Intelligent Information and Database Systems

One of the fundamental challenges of distributed multi-agent systems relates to the problem of knowledge integration – where a collective stance of the distributed system needs to be determined and justified. Assuming a simple multi-agent system we present an intuitive and consistent approach to the integration of imprecise descriptions of individual beliefs about the external world. In particular, focusing our attention to simple modal statements of certainty, i.e., possibility, belief and knowledge, we state rational set of common-sense postulates against the integration results. Further, utilising the Grounding Theory, as the means for appropriate grounding of statements, and incorporating the approach with theory of mind, as the underlying mechanism for interpretation of statements, we introduce a two-stage integration procedure. Finally, we prove the appropriateness of the proposed solution, show its basic properties, and provide a simple computational example.

In the paper, the power system topology verification with use of multiagent systems is considered. Two multiagent systems are taken into account. One of these systems is a modification of the second one. In the modified system, there are additionally so-called substation agents. Stages of analysis, design and investigation of performance characteristics of presented multiagent systems are described in the paper. The goal of the paper is presentation of performance characteristics of the mentioned multiagent systems in terms of probabilistic characteristics of agent activity and created messages. The carried out investigations show that the modified multiagent system has much better features than the earlier system.

In this paper, we propose a model of an Intelligent Multi-Agent System based on three distributed knowledge bases for solving problems automatically. Besides, we present architectures of agents in the system. We also illustrate an application of this model in three fields: plane geometry, 2D analytic geometry, and algebra. In our application, we use JADE platform, Maple, MathML, XML, …Finally, we show a method to test effects of the system developed from the model of MAS proposed.

A critical issue in patient planning is to determine whether the medical processes of a patient can be completed by a time constraint based on the available resources in hospitals. The problem is a Temporal Constraint Satisfaction Problem (TCSP). The objectives of this paper are to propose a viable and systematic approach to develop a distributed cooperative problem solver for TCSP and estimate the shortest and the longest completion time for handling a patient in the presence of uncertainty based on Multi-agent systems (MAS) architecture. Our approach combines MAS with a subclass of time Petri net (TPN) models to solve TCSP. Existing analysis methods of TPN based on state classes cannot be applied directly due to distributed architecture of MAS. In this paper, a temporal analysis method based on MAS architecture is proposed. Our temporal analysis method efficiently deduces the earliest and latest completion time of a patient based on interaction between agents.

This paper introduces an agent-based framework to modeling and simulate assessing rice area infested by brown plant hoppers from field to regional scales using agent-based and upscaling approach. Because detail levels of infestation rice area information are different from small scale to large scale, infestation rice area information are collected from fields in the Mekong Delta region-Vietnam and are upscaled to province scale, results are validated by province estimation reports. From the results, infestation rice area information can be used as an indicator for assessing damage levels of rice crop seasons caused by Brown plant hoppers, agent-based model and uscaling method could be an usefull tool for aggregating information from field to decision-makers and support planning argicultural strategics.

Rapidly growing collection of communication services is now available on the Internet. A communication service is a component in a server that provides capabilities to clients. Services available on the Internet include: WWW browsing and content retrieval services software distribution service. A common way to implement these services is to develop each one as a separate program and then compile, link, and execute each program in a separate process. However, this “static” approach to configuring services yields inflexible, often inefficient, applications and software architectures. The main problem with static configuration is that it tightly couples the implementation of a particular service with the configuration of the service with respect to other services in an application. In this paper we propose a system for dynamically configuring communication services. Server will invoke and manage services based on time stamp of service. The system will reduce work load of sever all services in executed by different threads based on time services are executed, suspended and resumed. Different patterns are used designing of service administration pattern that are reflective monitoring, strategy and thread per connection. This paper satisfies the properties of autonomic system: For monitoring use reflective monitoring, Decision making we use strategy pattern. Thread per connection is used of executing service in different thread. The pattern is described using a java-like notation for the classes and interfaces. A simple UML class and Sequence diagrams are depicted.

Investigating the root causes of abnormal events is a crucial task for an industrial chemical process. When process faults are detected, isolating the faulty variables provides additional information for investigating the root causes of the faults. Numerous data-driven approaches require the datasets of known faults, which may not exist for some industrial processes, to isolate the faulty variables. The contribution plot is a popular tool to isolate faulty variables without a priori knowledge. However, it is well known that this approach suffers from the smearing effect, which may mislead the faulty variables of the detected faults. In the presented work, a contribution plot without the smearing effect to non-faulty variables was derived. An industrial example, correctly isolating faulty variables and diagnosing the root causes of the faults for the compression process, was provided to demonstrate the effectiveness of the proposed approach for industrial processes.

