Knowledge Discovery Techniques Research Articles

AN INTELLIGENT CLASSIFICATION MODEL FOR PHISHING EMAIL DETECTION

Phishing attacks are one of the trending cyber attacks that apply socially engineered messages that are communicated to people from professional hackers aiming at fooling users to reveal their sensitive information, the most popular communication channel to those messages is through users emails. This paper presents an intelligent classification model for detecting phishing emails using knowledge discovery, data mining and text processing techniques. This paper introduces the concept of phishing terms weighting which evaluates the weight of phishing terms in each email. The pre processing phase is enhanced by applying text stemming and WordNet ontology to enrich the model with word synonyms. The model applied the knowledge discovery procedures using five popular classification algorithms and achieved a notable enhancement in classification accuracy, 0.991 accuracy was achieved using the Random Forest algorithm and 0.984 using J48, which is to our knowledge the highest accuracy rate for an accredited data set. This paper also presents a comparative study with similar proposed classification techniques.

An Intelligent Classification Model for Phishing Email Detection

Phishing attacks are one of the trending cyber-attacks that apply socially engineered messages that are communicated to people from professional hackers aiming at fooling users to reveal their sensitive information, the most popular communication channel to those messages is through users’ emails. This paper presents an intelligent classification model for detecting phishing emails using knowledge discovery, data mining and text processing techniques. This paper introduces the concept of phishing terms weighting which evaluates the weight of phishing terms in each email. The pre-processing phase is enhanced by applying text stemming and WordNet ontology to enrich the model with word synonyms. The model applied the knowledge discovery procedures using five popular classification algorithms and achieved a notable enhancement in classification accuracy; 99.1% accuracy was achieved using the Random Forest algorithm and 98.4% using J48, which is –to our knowledge- the highest accuracy rate for an accredited data set. This paper also presents a comparative study with similar proposed classification techniques.

Process Pharmacology: A Pharmacological Data Science Approach to Drug Development and Therapy.

A novel functional‐genomics based concept of pharmacology that uses artificial intelligence techniques for mining and knowledge discovery in “big data” providing comprehensive information about the drugs’ targets and their functional genomics is proposed. In “process pharmacology”, drugs are associated with biological processes. This puts the disease, regarded as alterations in the activity in one or several cellular processes, in the focus of drug therapy. In this setting, the molecular drug targets are merely intermediates. The identification of drugs for therapeutic or repurposing is based on similarities in the high‐dimensional space of the biological processes that a drug influences. Applying this principle to data associated with lymphoblastic leukemia identified a short list of candidate drugs, including one that was recently proposed as novel rescue medication for lymphocytic leukemia. The pharmacological data science approach provides successful selections of drug candidates within development and repurposing tasks.

Rough – Granular Computing knowledge discovery models for medical classification

Medical domain has become one of the most important areas of research in order to richness huge amounts of medical information about the symptoms of diseases and how to distinguish between them to diagnose it correctly. Knowledge discovery models play vital role in refinement and mining of medical indicators to help medical experts to settle treatment decisions. This paper introduces four hybrid Rough – Granular Computing knowledge discovery models based on Rough Sets Theory, Artificial Neural Networks, Genetic Algorithm and Rough Mereology Theory. A comparative analysis of various knowledge discovery models that use different knowledge discovery techniques for data pre-processing, reduction, and data mining supports medical experts to extract the main medical indicators, to reduce the misdiagnosis rates and to improve decision-making for medical diagnosis and treatment. The proposed models utilized two medical datasets: Coronary Heart Disease dataset and Hepatitis C Virus dataset. The main purpose of this paper was to explore and evaluate the proposed models based on Granular Computing methodology for knowledge extraction according to different evaluation criteria for classification of medical datasets. Another purpose is to make enhancement in the frame of KDD processes for supervised learning using Granular Computing methodology.

