This unit discusses the practical applications of text mining as well as some text mining tools. These include different categories and classifications of text miners, with their examples and uses. The uses of some text miner tools for the clustering and classifications of text are also discussed.
Author(s)
Details
Adebola
K. Ojo
Department of Computer Science, University of Ibadan, Ibadan,
Nigeria.
Please see the book here:- https://doi.org/10.9734/bpi/mono/978-81-972870-5-3/CH12
In this section, we will provide you with a brief overview
of various text mining tasks which are commonly used for analyzing large
volumes of unstructured textual data. Text classification, grouping, entity
extraction, fine-grained taxonomies, sentiment analysis, document
summarization, and entity relation modeling are some of these activities. Text
categorization involves organizing text into predefined categories based on its
content. Clustering is the process of grouping similar documents together based
on their intrinsic characteristics. Entity extraction involves identifying and
extracting key elements such as people, places, and organizations from text.
Granular taxonomies are hierarchical structures used for organizing textual
data. Determining the general sentiment of a text, whether it be favorable,
negative, or neutral, is the goal of sentiment analysis. Making a summary of a
longer material is called document summarizing. Lastly, the act of determining
the connections between various named entities that are stated in a text is
known as entity relation modeling.
Author(s) Details
Adebola K. Ojo
Department of Computer Science, University of Ibadan, Ibadan, Nigeria.
Please see the book here:- https://doi.org/10.9734/bpi/mono/978-81-972870-5-3/CH8
Technological development and improvements have accelerated
the Sensor node design in terms of low power consumption, and low cost and also
exhibit multifunctional and untethered communication in short distances. The
capabilities of the sensor nodes like sensing, data collecting and processing,
transferring, etc have assisted and greatly transformed the design, advancement
and deployment strategies of Wireless Sensor Networks, where all the nodes
collectively collaborate for the target application. The sensor nodes with the
help of the sensors attached to them monitor various parameters and transmit
the acquired data via wireless medium to a distant node which is called a sink
or base station. The main function of the sensor network is to gather sensor
data from the area/region of event occurrence and transmit it to the sink node.
The nodes in the sensor network work in a collective manner, thus making it
different from the ad-hoc networks.
A node is able to sense an event within a specified range.
The strength of the event signal is also deterministic of which sensor nodes
can sense the event. To ensure that no event is missed being reported, sensors
in a WSN are densely deployed. The sensor node’s position is generally not
fixed; the nodes might be randomly distributed around the phenomenon. So there
is a need for protocols to have self-organizing capabilities, also the nodes
must work in a co-operating manner to achieve effective information transfer.
Sensor nodes have processing capabilities of their own; they locally process
the data and then transfer it. The sensor nodes deployed in WSN have limited
computational capacities, memory and power compared to ad-hoc networks. Sensor
nodes in WSN may not have unique global identification as it can cause overhead
in communication, also owing to a high density of sensor nodes, global
identification can be challenging. In WSN there exist challenges in areas viz.,
scalability, cost, power, self-organization, interoperability, data
compression, etc.
Sensor nodes aggregate the sensed data before transmitting
it. Typically aggregation is performed at representative nodes or in the
gateway nodes while the data packets are in transit to sink. Data aggregation
protocols aim to remove redundant data, thus enhancing the lifetime of the
sensor network. In a typical WSN, data is transmitted in a multi-hop fashion;
nodes send data to their neighbors which are nearer to the sink. Nodes that are
closely placed are likely to sense the same data and thus cause redundancy.
Based on the application requirement, sensors either
transmit the data whenever an event is detected or periodically. These WSN
characteristics and varied application areas motivate a sensor MAC which is
operationally different from existing traditional MACs. Sensor networks MAC
have node self-organization and energy conservation as their primary goal. The
wireless channel access plays an important role in forwarding the data frames
to the sink. Many MAC protocols are proposed for efficient channel access. WSN
MAC techniques control and coordinate the radio component so that the network
is energy efficient thereby improving lifetime considerably.
