Association Rule Mining System Research Articles

Risk prediction and early warning of pilots’ unsafe behaviors using association rule mining and system dynamics

Risk prediction of pilots' unsafe behaviors is of great significance for preventing unsafe aviation incidents. However, there is a lack of effective and precise approach of dealing with this problem. This study proposes a hybrid approach of combining association rule mining (ARM) method and system dynamics (SD) model to predict and warn pilots' unsafe behaviors. Firstly, the risk factors are identified and classified according to the historical incident data by the human factor analysis and classification system. Then, the association rules between risk factors and unsafe behaviors are obtained by ARM. Finally, the SD is adopted to construct the risk prediction and early warning model for pilots' unsafe behaviors, and the applicability and effectiveness of the model are verified by the actual data from 2016 to 2020. The results of ARM show that there are 48 risk factors affecting pilots' unsafe behaviors, and 142 key association rules are formed between these risk factors and unsafe behaviors; environmental influences, pilots' adverse states, and organizational influences are all strongly related to pilots' unsafe behaviors, and the impact of environmental influences on unsafe behaviors mainly rely on the interaction with other factors. The results of SD demonstrate that although both of the cognitive error risk and the decision error risk show an increasing trend, the former increases slowly and keeps at the level of no alarm while the latter is growing faster and its risk may increase from no alarm to critical alarm; the operation error risk and the violation risk both show a downward trend, and finally remain at the no alarm level; the risk of pilots' unsafe behaviors has a stable fluctuation trend, and the risk values of most time are in the threshold ranges of major warning and critical warning. This work provides theoretical guidance for the decision-makers to develop measures to reduce incidents caused by pilots’ unsafe behaviors.

Privacy-preserving association rule mining via multi-key fully homomorphic encryption

Association rule mining is an efficient method to mine the association relationships between different items from large transaction databases, but is vulnerable to privacy leakage as operates over users’ sensitive data directly. Privacy-preserving association rule mining has emerged to protect users’ privacy during rule mining. Unfortunately, existing privacy-preserving association rule mining schemes suffer from high overhead, fail to support multiple users, and are challenging to prevent collusion attacks between twin-server. To this end, in this paper, we propose a privacy-preserving association rule mining solution via multi-key fully homomorphic encryption over the torus (MKTFHE), which efficiently supports multiple users through a single server only. Specifically, we first construct some multi-key homomorphic gates based on MKTFHE. Then, we designed a series of privacy-preserving computational protocols based on multi-key homomorphic gates. Finally, we build a privacy-preserving association rule mining system with a single cloud server to support multiple users. Moreover, privacy analysis and performance evaluation demonstrate our proposal is efficient and feasible. In contrast to existing solutions, the proposed scheme outperforms encryption and communication, saving approximately 8.5% running time.

Analysis of Run-Off-Road Accidents by Association Rule Mining and Geographic Information System Techniques on Imbalanced Datasets

Run-off-road (ROR) accidents cause a large proportion of fatalities on roads. Exploring key factors is an effective method to reduce fatalities and improve safety sustainability. However, some limitations exist in current studies: (1) Datasets of ROR accidents have imbalance problems, in which the samples of fatal accidents (FA) are much less than non-fatal accidents (NFA). Data mining methods on such imbalanced datasets make the results biased. (2) Few studies conducted spatial analysis of ROR accidents in visualization. Therefore, this study proposes an association rule mining (ARM)-based framework to analyze ROR accidents on imbalanced datasets. A novel method is proposed to address the imbalance problem and ARM is applied to analyze accident severity. Geographic information system (GIS) is adopted for spatial analysis of ROR accidents. The proposed framework is applied to ROR accidents in Victoria, Australia. Six FA factors and seven NFA factors are identified from two-item rules. The results of three-item rules indicate factors acting interactively increase the likelihood of FA or NFA. Hot spots of ROR accidents are presented by GIS maps. Effective measures are accordingly proposed to improve road safety. Compared with traditional data-balancing methods, the proposed framework has been validated to provide more robust and reliable results on imbalanced datasets.

