Riyanarto Sarno

Decision mining for multi choice workflow patterns

Decision mining is combination of process mining and machine learning technique to retrieve information about how an attribute in a business process affects a cases route choice. It identifies decision point by looking for XOR-splits in petri-net workflow model and analyzing rules for each choice based on available attributes using decision tree. Problem emerges when decision mining technique is used on a workflow that does not use either XOR or AND splits, for example OR-split gateway logic. OR-split does not have explicit representation in petri nets and it makes decision mining algorithm cannot find its decision point. Workflow pattern that uses OR-split as its splitting logic is multi choice. Multi choice does not have its own explicit representation in form of petri net and it is problematic to apply decision mining to this workflow pattern. To make multi choice can be analyzed by decision miner some modification needs to be applied to the petri net representation of this pattern. This paper proposes modification of OR-split gateway representation in petri net. The new representation of OR-split uses combination the existing XOR-split and AND-split to make the model easier to be analyzed using decision miner. The proposed modification do not affect the conformance of event log and process model, but will allow each choice branch to be checked by decision miner.

Clustering of ERP business process fragments

Business process management technology at present has been developed and applied both in small and in large scale. Many companies and organizations use, for instance, Enterprise Resource Planning (ERP) or other business process-oriented system. In this paper, a clustering method in business process model based on its similarity is proposed. This clustering aims to group some similar business processes to form a common business process. A new business process, as a result, can be composed based on similar common business process in order to increase reusability. It is done according to similarity value among business processes that is by calculating the similarity based upon structural and behavioral similarity method. Meanwhile, the clustering process uses a graph partition approach. This research then shows that the clustering result of business process is precise at certain threshold value.

Business process composition based on meta models

Nowadays, Business Process Management (BPM) technology has been developed and widely applied in small and large scale. There are many companies and organizations implemented business process oriented information systems. Therefore, the flexibility issues become very important in managing large quantity of business processes because business processes are always changing overtime. There is a need of flexible business process models so that every possible changes can be done easily every time. Flexible business process models are represented by a reconfigurable model. In this paper, we proposed a method to manage business process model variations in order to efficiently develop the business process repository. The variation management is done by storing the common business processes and its variation in the repository. Then, we also propose meta models which contain information about the models stored in the repository. The results show that the meta models support the composition process of a new business process based on the common business process. The variation management has achieved to reduce storage redundancy up to 82%.

Decomposition using Refined Process Structure Tree (RPST) and control flow complexity metrics

Process mining is a technique that aims to gain knowledge of the event log. The amount of data in the event log is very influential in the Process mining, because it contains millions of activities that shape the behavior of a company. The three main capabilities possessed by mining process is a discovery, conformance, and enhancement. This paper, we present an approach to decompose business processes using Refine Process Structure Tree (RPST). By breaking down a whole into sub models Business Processes (fragments) to the smallest part (atomic) can facilitate the analysis process and can easily be rebuilt. To measure the level of complexity in the model fragment and atomic models we use complexity Control flow metrics. Control flow complexity metrics have two main approaches that are count based measurement and execution path based measurement path. Count based measurement used to describe a static character, while an execution path based measurement used to describe the dynamic character of each model fragment or atomic models (bond fragment).

Improving the accuracy of COCOMO's effort estimation based on neural networks and fuzzy logic model

Constructive Cost Model II (COCOMO II) is investigated as the most popular model for software cost estimation. COCOMO II depends on several variables or Cost Drivers (CD). This research investigates the role of Effort Multiplier (EM) and Line of Code (LOC) to improve the accuracy of cost estimation. Fuzzy Logic has been implemented to the COCOMO II to represent the EM. Furthermore, in order to produce better estimation, this research uses Gaussian Membership Function to redesign the Effort Multiplier by studying the behavior of COCOMO II. This research also applies Neural Network (NN) approach to increase the accuracy of software effort estimation by training the software development datasets. The result is proposed model gives contribution to decrease error significantly.