There are problems associated with the existing WSN systems.
The existing schemes for data aggregation mechanisms are weakly related to data
correlation and data redundancy, leading to poor data quality. The majority of
the research work emphasizes that the clusterhead be elected based on node
energy, which may not prove fruitful for data correlation-based aggregation.
Data aggregation schemes employed in current techniques fail to perform data
analysis cost-effectively. MAC schemes result in congestion in the nodes
surrounding the base station, which should be eased to achieve better network
performance.
To overcome these challenges we present a framework to
enhance the data quality during the aggregation process. It is a novel and
simple clustering algorithm that performs the selection of the clusterhead
based on the data correlation factor. We also propose a novel hybrid MAC
technique called Improved Funneling MAC for effective resource management. Both
the protocols are implemented in MATLAB and simulation results are presented.
Implementations are compared with the existing schemes and it is found that our
implementations contribute to improved performance.
Author(s) Details
Dr. Anand Gudnavar
Department of Computer Science and Engineering, Jain College of Engineering
and Research, Belagavi, India.
Dr. Prakash Sonwalkar
Department of Computer Science and Engineering (AIML), Jain College of
Engineering and Research, Belagavi, India.
Dr. Keerti Narega
Department of Computer Science and Engineering, Graphic Era Deemed to be
University, Dehradun, India.
Please
see the book here:- https://doi.org/10.9734/bpi/mono/978-93-48006-44-8
Cancer remains one of the most formidable health challenges globally, and India has seen a steady rise in cancer incidence rates over the years. The ability to effectively analyze large-scale cancer datasets is crucial for understanding disease patterns, improving diagnosis, and guiding public health interventions. In this research, advanced data mining techniques were leveraged, specifically classification and clustering, to examine cancer incidence patterns in the aftermath of the Bhopal Gas Tragedy, a catastrophic industrial disaster. Our study focuses on comparing the incidence rates of Tobacco-Related Cancer (TCR) and Non-Tobacco-Related Cancer (Non-TCR) in two distinct regions of Bhopal, which were partitioned after the tragedy.
Using over 40 years of data from the Population-Based Cancer
Registry (PBCR) of Bhopal, data mining methodologies were applied to uncover
hidden patterns and correlations within the cancer incidence data. The study
seeks to explore the long-term impact of environmental exposure on cancer
prevalence, particularly the difference in cancer types between the two
regions. By employing the WEKA tool, a well-established platform for machine
learning and data mining, cancer cases were systematically classified and
significant insights were extracted from the data.
Our findings reveal notable differences in cancer incidence
between the two regions, offering insights into how environmental factors,
lifestyle choices, and socio-economic conditions may influence cancer
development. The study highlights the value of data-driven approaches in health
care, particularly as a decision support system for medical analysts. These
insights not only contribute to the understanding of cancer epidemiology in
Bhopal but also underscore the importance of continuous health monitoring in
populations affected by industrial disasters. Furthermore, the methodology
applied in this study serves as a foundation for future research aimed at
improving cancer prevention, early detection, and personalized treatment
strategies in similar contexts.
Author (s) Details
Sanjeev Gour
Department of Computer Science, Medicaps University, Indore, India.
Rajendra Randa
Department of Computer Science, Medicaps University, Indore, India.
Please see the book here:- https://doi.org/10.9734/bpi/mcscd/v10/3147
Data aggregation represents a fundamental process within
wireless sensor networks, facilitating the transmission of environmental data
to end-users via base stations. Despite its critical role, data aggregation
often receives less attention compared to routing and energy optimization
challenges. This work presents a comprehensive review of existing data
aggregation schemes, with a specific emphasis on correlational-based
approaches. Our analysis reveals a significant gap in research dedicated to
correlational-based data aggregation techniques. Furthermore, existing methods
tend to overlook crucial factors such as data quality, computational
complexity, and appropriate benchmarking. Addressing these unresolved issues is
essential for enhancing the reliability and quality of data aggregation
processes in wireless sensor networks. This chapter outlines the key challenges
and opportunities for future investigations in this domain.