Risks and Predictors of Non-Alcoholic Liver Disease Progression Using Association Rules Mining

Non-alcoholic Steatohepatitis Disease (NASH), a progression of Non-alcoholic Fatty Liver Disease (NAFLD), occurs in case of the increase of fat accumulation in the liver. The disease can next progress to fibrosis, cirrhosis or liver cancer. The most accurate way to diagnose NAFLD progression into NASH is through a liver biopsy. This is painful, expensive and difficult to repeat several times to monitor the fibrosis progression. Thus, finding a non-invasive solution through markers can reliably help tracking the disease progression. The objective of this study is to assess the diagnostic and prognostic performance of serum markers to monitor liver disease progression in comparison to findings by liver biopsy. An association rule mining system is proposed using a Frequent Pattern mining algorithm to reach this objective. An Egyptian cohort consisting of 2300 NAFLD and NASH patients is included in an experimental study, where the results showed that the blood tests and serum markers, PIIINP and ELF, can predict the progression of NAFLD into NASH, and can discriminate between the different stages of NASH with confidence value 0.9. The presented results indicate an advantageous promising non-invasive solution in medicine for predicting of the disease and its progression, while avoiding alternative biopsy exposition

A fuzzy association rule-based knowledge management system for occupational safety and health programs in cold storage facilities

PurposeThis paper aims to improve operational efficiency and minimize accident frequency in cold storage facilities through adopting an effective occupational safety and health program. The hidden knowledge can be extracted from the warehousing operations to create the comfortable and safe workplace environment.Design/methodology/approachA fuzzy association rule-based knowledge management system is developed by integrating fuzzy association rule mining (FARM) and rule-based expert system (RES). FARM is used to extract hidden knowledge from real operations to establish the relationship between safety measurement, personal constitution and key performance index measurement. The extracted knowledge is then stored and adopted in the RES to establish an effective occupational and safety program. Afterwards, a case study is conducted to validate the performance of the proposed system.FindingsThe results indicate that the aforementioned relationship can be built in the form of IF-THEN rules. An appropriate safety and health program can be developed and applied to all workers, so that they can follow instructions to prevent cold induced injuries and also improve the productivity.Practical implicationsBecause of the increasing public consciousness of occupational safety and health, it is important for the workers in cold storage facilities where the ambient temperature is at/below 10°C. The proposed system can address the social problem and promote the importance of occupational safety and health in the society.Originality/valueThis study contributes to the knowledge management system for improving the occupational safety and operational efficiency in the cold storage facilities.

의료 정보 추출을 위한 TF-IDF 기반의 연관규칙 분석 시스템

u-Health에 대한 관심과 IT 기술의 발전에 따라 의료 정보를 적극적으로 활용하고자 하는 요구가 커지고 있으며, 이에 대해 텍스트 형태의 의료 정보 데이터에 연관규칙 기법을 적용하여 질병과 증상과의 관계를 추론하는 시스템에 대한 연구들이 이루어지고 있다. 그러나 일반적인 연관규칙 기법을 의료 정보 데이터에 그대로 적용할 경우, 이전에는 새로운 연관규칙들보다 일반적이며 의미없는 연관규칙들이 많이 생성되는 문제가 발생한다. 또한 필터링으로 인해 빈번하게 함께 발생하지는 않지만 의학적으로 의미있는 항목들의 연관 규칙을 발견할 수 없다는 한계점을 가지게 된다. 본 논문에서는 의료데이터 특성을 고려하여 빈번한 항목과 빈번하지 않지만 의학적으로 의미 있는 항목들을 대상으로 연관규칙을 구성하여 의료 전문가의 의사 결정에 도움을 주기 위한 시스템을 제안한다. 제안 시스템은 의료 기록 데이터에서 용어들을 TF-IDF기반으로 가중치를 부여하고 기존 FP-Growth 알고리즘을 확장하여 TF-IDF 가중치를 고려한 빈번하게 발생하거나 빈번하지 않지만 의미 있는 연관규칙을 구성한다. 특정 질의 데이터가 입력되면 해당 데이터에 나타난 연관 규칙들의 유사도를 의학분야 온톨로지를 이용하여 평가하여 해당 데이터의 내용과 관련된 후보 질병들을 추론한다. 추론된 후보 질병명은 의료 전문가에게 의사 결정의 참고 자료로 제공된다. 실제 임상 진료 및 처방 기록 데이터에 대해 제안 시스템을 적용해 본 결과, 본 제안 시스템을 통해 도출한 연관 규칙이 기존 FP-Growth 알고리즘을 적용했을 때 보다 더 구체적인 질병과 증상과의 관계들을 포함함을 확인할 수 있었다. 또한 본 제안 시스템은 자유형식의 의료 및 병리데이터를 마이닝하고 후보 질병들을 가중치 기반으로 보여주므로, 의료 기록 정보로부터 질병 관련 새로운 정보를 획득하고 의료진의 의사 결정에 도움을 주는 시스템으로 활용될 수 있다. Because of the recent interest in the u-Health and development of IT technology, a need of utilizing a medical information data has been increased. Among previous studies that utilize various data mining algorithms for processing medical information data, there are studies of association rule analysis. In the studies, an association between the symptoms with specified diseases is the target to discover, however, infrequent terms which can be important information for a disease diagnosis are not considered in most cases. In this paper, we proposed a new association rule mining system considering the importance of each term using TF-IDF weight to consider infrequent but important items. In addition, the proposed system can predict candidate diagnoses from medical text records using term similarity analysis based on medical ontology.