A dominance based rough set classification system for fault diagnosis in electrical smart grid environments

Nowadays, power grid monitoring systems are shifting towards more disseminating and distributive operations. The diagnosis of faults using knowledge discovery techniques has become an essential component of the process, accounting to the challenges faced by the electrical power monitoring systems. The system's operators employ various energy management techniques which play important roles in the overall management, reliability and operational sustainability of smart grid utilities. Sometimes, these systems are disrupted by events like a short circuit in the system or any external incursion that could pose a threat to the public safety as well as to the critical infrastructure of the grid. This paper aims at providing a robust system for fault diagnosis using the status of an intelligent electronic device and circuit breakers which can be tripped by any kind of fault. To protect the system from vulnerabilities and different kinds of faults, a multilayered fault estimation classifier, based on the Dominance based rough set is proposed. This technique provides an effective solution to the system's operator for the proper diagnosis of the faults and intrusions by classifying the state of the system. In addition to this, the operator can take preventative measures to protect the system from further damage.

The SpatialARMED Framework: Handling Complex Spatial Components in Spatial Association Rule Mining

Recent research has identified spatial association rule (SAR) mining as a promising technique for geographic pattern mining and knowledge discovery. Nevertheless, important spatial components embedded in the studied phenomenon, in particular complex spatial functional relations such as neighborhood effects and spatial spillover effects have largely been neglected. This article unravels this specific problem to enhance the effective application of SAR mining practices in spatial data analytics. The main discussion focuses on the specification of complex spatial components by means of spatial dependence properties of the data and on how to integrate them in the process of SAR mining. A comprehensive framework dubbed SpatialARMED is proposed for the effective extraction of spatial patterns. The framework is then showcased through its application to crime analysis.

A Novel Decision Tree Framework using Discrete Haar Wavelet Transform

Data Mining is a popular knowledge discovery technique. In data mining decision trees are of the simple and powerful decision making models. One of the limitations in decision trees is towards the data source which they tackle. If data sources which are given as input to decision tree are of imbalance nature then the efficiency of decision tree drops drastically, we propose a decision tree structure which uses discrete haar wavelet transformation technique along with a filter. In this paper, we propose a novel method WT Tree based on above strategy. Extensive experiments, using C4.5 decision tree as base classifier, show that the performance measures of our method is comparable to state-of-the-art methods.

Applying data mining techniques to medical time series: an empirical case study in electroencephalography and stabilometry

One of the major challenges in the medical domain today is how to exploit the huge amount of data that this field generates. To do this, approaches are required that are capable of discovering knowledge that is useful for decision making in the medical field. Time series are data types that are common in the medical domain and require specialized analysis techniques and tools, especially if the information of interest to specialists is concentrated within particular time series regions, known as events.This research followed the steps specified by the so-called knowledge discovery in databases (KDD) process to discover knowledge from medical time series derived from stabilometric (396 series) and electroencephalographic (200) patient electronic health records (EHR). The view offered in the paper is based on the experience gathered as part of the VIIP project.11This work was partially supported by the Spanish Ministry of Education and Science as part of the 2004–2007 National R&D&I Plan through the VIIP Project (DEP2005–00,232-C03).Knowledge discovery in medical time series has a number of difficulties and implications that are highlighted by illustrating the application of several techniques that cover the entire KDD process through two case studies.This paper illustrates the application of different knowledge discovery techniques for the purposes of classification within the above domains. The accuracy of this application for the two classes considered in each case is 99.86% and 98.11% for epilepsy diagnosis in the electroencephalography (EEG) domain and 99.4% and 99.1% for early-age sports talent classification in the stabilometry domain. The KDD techniques achieve better results than other traditional neural network-based classification techniques.

‘Publish or Perish’ in African Studies: New Ways to Valorise Research

AbstractNew modes of publishing African Studies accelerate research, creating digital repositories, but valuing the outcomes is complex, and contested. For tenure of both librarians and academics, refereed publications remain the Holy Grail. Fears of quality decline accompany a rise in self-publishing and inverse social media brevity. Valuing works requires reliable sources, but permanency concerns remain: online news in Africa can quickly disappear, as evident with Ebola reporting; and if you trust Google, in African languages some news never occurred! But at the same time, new modes of publishing and communication allow wider review, and the contesting of canon. Twenty years ago, Africanists hoped open access credentialing was just around the corner. This has not eventuated but a measure of recognition is conceded as society itself goes online. Blogs and open access journals will not clinch jobs, but they can let scholars leave an imprint. Graduate digital courses with blogs replacing essays and completed projects offered back to libraries show a trend to capacitise a new generation of teachers, valorising research. Librarian job descriptions insist on digital skills. Granting bodies privilege the digital. Publishers go online. African studies associations hold digital workshops. If output in core media maintains centrality in ranking and citation then there is now a merging of forms. New books appear, some to acclaim, using digital sources, new knowledge discovery techniques emerge. Across all these trends, there will remain a need to balance digital and print collected by libraries, and refereed writing with more spontaneous communication, but their boundary is likely to narrow.