Author(s) Details:
Anand Gudnavar,
Department of CSE, Jain College of Engineering and Research,
Belagavi, Karnataka, India.
Virupaxi Dalal,
Department
of ECE, Jain College of Engineering and Research, Belagavi, Karnataka, India.
Raghavendra Maggavi,
Department of ECE, Jain College of Engineering and Research,
Belagavi, Karnataka, India.
Veeresh Hiremath,
Department of ECE, Jain College of Engineering and Research,
Belagavi, Karnataka, India.
Please see the link here: https://stm.bookpi.org/RUMCS-V4/article/view/14140
This paper focuses on the problem of using machine learning techniques on cancer/tumor prediction. One of the biggest applications of big data and machine learning is in the field of medical domain. Consequently, a health care organization that uses the techniques of machine learning and big data to treat patients see fewer mishaps or gets enough time to deal with them in advance. Classification and prediction of the images are fairly easy task for humans, but it takes more effort for a machine to do the same. Machine learning helps to attain this goal. It automates the task of classifying a large collection of images into different classes by labelling the incoming data and recognizes patterns in it, which is subsequently translated into valuable insights. Furthermore, the prediction of Osteosarcoma case for one of the four classes of tumor namely Non tumor, Non-Viable tumor, viable tumor, Viable: Non-Viable tumor has to be done. The quantitative analysis is done using various machine learning libraries of python. The three classification algorithms used for image analysis are random forest, SVM, and logistic regression. The metrics used for performing perspective analysis are precision, recall and F1 Score. The domain of medical imaging helps providing important information on anatomy and organ function subsequently detecting disease states. Although the characteristics of medical data make its analysis a big challenge notwithstanding that machine learning techniques could make the task easier. The results show that the random forest algorithm has performed best amongst the three classification algorithms when given with less complicated scenario, with prediction accuracy, precision, recall and f1 score of 100%. But the performance of every classification algorithm degrades when provided with the cases of Osteosarcoma which has got more complicated scatter graph. However, the logistic regression retains its performance by predicting tumor cases with 99% accuracy. For future scope, various other machine learning algorithms can be applied to observe their performance on the same set of features extracted from the cancer image data set.
The interconnected system algorithms Self-Organizing Map (SOM) and Affecting animate nerve organs Gas (NG) use unsupervised competing learning. These methods have the fault-finding virtue of maintaining the topological structure of the data, that means that data that are enclose the input distribution are plan to neighboring positions in the network or output. This characteristic form them intriguing to search in terms of dossier clustering. A crucial characteristic resolving vast amounts of data manually maybe challenging and time-consuming. Suitable way, technologies for analyzing and visualizing large multidimensional data sets are necessary. We introduce a order for comparing and visualizing the SOM and NG in this branch. We describe these algorithms first, and then we create a pictorial comparison betwixt them. The protein mass spectrometry dossier clustering is then elucidated using these imagination approaches.
Author(s) Details:
Terje Solsvik Kristensen,
Department
of Informatics, Western Norway University of Applied Sciences, Bergen, Norway
and BIC AS, Norway.
Please see the link here: https://stm.bookpi.org/RATMCS-V7/article/view/12831
This study presents an creative security model for wireless approach in 5G networks, referred to as 5GDoSec. Considering that a concern inside the security of 5G network access refers to Distributed Denial of Service (DOS) attacks accredit its orientation towards the Internet of Things (IoT), a safety model is put forth. This novel model offers a judgment to this predicament, demanding slightest user dossier, user-friendly operation, modernized training and arrangement, as well as modest computational demands and irregular adaptability. The basic target of this model search out identify potential trespassers and malicious actors through the request of Deep Neural Networks coupled with machine intelligence methodologies. The methodology trails an evolutionary process established prototypes where an alone security model is buxom through data analysis. This approach influences access dossier collected from a specific effort point that aggregates, profiles, and classification authorized network users. The aim search out discern, established access metrics and alive durations, those individuals that ability pose a security risk. The adaptable type of the 5GDoSec model has been tentatively demonstrated and stands as a dependable method of accurately classification hazardous users. Empirical confirmation, gauged through the DaviesBouldin index, underlines its superiority over alternative methods such as Kmeans and Linkage.