An Intelligent Type-II Diabetes Mellitus Diagnosis Approach using Improved FP-growth with Hybrid Classifier Based Arm

Diabetes mellitus has turned out to be a common chronic disease that affects between 2 and 4% of the total population. Recently, most of the system uses association rule mining for diagnosing type-II diabetes mellitus. The most vital concern of association rules is that rules are derived from the complete data set with no validation on samples. Previously, Association rule based Modified Particle Swarm Optimization and Least Squares Support Vector Machine classification is introduced with the capability to lessen the number of rules, looks for association rules on a training set and at last validates them on an independent test set. On the other hand, it only employs categorical data. In case of Type-II Diabetes Mellitus medical diagnosis, the exploitation of continuous data might be essential. With the aim of solving this complication, Improved Frequent Pattern Growth (IFP-Growth) with Hybrid Enhanced Artificial Bee Colony-Advanced Kernel Support Vector Machine (HEABC-AKSVM-IFP Growth) classification based Association Rule Mining (ARM) system is proposed in this study to create rules. This study introduces improved FP-growth to effectively derive frequent patterns including from a vague database in which items possibly will come into view in medical database. Then, HEABC-AKSVM-IFP Growth classifier is employed to create the association rules from the frequent item sets, also keeping away from the rule redundancy and inconsistencies at the time of mining process. Then, results are simulated and evaluated against few classification techniques in terms of classification accuracy, number of derived rules and processing time.

A partition enhanced mining algorithm for distributed association rule mining systems

The extraction of patterns and rules from large distributed databases through existing Distributed Association Rule Mining (DARM) systems is still faced with enormous challenges such as high response times, high communication costs and inability to adapt to the constantly changing databases. In this work, a Partition Enhanced Mining Algorithm (PEMA) is presented to address these problems. In PEMA, the Association Rule Mining Coordinating Agent receives a request and decides the appropriate data sites, partitioning strategy and mining agents to use. The mining process is divided into two stages. In the first stage, the data agents horizontally segment the databases with small average transaction length into relatively smaller partitions based on the number of available sites and the available memory. On the other hand, databases with relatively large average transaction length were vertically partitioned. After this, Mobile Agent-Based Association Rule Mining-Agents, which are the mining agents, carry out the discovery of the local frequent itemsets. At the second stage, the local frequent itemsets were incrementally integrated by the from one data site to another to get the global frequent itemsets. This reduced the response time and communication cost in the system. Results from experiments conducted on real datasets showed that the average response time of PEMA showed an improvement over existing algorithms. Similarly, PEMA incurred lower communication costs with average size of messages exchanged lower when compared with benchmark DARM systems. This result showed that PEMA could be efficiently deployed for efficient discovery of valuable knowledge in distributed databases.

Research on Employment Data Mining for Higher Vocational Graduates

In order to make effective use a large amount of graduate data in colleges and universities that accumulate by teaching management of work, the paper study the data mining for higher vocational graduates database using the data mining technology. Using a variety of data preprocessing methods for the original data, and the paper put forward to mining algorithm based on commonly association rule Apriori algorithm, then according to the actual needs of the design and implementation of association rule mining system, has been beneficial to the employment guidance of college teaching management decision and graduates of the mining results.