In the paper problem of planning training protocol with taking into account limitations on the training intensity due to the health problems of the exerciser is considered. In the first part of the work short introduction to existing solutions in the area of eHealth applications is given. Next, architecture of the eHealth system to support exerciser training is discussed. The main functionalities of proposed system are pointed out and challenges are highlighted. The concept of context-awareness and personalization is stressed. At the end the problem of model based optimisation of the training protocol is formulated.

This paper proposes a solution algorithm to locate the facial features on the human face images. First, the proposed algorithm determines the face region based on skin-tone segmentation and morphological operations. Then, we locate the facial features (i.e. brows, eyes, and mouth) by their color information. Finally, this algorithm set the shape control points based on the Facial Animation Parameters in the MPEG-4 standard on the located facial features. Results of experiments to the face images show that the proposed approach is not only robust but also quit efficient.

In class teaching with a large number of students, teachers lack sufficient time in understanding individual student learning situation. The framework of learning activity in this study is based on the Learning Diagnosis Diagram. Before conducting learning activities, teachers must prepare Learning Diagnosis Diagrams. This work proposes an adaptive Learning Diagnosis Diagram to consider differences among students. The proposed system provides a personalized Learning Diagnosis Diagram for individual students and adjusts learning phases to automatically fit student achievement. The learning evaluation demonstrates the effectiveness of the proposed method. The evaluation shows that the Learning Diagnosis Diagram can provide an adaptive learning environment for students.

In Technical Image Press Association (TIPA) Awards 2009[1], Sony DSC-HX-1 received a honor of “Best Super zoom D-camera”. In fact, Sony HX-1 is recognized not only having a 20x optical super zoom lens, but offering many special video effects. It can be a new trend to judge current DSLR-like camera, such as Fujifilm FinePix HS10 received the same Award 2010[2] with same reasons again. Theoretically, it is a new integration technology for video virtually reality, which provide multiple platform users for video camera, video game, and mobile phone all together. Administers from variety of fields begin to think how to integrate some hot-selling video scene effects from all possible mobile video products, developing this “dazzling” virtually reality (VR) imagination beyond limitation, to attract more potential consumers, which can be vital for small businesses to survive in the future. In our experiment, we collect more than 300 cases from the telephone survey during March 2011 to Aug 2011. Total of 212 cases comply with the conditions. To probe mainly into the relationship between new generation video effects confidence level and 3 potential consumers: Amateur Photographer (AP), Senior Photographer (SP), and college student (CS). In our experiment, we collect more than 300 cases from the telephone survey during March 2011 to Aug 2011. Total of 212 cases comply with the conditions. To probe mainly into the relationship between new generation video effects confidence level and 3 potential consumers: Amateur Photographer (AP), Senior Photographer (SP), and college student (CS). That is the reason we are probe into this highly competitively market with brilliant creative design, and hope to offer an objective suggestion for both industry and education administers. .

The purpose of this article is to incubate the establishment of semantic web services for next generation academic electronic library via Web 3.0. The e-library users will be extremely beneficial from the forthcoming semantic social web mechanism. Web 3.0 will be the third generation of WWW and integrate semantics web, intelligent agent, and Distributed Artificial Intelligence into the ubiquitous networks. On top of current library 2.0 structures, we would be able to fulfill the Web 3.0 electronic library. We design the deployment of intelligent agents to form the semantic social web in order to interpret linguistic expressions of e-library users without ambiguity. This research is conducting the pioneering research to introduce the future and direction for the associate academic electronic library to follow the proposed guidelines to initiate the construction of future library system in terms of service-oriented architecture. This research article is the pioneering practice of future academic digital libraries under Web 3.0 structures.

In this paper, we use fuzzy reasoning techniques and the domain ontology for anti-diabetic drugs selection. We present an anti-diabetic drugs recommendation system based on fuzzy rules and the anti-diabetic drugs ontology to recommend the medicine and the medicine information. The experimental results show that the proposed anti-diabetic drugs recommendation system has a good performance for anti-diabetic drugs selection.

This paper presents an image resizing system, the purpose is to build an image resizing system based on aesthetic composition including rule of thirds and subject position. With global operations, using the traditional scaling and non-traditional content-aware image resizing do the global operation, supplemented by photo rating for adaptive adjustment to reduce user operation with clear quantitative criteria as a basis for adjustment. In non-traditional content-aware image resizing, two algorithms are used, Seam Carving for Content-Aware Image Resizing and Adaptive Content-Aware Image Resizing, to adjust path detection with Otsu’s Method according to the above-mentioned algorithm diagram.