Author(s) Details:
Sebastian Camilo Vanegas Ayala,
Faculty
of Engineering, Universidad Distrital Francisco José de Caldas, Intelligent
Internet Research Group, Bogotá D.C., Colombia.
Octavio
José Salcedo Parra,
Faculty
of Engineering, Universidad Distrital Francisco José de Caldas, Intelligent
Internet Research Group, Bogotá D.C., Colombia and Department of Systems and
Industrial Engineering, Faculty of Engineering, Universidad Nacional de
Colombia, Bogotá D.C., Colombia.
Brayan Leonardo Sierra Forero,
Faculty of Engineering, Universidad Distrital Francisco José de
Caldas, Intelligent Internet Research Group, Bogotá D.C., Colombia.
Please see the link here: https://stm.bookpi.org/RADER-V8/article/view/11955
The present study proposed to improve the healing supply distribution routing. Repetition, delay, and cross-precinct sequence of childbirth cause extra transportation cost. The distribution killing planning in home-transmittal industry plays significant belongings to the customer delight and the total logistics costs. Therefore, it is important to have a well-devised distribution whole for the home-delivery company to deal with its manufacturing features and practical needs and to assuage the marketing demand necessities. We proposed a clustering approach to increase the efficiency. We group all the points into various clusters, then routing the course. By allocating to the encircling individual points separately, a distribution centre inside the apiary-shaped establishment can significantly decrease the distance and reduce the time. The aim of logistics search out provide high-quality duty that meets the needs of end users. Therefore, destroying optimization with grouping approach can help to improve the act. The distribution distance can be decreased from 52 km to 29.4 km, i.e., 43.46 . The cost bettering in terms of time dropped off from 130 minutes to 61 proceedings, i.e., 53.08 . For problems to a degree errors in the manual effort of data and repeated renewal, personnel instruction and training can be heartened in the future, standard SOP maybe implemented to reduce mistakes, and replenishment can again move towards better planning of supporting measures.
Author(s) Details:
Chuang-Chun Chiou,
Department
of Industrial Engineering and Enterprise Information, Tunghai University No.
1727, Sec. 4, Taiwan Blvd., Xitun Dist., Taichung City 407, Taiwan.
Wen-Hui
Ouyang,
Department
of Industrial Engineering and Enterprise Information, Tunghai University No.
1727, Sec. 4, Taiwan Blvd., Xitun Dist., Taichung City 407, Taiwan.
Tzu-Yang Lin,
Department of Industrial Engineering and Enterprise Information,
Tunghai University No. 1727, Sec. 4, Taiwan Blvd., Xitun Dist., Taichung City
407, Taiwan.
Please see the link here: https://stm.bookpi.org/RADER-V1/article/view/10089
Recent years have seen tremendous advancements in multimedia applications due to the growth of portable digital devices. The development of multimedia applications has coincided with an increase in the need for 3D technologies. Poor image quality and the growing temporal complexity are the two biggest issues with two-dimensional to three-dimensional image conversion. By addressing spectral and spatial challenges, pre-processing techniques on noisy, blurry texture images with low resolution can enhance visual perception. The study describes the issues with the augmentation schemes. In addition to a number of interpolation methods, the paper provides augmentation methods that make use of Discrete Wavelets and Stationery Wavelet Transforms as preprocessing tools. For 3D conversion, an Improved Simple Linear Iterative Clustering (ISLIC) method with Statistical Region Merging (SRM) is suggested.