Mining the Quantified Self: Personal Knowledge Discovery as a Challenge for Data Science.

The last several years have seen an explosion of interest in wearable computing, personal tracking devices, and the so-called quantified self (QS) movement. Quantified self involves ordinary people recording and analyzing numerous aspects of their lives to understand and improve themselves. This is now a mainstream phenomenon, attracting a great deal of attention, participation, and funding. As more people are attracted to the movement, companies are offering various new platforms (hardware and software) that allow ever more aspects of daily life to be tracked. Nearly every aspect of the QS ecosystem is advancing rapidly, except for analytic capabilities, which remain surprisingly primitive. With increasing numbers of qualified self participants collecting ever greater amounts and types of data, many people literally have more data than they know what to do with. This article reviews the opportunities and challenges posed by the QS movement. Data science provides well-tested techniques for knowledge discovery. But making these useful for the QS domain poses unique challenges that derive from the characteristics of the data collected as well as the specific types of actionable insights that people want from the data. Using a small sample of QS time series data containing information about personal health we provide a formulation of the QS problem that connects data to the decisions of interest to the user.

Multifactor Affiliation Analysis: A Multifactor Dimensionality Reduction based Learning Model for Knowledge Discovery and Similarity Measure in 2-way Data Classification

Extracting useful information from the datasets of high dimension and representing the learnt knowledge in an efficient way is a challenge in knowledge discovery and data mining. Although many pattern recognition, knowledge discovery and data mining techniques are available in literature, there is a need for techniques that represent the high dimensional data in a low dimension by preserving useful information for supervised learning. In this work, we design a novel model which effectively captures both inter-feature and intrafeature relationships in the sample space for knowledge discovery by performing dimensionality reduction, using a modified version of multi-factor dimensionality reduction based algorithm. The model uses the learnt knowledge to quantify the similarity of a test sample with respect to a specific class. The evaluation of the model on Fisher’s IRIS dataset containing 50 samples each from three types of IRIS species- setosa, versicolor and verginica, shows that the designed model explores the data set for useful information and assigns test samples to a specific class with measurable similarity indices.

Incorporation of abstraction capability in a logic-based framework by using proximity relations

This paper aims to incorporate a knowledge discovery technique into the Proximity-based Logic Programming paradigm in order to generate background knowledge (conceptual hierarchies) in a semi-automatic way which may lead to an efficient and desirable abstraction process among the symbols (represent ing concepts) from a first-order language and to the discovery of generalized relationship among them i.e. a logic-based framework with the capability of abstraction. This method makes use of the concept of λ-block characterizing the notion of equivalence when working with proximity relations. When the universe of discourse is composed of concepts which are related by proximity, the sets of λ-blocks extracted from that proximity relation can be seen as hierarchical sets of concepts grouped by abstraction level. Then, each group (forming a λ-block) can be labeled, with user help, by means of a more general descriptor in order to simulate a generalization process based on proximity. Thanks to this process, the system can learn concepts that were unknown initially and reply to queries that it was not able to answer. The novelty of this work is that it is the first time a method, with analogous features to the one aforementioned, is implemented inside a fuzzy logic programming framework. Certainly, in order to check the feasibility of the method, we have developed a software tool which have been integrated into the Bousi~Prolog system. Finally, this work presents a method to get a set of recommended abstract descriptors by using WordNet. This allows to improve the original generalization mechanism, helping the user in the task of selecting a convenient abstraction. Also, the overall method can be seen as a technique that facilitates the tuning of term ontologies.

