is a wealth of literature attesting to the extensive research on mining non-qualitative attributes (such as part-of relations) from text and the web, but very little on mining qualitative attributes (such as size, color, taste, etc.). It became increasingly difficult for traditional intelligence data-processing technologies to meet the demands of these tasks. The accuracy of segmenting words in intelligence texts was significantly enhanced by our suggested intelligence dictionary-based word segmentation technique and the construction of an intelligence-specific dictionary.

Recently developed e-learning recommendation systems aid students and instructors. These conditions need specialist online learning for students and educators. Split-and-conquer clustering based on a strategy creates a clever recommender for learners' requirements, preferences, and abilities. An automatic self-learning recommender evaluates learners using a separate-and-conquer analysis. Cluster-based linear pattern mining may reveal learner functions. The application suggests properly utilising common pattern scores. This technique was evaluated across different learner groups and datasets to provide relevant learning tasks based on learning style, interest classification, and competence. The recommended cluster-based recommender improved trial recommendation performance, helping learners complete more courses than the no-recommender group. Students scored the suggestion tool over 65% in all areas. The recommender increased metric values for bigger learners in the simulation. Statistics showed huge disparities in student measures exceeding 1000. A computational framework using |L| (recommendation list size) and student qualities explained the disparities. The kids liked the recommender's speed and accuracy.

By analysing the geographical, chronological, and semantic components of geographic data, it is possible to reconstruct users' real route itineraries and get insights into their preferences and behavior. To analyse tourist traffic patterns at A-level scenic spots in Jiangsu and Zhejiang across time and space, this research collects and preprocesses Weibo check-in data for these locations. The author used a temporal perspective, examining how check-in data fluctuated between 2016 and 2018 and how it differed on weekends, holidays, and weekdays. The acquired data were subjected to a spatial kernel density analysis, which revealed the most active regions. Lastly, the vacation travel mode and characteristics were identified through an examination of spatial and locational flows and their orientations. The results of this study provide the groundwork for the growth of tourism.

New avenues for intelligence collection have been made possible by the advent and proliferation of the internet from the perspective of military intelligence activities, but the ever-increasing amount of information stored on it has put a strain on traditional methods of intelligence analysis. The current state of intelligence analysis must be corrected by expanding and improving military intelligence analysis practices. The control node to intelligence agency analysis must be abandoned if the military is to obtain the depth of intelligence knowledge it requires, if the technology of data mining is to be introduced into the analysis conducted by intelligence agencies, and if a data mining model is to be used to construct the network of military and intelligence analysis. The model's algorithm for intelligence analysis, which is based on semantic analysis rather than traditional types of association analysis, may improve the efficiency and accuracy of military intelligence gathering as a result.

Finding relevant information online has gotten increasingly difficult as the amount of data available on the internet has grown exponentially. In high-data-density, complex-domain settings, the recommendation system may be a big aid to users in making decisions. In the recommendation system, several approaches have been presented and collaborative filtering is a common practice. The cold-start problem is one of the remaining issues with collaborative filtering approaches. To address this issue, we offer a movie recommendation system that uses social network analysis and collaborative filtering. We use user preferences like age, gender, and profession to generate a connection matrix, and then that matrix is used to cluster people using community identification based on edge betweenness centrality. The suggested system then proposes movies to new members based on the preferences of the existing users in the group. Utilizing MAE, we demonstrate the superiority of the suggested technique.

A company's postmanagement entails assessing, selecting, and employing individuals who logically and scientifically match the post requirements as well as are a good fit for the company's growth into their appropriate positions. The data that most accurately portrays the market need for data analysis skills is the job recruiting information published on the employment website. Nevertheless, this work uses text mining to analyze the information in online recruiting materials. This research serves as a valuable resource for analyzing job search data, since most of the material is presented in textual form. Workplace matching is the focus of this research, which draws on text mining as well as multi-criteria decision-making. With 66.45% of the candidates coming from the Internet business, technical positions within text mining are mostly concentrated there. Technical data analysis is not as in demand in other sectors; in fact, demand is below 15% across the board. Education continues to dominate as a demand supplier, accounting for 30.53% of the market, despite the dispersed and expansive nature of the business data analysis industry. But at 24.28% and 22.34%, real estate and media are in close proximity. The service sector will have a greater need for business data analysts than technical ones, in addition to the aforementioned three sectors. During the data-gathering step, the recruiting website's data is analyzed to extract job details, including pay, qualifications, required work experience, and the total number of recruits. ¬The data gathering step is carried out by this paper, which crawls the recruiting website for job information. Along with further details about the post, such as the number of recruits, the position's remuneration, the qualifications required, and the required work experience, this study may aid in analyzing recruitment data published by the website and swiftly reflect data on the market need for data analysis abilities.