Comparison of different Neural Network architectures for software cost estimation

This research will observe the use of Artificial Neural Networks (ANN) for estimating software cost. Constructive Cost Model (COCOMO) is the most famous estimating model for software cost, which will be used in this research. The model estimates software cost by calculating several variables which are created by expert with some equations. Furthermore, ANN helps to estimate COCOMO effort accurately. This research offers multilayer feed-forward neural network to adjust COCOMO effort estimation parameters. Also, an algorithm such as Back-propagation is applied to improve the architecture by comparing actual effort with estimated effort and updating the network. However, there are several types of neural network architecture. This research tries to compare several types of architecture by testing each architecture model to dataset. This paper concerns with two different architectures. The difference of this two architecture is basic architecture only uses effort multipliers as input layer while modified architecture divides input layer into two categories such as effort multipliers and scale factors. The result is the proposed model increases the accuracy and each model has different result.

Business process anomaly detection using ontology-based process modelling and Multi-Level Class Association Rule Learning

Many companies in the world have used the business process management system (BPMS). This system is used to manage and analyze the running business process in the company. Every business process has a possibility to have changes in its realization. Those changes generate some variations of the business process. The variations, can be in line with the companys principles and or become an anomaly for the company. These anomalies can cause frauds which make some losses for the company. In order to reduce the losses, business process anomaly detection method is needed. This paper proposed ontology-based process modeling to model and capture the business process anomalies and the method of multi-level class association rule learning (ML-CARL) to detect fraud in business process. From the experiment which have been done in this paper, the accuracy of 0.99 was obtained from the ML-CARL method. It could be concluded that ontology-based process modeling and the ML-CARL method can detect business process anomalies well.

Weighted Ontology and weighted tree similarity algorithm for diagnosing Diabetes Mellitus

Application knowledge base for diabetes such as expert systems has been developed, but generally using conventional methods that have limitations in representing knowledge. Ontology supports the search of data / information by defining the concept of convergent intended by the user. This study using Diabetes Mellitus Classification based diabetes disease diagnosis from World Health Organization Geneva. This system receives input patient data from user. Then, system will build the patient ontology to represent patient knowledge. We are connecting Java applications to Protégé using OWL API. Then, system will calculate the weight of an ontology based on density. This system use JENA Inference Engine and working memory area for reasoning. The system would then do process similarity matching with Ontology Diabetes Mellitus using weighted tree similarity algorithm. Ontology has the highest similarity value will be the proposed diagnosis. Results of this study show that the representation in the form of OWL ontology using weighted ontology and weighted tree similarity algorithm can be used to represent knowledge about diabetes mellitus.

Evaluation maturity index and risk management for it governance using Fuzzy AHP and Fuzzy TOPSIS (case Study Bank XYZ)

Risk management maturity index should be measured to determine whether the application of risk management within the organization succeed or not. Assessment of risk management maturity index is very important because it allows the identification of strengths and weaknesses of the organization that can be used to improve corporate governance and organization risk managementt. Many measurement methods that have been used, but when but when we get a less good data, this will cause some problems. Existing data are sometimes inadequate for problems in real life. For that we need a new model to perform a measurement. In this study using fuzzy logic models in making a decision of structured preference maker. Fuzzy theory helps in measuring the concept of uncertainty related to human which is subjective. Two applications of fuzzy namely Fuzzy AHP is used to determine the weight of the specified criteria and Fuzzy TOPSIS to rank of selected alternatives. This study uses a case study of Bank XYZ as an object of research. The results of this research to get Skills and Expertise (SE): 0.041641. For the calculation of risk management SW/HW (Slow Connection): 0.87410948

Weighted ontology for subject search in Learning Content Management System

Learning Content Management System (LCMS) is a powerful tool for supporting distance learning. One of the LCMS development problems is subject searching. Most users have no idea about subject name of what they are looking for. They only know about subject contents. Nowadays, search engine embedded with LCMS gives result based on string matching keyword based. Precision and recall of this method is low. This research proposes subject name search based on document content using weighted ontology. Ontology is built from extracted term. Each term is given a weight based on the number of its relation. User query is expanded based on its synonym in WordNet. It is also weighted and taken into account of its similarity with course ontology. System retrieves similar or same subjects based on user query. Precision and recall of weighted ontology search result is better than string matching keyword search.