Provide a Method for Increasing the Efficiency of Learning Management Systems using Educational Data Mining

In recent years, data mining approach has been used for optimizing the mass of information on educational organizations and extracting knowledge from these data. This approach develops knowledge discovery techniques from the data in the learning environment, especially from the university students. Applying data mining techniques on the educational data will result in the extraction of useful information and knowledge from them. This information then could be used to improve the learning process and increase the efficiency and performance of students. Therefore, the purpose of this research is to provide a method for increasing the efficiency of learning management system using educational data mining. To this end, a dataset was collected from the students in an E-learning center and the techniques such as classification, clustering and association rules mining were implemented on the specified parameters. Finally, the obtained results are used to improve the management of virtual education systems.

The Impact of Driving Styles on Fuel Consumption: A Data-Warehouse-and-Data-Mining-Based Discovery Process

This paper discusses the results of applied research on the eco-driving domain based on a huge data set produced from a fleet of Lisbon's public transportation buses for a three-year period. This data set is based on events automatically extracted from the control area network bus and enriched with GPS coordinates, weather conditions, and road information. We apply online analytical processing (OLAP) and knowledge discovery (KD) techniques to deal with the high volume of this data set and to determine the major factors that influence the average fuel consumption, and then classify the drivers involved according to their driving efficiency. Consequently, we identify the most appropriate driving practices and styles. Our findings show that introducing simple practices, such as optimal clutch, engine rotation, and engine running in idle, can reduce fuel consumption on average from 3 to 5l/100 km, meaning a saving of 30 l per bus on one day. These findings have been strongly considered in the drivers' training sessions.

An optimized fuzzy method for finding actions

Most of machine learning and data mining techniques only focus on finding patterns satisfying technical interestingness measures. They do not consider business interestingness measures in the mining process, so most of mined patterns may not be usable in corresponding domain. Moreover, the techniques do not interpret the patterns in order that business people know what actions to take to support their business decisions. They require a considerable amount of extra work by domain experts to interpret the patterns. Actionable Knowledge Discovery techniques deliver technical and business significant patterns and support automatically converting patterns into deliverables in business-friendly and operable forms such as actions. In this paper, we incorporate fuzzy set theory and attribute flexibility for the enhancement fuzzy cost-effective action mining algorithm (F-CEAMA). We present an algorithm that suggests actions regarding to their practical applicability. The algorithm takes into account attribute flexibility and fuzzy cost of actions and attempts to maximize the fuzzy net profit. The contribution of the work is in taking the output from fuzzy decision tree, using flexibility of its attributes, and producing novel and actionable knowledge through automatic fuzzy post processing. The performance of proposed algorithm is compared to F-CEAMA using several real-life datasets taken from the UCI Machine Learning Repository. Experimental results show that the proposed algorithm outperforms F-CEAMA not only in finding more actions but also in finding actions with more fuzzy net profit.

A Comment on “A Similarity Measure for Text Classification and Clustering”

A similarity measure namely, similarity measure for text processing (SMTP) is proposed by Lin et al. [1] for knowledge discovery on text collection. The proposed measure considered the three cases for similarity measurements between the pairs of documents. These cases are based on absence and presence of features in the pair of text documents. The first case covers the features appearing in both of the documents, second case covers the features appears in only one document and the third case covers the features appears in none of the documents. The proposed similarity measure considered to be ideal for finding similarity between the pair of text documents on the basis of presence or absence of features available in text documents, however, while exploring the SMTP similarity measurement it is found that the case of measuring similarity between the pair of similar documents is not covered. The objective of this work is to highlight this gap and propose a minor change to make the SMTP a complete similarity measurement technique for knowledge discovery in line with the other standard similarity techniques.

Future Trends of Data Mining in Predicting the Various Diseases in Medical Healthcare System