A crucial function for the size system is played in the sector of producing clothes. This study work aims to present a robust methodology that might be applied for creating sizing systems using data mining methods and Indian anthropometric data. A new method of two-stage data mining has been used to identify the shirt size type of Indian men. There were two phases to this strategy. Cluster analysis would be the first stage, and cases would then be sorted according to the cluster results in order to extract the most important classification algorithms for shirt size as the second stage. Based on the chest size discovered during the data mining process, a sizing system was created for the age range of 25 to 66 years for Indian men. The specification of the size label is a crucial component of the sizing system since it helps buyers easily locate the appropriate clothing size for further consideration.
Background: K-Means is a popular partition-based clustering technique that divides an input dataset into a collection of groups. K-Means is a popular technique because of its simplicity and quickness in grouping large amounts of data. Because of the massive quantity of electronic data generated, data clustering techniques have had to be modified in order to process it. When dealing with massive data, we may improve the performance of K-Means by using a distributed computing environment. The MapReduce paradigm may be used with K-Means to create a distributed computing environment and improve time efficiency. The number of clusters, 'K,' must be pre-specified as an input to the algorithm in order for K-Means to work. This advance calculation and definition of cluster number generally leads to "forced" clustering of data in the lack of sufficient domain expertise, or for a new and unknown dataset, and correct clustering does not emerge.
Method: In this
research, we provide a novel K-Means-based method that accepts just a numerical
dataset as input and produces the required number of clusters on the fly using
the MapReduce programming style.
In this chapter, we propose Merge-Filter Representative-based Clustering (Merge-Filter-RC), a general domain-independent method for finding near-duplicate records within and across different data sources. Following that, we develop three nearly optimal classes of algorithms known as All-Three algorithms: constant threshold (CT), variable threshold (VT), and function threshold (FT). Merge-Filter-RC and All-Three form the backbone of this effort. Merge-Filter-RC recursively divides and merges near-duplicates into hierarchical clusters with prototype representatives to dis- till locally and globally near-duplicates. Each cluster is distinguished by one or more dynamically refined representatives. To limit the number of pairwise comparisons and hence the search space, representatives are utilised for further similarity comparisons. Furthermore, we describe the findings of the comparisons as "very similar," "similar," and "not comparable." We augment All-Three methods with a more complete reexamination of the original well-tuned characteristics of Monge-(ME) Elkan's foundational work, which we avoided by employing an affine variation of Smith-(SW) Waterman's similarity measure. We conducted multiple trials and comprehensive research on real-world benchmarks as well as synthetically created data sets to demonstrate that All-Three algorithms based on the Merge-Filter-RC technique surpass Monge-algorithmic Elkan's in terms of accuracy in finding near-duplicates. Furthermore, All-Three methods are as computationally efficient as Monge-technique. Elkan's.
Medical imaging, network traffic analysis, environmental systems, and other sectors have benefited from the development of diverse data mining methods. The environment system is now the most essential topic of concern for individuals in today's society, as it has a daily impact on human lives. ES elements such as earthquakes, soil erosion, deforestation, rising summer temperatures, rain fall density/intensity, flood occurrences, and the most significant is the impact of all of these ES factors on human people and their behaviour, both directly and indirectly. Data mining methods can be used to uncover patterns in data that is widely spread, heterogeneous, sparse, multidimensional, and heterogeneous, such as data from the Environment System. This study provides a brief overview of the key phases, techniques, and processes involved in developing and dealing with ES data, which are critical in the development of a data mining tool for detecting and understanding patterns in environmental system data sets. The data mining techniques used in the design of ES Tool span from processing crude data sets to translating them into patterns for examination.
Author
(S) Details
M. S. Chaudhari
Department
of Computer Science & Engineering, Priyadarshini Bhagwati College of
Engineering, RTM Nagpur University, Nagpur, Maharashtra, India.
N. K. Choudhari
Priyadarshini
Bhagwati College of Engineering, RTM Nagpur University, Nagpur, Maharashtra,
India.