In artificial intelligence, knowledge models play an important role in designing the knowledge base systems and the expert systems. The quality of the intelligent systems depends heavily on the knowledge base built on the models. This paper presents a model of knowledge called functional knowledge model (FKM). This model is used to represent knowledge domains about functions and computing relations between them in different real applications. Besides, this paper also proposes a technique for solving an important problem in functional knowledge domains - simplification of functional expressions. Functional knowledge model is applied to construct an automatic system for simplifying trigonometric expressions in high school. This system can reason automatically from knowledge and provides a step by step solution that similar to the way of human thinking.

The main aim of this research is to deal with enriching conceptual semantic by expanding local conceptual neighbor. The approach consists of two phases: neighbor enrichment phase and matching phase. The enrichment phase is based on analysis of the extension semantic the ontologies have. The extension we make use of in this work is generated an contextually expanded neighbor of each concept from external knowledge sources such as WordNet, ODP, and Wikimedia. Outputs of the enrichment phase are two sets of contextually expanded neighbors belonging to these two corresponding ontologies, respectively. The matching phase calculates similarities between these contextually expended neighbors, which yields decisions which concepts are to be matched.

In this paper, a novel density,

M

-density, was proposed. Based on this new density, a novel composition forecasting model was also proposed. For comparing the forecasting efficiency of this new density with the well-known density,

N

-density, a real data experiment was conducted. The performances of Choquet integral composition forecasting model with extensional L-measure,

λ

-measure and P-measure, by using

M

-density and

N

-density, respectively, a ridge regression composition forecasting model and a multiple linear regression composition forecasting model and the traditional linear weighted composition forecasting model were compared. Experimental result showed that the Choquet integral composition forecasting model with respect to extensional L-measure based on

M

-density outperforms other composition forecasting models. Furthermore, for each fuzzy measure, including the L

E

-measure, L-measure,

λ

-measure and P-measure, the

M

-density based Choquet integral composition forecasting model is better than the

N

-density based.

This paper presents the design of an intelligent cart system to be used in a typical airport. The intelligent cart system consists of a set of mobile software agents to control the cart and provides a novel method for alignment. If the carts gather and align themselves automatically after being used, it is beneficial for human workers who have to collect them manually. To avoid excessive energy consumption through the collection of the carts, in the previous study, we have used ant colony optimization (ACO) and a clustering method based on the algorithm. In the current study, we have extended the ACO algorithm to use the vector values of the scattered carts in the field instead of mere location. We constructed a simulator that performs ant colony clustering using vector similarity. Waiting time and route to the destination of each cart are made based on the cluster created this way. These routes and waiting times are conveyed by the agent to each cart, while making them in rough lines. Because the carts are clustered by the similarity of vectors, we have observed that several groups have appeared to be aligned. The effectiveness of the system is demonstrated by constructing a simulator and evaluating the results.

Fuzzy ontology are often more useful than non-fuzzy ontologies in knowledge modeling owing to the possibility for representing the incompleteness and uncertainty. In this paper we present an approach to fuzzification on the instance level of ontology using multi-value and multi-attribute structure. A consensus-based method for fuzzy ontology integration is proposed.

The research and development of various autonomous robots have been conducted for seeking methods of making multiple robots cooperate efficiently. In this paper we report an experiment to control multiple robots by using a set of mobile agents. Our previous study succeeded in making autonomous robots roughly gather by using ACC (Ant Colony Clustering), while suppressing energy consumption. The robots that are gathered are in arbitrary shapes. It is easier for human laborer to collect then if some of them form lines. We have studied to make the robots, which are roughly gathered by using ACC, form short lines. In this paper, we propose the line forming technique of the autonomous robots to achieve the above-mentioned purpose. We have constructed a simulator to show the movements of many robots based on the data collected from a few real robots. The results of the simulation demonstrate the effectiveness of the technique.

This paper presents a design and evaluates the performance of an efficient charging scheduler for electric vehicles, aiming at reducing the peak load of a fast charging station while meeting the time constraint of all charging requests. Upon the task model consist of actuation time, operation length, deadline, and a consumption profile, the proposed scheduler fills the allocation table, by which the power controller turns on or off the electric connection switch to the vehicle on each time slot boundary. For the sake of combining the time-efficiency of heuristic-based approaches and the iterative evolution of genetic algorithms, the initial population is decided by a heuristic which selects necessary time slots having the lowest power load until the previous task allocation. Then, the regular genetic operations further improve the schedule, additionally creating a new chromosome only from the valid range. The performance measurement result obtained from a prototype implementation shows that our scheme can reduce the peak load for the given charging task sets by up to 4.9 %, compared with conventional schemes.