Using a Lattice Intension Structure to Facilitate User-Guided Association Rule Mining

Narrowing down the computational space is a key factor in improving the efficiency of an association rule mining system. One approach to achieve this is to let the user guide the association rule mining process by enabling the user to specify the types of association rules that he/she might be interested in. Instead of computing all that can be computed, the system limits its association rule mining process to the discovery of only the association rules that may be of interest to the user, therefore, reducing the computational space and complexity. In this paper, we introduce a new approach for achieving this by using a new structure called lattice intension structure.

Modeling interestingness of streaming association rules as a benefit-maximizing classification problem

In a typical application of association rule learning from market basket data, a set of transactions for a fixed period of time is used as input to rule learning algorithms. For example, the well-known Apriori algorithm can be applied to learn a set of association rules from such a transaction set. However, learning association rules from a set of transactions is not a one time only process. For example, a market manager may perform the association rule learning process once every month over the set of transactions collected through the last month. For this reason, we will consider the problem where transaction sets are input to the system as a stream of packages. The sets of transactions may come in varying sizes and in varying periods. Once a set of transactions arrive, the association rule learning algorithm is executed on the last set of transactions, resulting in new association rules. Therefore, the set of association rules learned will accumulate and increase in number over time, making the mining of interesting ones out of this enlarging set of association rules impractical for human experts. We refer to this sequence of rules as “association rule set stream” or “streaming association rules” and the main motivation behind this research is to develop a technique to overcome the interesting rule selection problem. A successful association rule mining system should select and present only the interesting rules to the domain experts. However, definition of interestingness of association rules on a given domain usually differs from one expert to another and also over time for a given expert. This paper proposes a post-processing method to learn a subjective model for the interestingness concept description of the streaming association rules. The uniqueness of the proposed method is its ability to formulate the interestingness issue of association rules as a benefit-maximizing classification problem and obtain a different interestingness model for each user. In this new classification scheme, the determining features are the selective objective interestingness factors related to the interestingness of the association rules, and the target feature is the interestingness label of those rules. The proposed method works incrementally and employs user interactivity at a certain level. It is evaluated on a real market dataset. The results show that the model can successfully select the interesting ones.

The effect of threshold values on association rule based classification accuracy

Classification Association Rule Mining (CARM) systems operate by applying an Association Rule Mining (ARM) method to obtain classification rules from a training set of previously classified data. The rules thus generated will be influenced by the choice of ARM parameters employed by the algorithm (typically support and confidence threshold values). In this paper we examine the effect that this choice has on the predictive accuracy of CARM methods. We show that the accuracy can almost always be improved by a suitable choice of parameters, and describe a hill-climbing method for finding the best parameter settings. We also demonstrate that the proposed hill-climbing method is most effective when coupled with a fast CARM algorithm such as the TFPC algorithm which is also described.

Product portfolio identification based on association rule mining

It has been well recognized that product portfolio planning has far-reaching impact on the company's business success in competition. In general, product portfolio planning involves two main stages, namely portfolio identification and portfolio evaluation and selection. The former aims to capture and understand customer needs effectively and accordingly to transform them into specifications of product offerings. The latter concerns how to determine an optimal configuration of these identified offerings with the objective of achieving best profit performance. Current research and industrial practice have mainly focused on the economic justification of a given product portfolio, whereas the portfolio identification issue has been received only limited attention. This article intends to develop explicit decision support to improve product portfolio identification by efficient knowledge discovery from past sales and product records. As one of the important applications of data mining, association rule mining lends itself to the discovery of useful patterns associated with requirement analysis enacted among customers, marketing folks, and designers. An association rule mining system (ARMS) is proposed for effective product portfolio identification. Based on a scrutiny into the product definition process, the article studies the fundamental issues underlying product portfolio identification. The ARMS differentiates the customer needs from functional requirements involved in the respective customer and functional domains. Product portfolio identification entails the identification of functional requirement clusters in conjunction with the mappings from customer needs to these clusters. While clusters of functional requirements are identified based on fuzzy clustering analysis, the mapping mechanism between the customer and functional domains is incarnated in association rules. The ARMS architecture and implementation issues are discussed in detail. An application of the proposed methodology and system in a consumer electronics company to generate a vibration motor portfolio for mobile phones is also presented.