View Book:- https://stm.bookpi.org/RAMRCS-V1/article/view/4339
This research describes a novel way to employing artificial neural networks (ANNs) to improve transmission line protection. The suggested technique feeds four different neural network structures instantaneous voltages and currents on a transmission line during normal and fault conditions. The structures are then expertly merged to provide a system that can more effectively detect and diagnose shunt problems. The report goes into great detail about the design process as well as the many simulations that were run. The accuracy and mean square error (MSE) of the created system are examined, and the findings reveal that this approach is capable of identifying and classifying all probable shunt faults on the 33-kV Nigeria power lines in less than 1ms with a high level of precision. When evaluated under various shunt fault types with varying resistances and distances, the system's performance demonstrates that it can be used to improve distance line protection in 33-kV Nigeria power lines.
Clustering is a frequently used technique for grouping data objects based on similarity features. To generate similarity features, similarity metrics such as Euclidean, cosine, and others are employed. To locate clusters, traditional clustering methods such as k-means and other graph-based algorithms are often utilised. These methods, on the other hand, require user participation in order to determine the number of clusters. Traditional clustering algorithms partition data without first taking into account the number of clusters or cluster tendency. There is a risk of poor clustering performance when employing k-means or graph-based clustering methods with an intractable "k" value provided by the consumer. As a result, it is critical to focus on cluster tendency approaches for prior knowledge of the number of clusters in clustering. This paper discusses the numerous visual access tendency (VAT) approaches for calculating the number of clusters.
Author (s) DetailsA mobile ad hoc network (VANET) is a mobile ad hoc network in which network nodes are mobile road vehicles. In order to prevent accidents, the key problem facing VANET is the transmission of the safety message or Warning message between the cars. The large number of vehicles in a specific region can cause data congestion, resulting in a long delay in transmitting the messages. An effective way to relay messages with less delay and a high distribution ratio is to cluster the vehicles into groups. The selection of cluster heads significantly affects the efficiency of the clusters (CHs). In this paper, in order to optimize the efficiency of the network energy and connection quality, the creation of clusters and cluster head selection is formulated as an optimization problem. Two methods have been used to solve the problem: density-based clustering and the selection of cluster heads using the Differential Evolution (DE) Algorithm. The protection or warning signals can be conveniently transmitted to all vehicles by integrating these approaches by dynamically changing transmitting power and containment window size. Compared to and tested against the current AdvB and CBAPA protocols is the performance of the proposed ESMBS protocol. Various performance metrics, such as collision, packet delivery ratio, packet delay ratio, latency, transmission time and throughput, can be regulated by the proposed scheme. By reducing the risk of crashes, the proposed device transmits alarm or safety messages.
Author (s) DetailsThe Wireless Sensor Network (WSN) is evolving today to be a highly promising technology to be used in the future. Different energy-efficient routing protocols have been designed and further developed for the WSNs in order to facilitate the transmission of data to their respective destinations. The various clustering techniques are commonly perused by various researchers to increase their scalability goals and also their lifetime. Many protocols have been used to create a hierarchical structure to minimise the expense of the route at the time of development. To make some connection to the base station. This work effectively increases the energy lifetime and reliability of the network within the clustering protocols for which many protocols have been proposed. The Bat Algorithm (BA), the Bat algorithm, mutation, and the Genetic Algorithm are discussed (GA). This BAT algorithm has search capabilities to solve engineering problems for different applications. Viability has been found for the mutated BAT algorithms Observed in several tasks that were proven and shown by the empirical findings, thereby making the proposed method perform better compared to all schemes. The average packet delivery ratio of mutated BAT is 15.72 percent and 4.72 percent higher than GA and BAT respectively at the number of nodes 300. The average packet delivery ratio of mutated BAT is increased by 25.79 percent and 4.45 percent respectively by the number of nodes 1200 compared to GA and BAT. The average packet delivery ratio of mutated BAT is increased by 23.61 percent at the number of nodes 1800 and by 4.14 percent as compared to GA and BAT respectively.
Computer programming is a cognitive and formal problem solving process that can involve many possible
solutions. Thus, manual evaluation of programming exercises is an onerous task, in particular in the case of