The thriving medical applications of data mining in the fields of medicine and public health has led to the popularity of its use in knowledge discovery in databases (KDD). Data mining has revealed novel biomedical and healthcare acquaintances for clinical decision making that has great potential to improve the treatment quality of hospitals and increase the survival rate of patients. Disease diagnosis is one of the applications where data mining tools are establishing the successful results. Data mining intends to endow with a systematic survey of current techniques of knowledge discovery in databases using data mining techniques that are in use in today’s medical research. Discussion is made to enable the disease diagnosis and the breakthrough of hidden healthcare patterns from related databases is offered. Also, the use of data mining to discover such relationships as those between health conditions and a disease is presented. It further discusses about the tools that can be used for the processing and classification of data. This paper summarizes various technical articles on medical diagnosis and prognosis. It has also been focused on current research being carried out using the data mining techniques to enhance the disease(s) forecasting process. This research paper provides future trends of current techniques of KDD, using data mining tools for healthcare. It also confers significant issues and challenges associated with data mining and healthcare in general. The research found a growing number of data mining applications, including analysis of health care centers for better health policy-making, detection of disease outbreaks and preventable hospital deaths. The root causes of all diseases get closer towards drugs i.e. the foremost risk factor of all hilarious diseases. Drug addiction using WEKA has been used that brings into light concerning majority of drug abusers started abusing drugs at age below 20yrs.It is to make aware the druggist about the various diseases that are caused with heavy or long term intake of drugs in their life. So, to make an expert system that will awake the youth about precarious use of drugs and also alert the affected person.

Tool Wear Monitoring and Alarm System Based on Pattern Recognition With Logical Analysis of Data

This paper presents a new tool wear monitoring and alarm system that is based on logical analysis of data (LAD). LAD is a data-driven combinatorial optimization technique for knowledge discovery and pattern recognition. The system is a nonintrusive online device that measures the cutting forces and relates them to tool wear through learned patterns. It is developed during turning titanium metal matrix composites (TiMMCs). These are a new generation of materials which have proven to be viable in various industrial fields such as biomedical and aerospace. Since they are quite expensive, our objective is to increase the tool life by giving an alarm at the right moment. The proposed monitoring system is tested by using the experimental results obtained under sequential different machining conditions. External and internal factors that affect the turning process are taken into consideration. The system's alarm limit is validated and is compared to the limit obtained when the statistical proportional hazards model (PHM) is used. The results show that the proposed system that is based on using LAD detects the worn patterns and gives a more accurate alarm for cutting tool replacement.

Life beyond big data: governing with little analytics

The twenty-first-century rise of big data marks a significant break with statistical notions of what is of interest or concern. The vast expansion of digital data has been closely intertwined with the development of advanced analytical algorithms with which to make sense of the data. The advent of techniques of knowledge discovery affords some capacity for the analytics to derive the object or subject of interest from clusters and patterns in large volumes of data, otherwise imperceptible to human reading. Thus, the scale of the big in big data is of less significance to contemporary forms of knowing and governing than what we will call the little analytics. Following Henri Bergson's analysis of forms of perception which ‘cut out’ a series of figures detached from the whole, we propose that analytical algorithms are instruments of perception without which the extensity of big data would not be comprehensible. The technologies of analytics focus human attention and decision on particular persons and things of interest, whilst annulling or discarding much of the material context from which they are extracted. Following the algorithmic processes of ingestion, partitioning and memory, we illuminate how the use of analytics engines has transformed the nature of analysis and knowledge and, thus, the nature of the governing of economic, social and political life.

A reliability-based recommendation method to improve trust-aware recommender systems

Recommender systems (RSs) are programs that apply knowledge discovery techniques to make personalized recommendations for user’s information on the web. In online sharing communities or e-commerce sites, trust is an important mechanism to improve relationship among users. Trust-aware recommender systems are techniques to make use of trust statements and user personal data in social networks. The accuracy of ratings prediction in RSs is one of the most important problems. In this paper, a Reliability-based Trust-aware Collaborative Filtering (RTCF) method is proposed to improve the accuracy of the trust-aware recommender systems. In the proposed method first of all, the initial trust network of the active user is constructed by using combination of the similarity values and the trust statements. Then, an initial rate is predicted for an unrated item of the user. In the next step, a novel trust based reliability measure is proposed to evaluate the quality of the predicted rate. Then, a new mechanism is performed to reconstruct the trust network for those of the users with lower reliability value than a predefined threshold. Finally, the final rate of the unrated item is predicted based on the new trust network of the user. In other words, the proposed method provides a dynamic mechanism to construct trust network of the users based on the proposed reliability measure. Therefore, the proposed method leads to improve the reliability and also the accuracy of the predictions. Experimental results performed on two real-world datasets including; Epinions and Flixster, demonstrated that the proposed method achieved higher accuracy and also obtained reasonable user and rate coverage compared to several state-of-the-art recommender system methods.