One new piece of software that's making waves is the recommendation system, or RS. It uses knowledge-discovery methods to infer the user's goals and interests. It is becoming increasingly difficult to decipher the user's intent and activate the suggestion appropriately as data volume grows exponentially. This research proposes a new Deep Knowledge Grid (DKG) for RS construction and huge data analysis. The DKG uses a Deep Convolutional Neural Network (DCNN). This DKG explicitly models the KG's end-to-end high-order connectivities. Using an attention approach, it iteratively propagates embedded information from a node's neighbors—persons, objects, or characteristics—while determining the relative significance of those neighbors. From a conceptual standpoint, our DKG outperforms state-of-the-art KG-based recommendation methods that either use regularization to model high-order relations or extract paths to exploit them explicitly. Results on publicly available benchmarks show that KGAT outperforms state-of-the-art methods like RippleNet with Neural FM. The attention mechanism's benefits for interpretability and the efficacy of embedding propagation in high-order relation modeling have been demonstrated in subsequent studies.

The study's overarching goal is to provide a methodology for employing domain ontologies to get health-related data from the Internet. The Internet has made patient-doctor dialogues commonplace. Solutions like HealthTap and AskTheDoctors make it possible for people to ask physicians questions about their health. The knowledge gap between experts and laypeople, as well as the language barrier, mean that most people seeking medical treatment online still struggle to ask effective questions. It is difficult to glean information from these non-expert accounts since they usually do not use technical language. The underlying applications, such as information retrieval, become less efficient as a result of this. In this paper, we provide an ontology-driven method for employing a meta-model to glean insights from such incomplete descriptions. When people seeking medical care don't share the same terminology as their doctors, a meta-model may help. To access the extensive medical vocabulary, the meta-model is mapped to SNOMED CT. To increase the coverage of layman's words during information extraction, it is mapped with WordNet. A data extraction prototype that relies on syntactical similarities is put into place in order to evaluate the approach's potential. Using the gold standard corpora established in Task 1 of ShARe CLEF 2013, the method demonstrated encouraging results in identifying medical ideas in actual medical papers, with an F-score of 0.79.

Many social media forums (blogs, Facebook discussions, etc.) have emerged as a result of the proliferation of the Internet, and users are able to access these sites from their mobile devices. These sites serve as venues for users to express and communicate their thoughts and experiences on current events and other global concerns. Before making a purchase, many customers check out review websites. In order to find useful product reviews and user experiences, people often explore popular websites. It is simple for us to gather massive volumes of product data, both structured and unstructured, and then analyze it to extract the specific product details you need. Researchers are increasingly turning to sentiment analysis—also called opinion exploration or opinion mining—to gather and assess customer sentiment and viewpoints. Emotional text mining is introduced in this paper, with a focus on reducing data dimensions via principal component analysis and singular value decomposition, and on improving accuracy and reducing implementation time through the use of an additional feature strategy. The research begins by proposing a method for preparing data to classify sentiment. Secondly, it leverages new features to improve emotion categorization precision. Singular value decomposition and principal component analysis are used to reduce data in the third step. Finally, it creates five modules with different features and analyzes their performance, both with and without stemming. When compared to the alternatives, the experimental results show that the proposed method is the most accurate and has the potential to drastically reduce implementation time.

A user's interests and goals can be deduced with the use of a Recommendation System (RS), a new form of technology that employs knowledge discovery methods. User intent analysis and corresponding suggestion triggering become increasingly challenging in the face of exponential data growth. This work proposes a unique Deep Knowledge Graph (DKG) to do the necessary data analysis and construct the RS. DKG employs a Deep Convolutional Neural Network (DCNN). Our proposed DKG explicitly models the KG's end-to-end high-order connectivities. It recursively propagates the embeddings from a node's neighbors, which can be people, things, or traits, using an attention mechanism to assess the relative significance of the neighbours. In terms of theory, our DKG outperforms existing KG-based recommendation methods since it does not rely on regularization or an explicit representation of high-order relations. Empirical results on public benchmarks show that KGAT outperforms state-of-the-art methods like RippleNet and Neural FM. The benefits of the attention mechanism for interpretability, as well as the efficacy of embedding propagation for high-order relation modeling, have been shown in subsequent studies.

As the number and variety of data kept growing, conventional data processing technologies could not keep up with the demands of the big data setting. Because of this, new processing instruments tailored to specific needs have emerged. However, picking the ideal instrument could be difficult due to the abundance of options. To aid its mapping to different needs, this page provides a technical overview of the most common analytics tools, along with a literature review on the topic. Furthermore, we highlight the importance of choosing the correct tool for various uses, especially in smart city settings.

Creating a system to track the health of a machine is no easy feat. For this to work, a sufficiently large dataset of machine operation signals, operation-related background data, and diagnosis-related anecdotes is required. Nowadays, it is not difficult to address the two issues mentioned. Because it relies on non-numerical and imprecise information, the diagnostic experience is the most difficult to convey. However, in order to build a reliable monitoring system, the processing of the collected data is crucial. A system devoted to suggesting processing methods for condition monitoring is presented in this article's framework. A database and internal modules built using fuzzy logic are part of it. The technique proposes processing algorithms based on user-provided contextual information. The results of testing the suggested agent on two separate parallel gearboxes are also presented in this article. Processing algorithms using assigned model types are the outputs of the system. The findings demonstrate that the algorithms suggested by the system outperform those chosen at random. Accuracy gains ranging from 5% to 14.5% are achievable, according to the findings.