A motor control model based on reinforcement learning (RL) is proposed here. The model is inspired by organizational principles of the cerebral cortex, specifically on cortical maps and functional hierarchy in sensory and motor areas of the brain. Self-Organizing Maps (SOM) have proven to be useful in modeling cortical topological maps. The SOM maps the input space in response to the real-valued state information, and a second SOM is used to represent the action space. We use the Q-learning algorithm with a neighborhood update function, and an SOM for Q-function to avoid representing very large number of states or continuous action space in a large tabular form. The final model can map a continuous input space to a continuous action space.

Facial feature extraction plays an important step in automated visual interpretation and human face recognition. Detecting facial feature is a crucial role in a wide variety of application such as human computer interface, facial animation and face recognition, etc. The major objective of this paper is to review the recent developments on the methods of facial feature extraction. This study summaries different method for feature point extraction and their applications on face image identification and highlight the performance regarding these methods. The major goal of the paper is to provide a summary reference source for the researchers involved in facial feature extraction.

This study proposes a straightforward and efficient infant location system to reduce the potential risks of the theft and misuse hold. The system can recognize without difficulty different locations of newborn babies which are attached wristband active RFID tags. The system can accurately recognizes the locations of newborn babies by using decision tree classifiers after the active RFID readers has received different intensity of electromagnetic waves transmitted by active RFID tags.

This study proposes a straightforward and efficient intelligently remote infant monitoring system to reduce the potential risks of the theft, misuse hold and abnormal body temperature. The system can accurately recognizes the locations of newborn babies by using neural network classifiers after the active RFID readers has received different intensity of electromagnetic waves transmitted by active RFID tags. The newborn babies of temperature anomalies also can be diagnosed by the body temperature sensors and the proposed infant monitoring system. The remote infant monitoring system improved infant care and safety, reduced systems and human-based errors and enabled fast communicating with the clinical staff and families. This system can be used for infants at home or in a hospital nursery room.

At present, Radio Frequency Identification (RFID) technology has been widely accepted in hospitals to monitoring newborn babies.This study proposes a straightforward and efficient infant monitoring system to reduce the potential risks of the theft, misuse hold and abnormal body temperature. The system can recognize without difficulty different locations of newborn babies which are attached wristband active RFID tags. The system can accurately recognizes the locations of newborn babies by using Bayesian network classifier after the active RFID readers has received different intensity of electromagnetic waves transmitted by active RFID tags. The infant monitoring system also can detect temperature anomalies of newborn babies real time by the body temperature sensors.

The purpose of this study is to present a fuzzy diagnosing machine fault to support the developing machine diagnosis system. The fuzzy evaluation is used to process the problems of which the fault causes and the symptoms are dealing with the uncertainty environment. In this study, we propose two propositions to treat the machine diagnosis fault.

We focus on a potential capability of Exploitation-oriented Learning (XoL) in non-Markov multi-agent environments. XoL has some degree of rationality in non-Markov environments and is also confirmed the effectiveness by computer simulations. Penalty Avoiding Rational Policy Making algorithm (PARP) that is one of XoL methods was planed to learn a penalty avoiding policy. PARP is improved to save memories and to cope with uncertainties, that is called Improved PARP. Though the effectiveness of Improved PARP has been confirmed on computer simulations, there is no result in real world environment. In this paper, we show the effectiveness of Improved PARP in real world environment using a keepaway task that is a testbed of multi-agent soccer environment.

The similarity search problem in streaming time series has become a hot research topic since such data arise in so many applications of various areas. In this problem, the fact that data streams are updated continuously as new data arrive in real time is a challenge due to expensive dimensionality reduction recomputation and index update costs. In this paper, adopting the same ideas of a delayed update policy and an incremental computation from IDC index (Incremental Discrete Fourier Transform(DFT) Computation – Index) we propose a new approach for similarity search in streaming time series by using MP_C as dimensionality reduction method with the support of Skyline index. Our experiments show that our proposed approach for similarity search in streaming time series is more efficient than the IDC-Index in terms of pruning power, normalized CPU cost and recomputation and update time.

In this paper, a hybrid decision forest is constructed by double randomization of the original training set. In this decision forest, each individual base decision tree classifiers are incorporated with an additional classifier model, the

Logitboosted

decision stump. In the first randomization, the resamples to train the decision trees are extracted; in the second randomization, second set of resamples are generated from the out-of-bag samples of the first set of resamples. The boosted decision stumps are constructed on the second resamples. These extra resamples along with the resamples on which the base tree classifiers are trained, approximates the original training set. In this way we are utilizing the full training set to construct a hybrid decision forest with larger feature space. We have applied this hybrid decision forest in two real world applications; a) classifying credit scores, and b) short term extreme rainfall forecast. The performance of the hybrid decision forest in these two problems are compared with some well known machine learning methods. Overall results suggest that the new hybrid decision forest is capable of yielding commendable predictive performance.

In the paper three learning and inference ordering approaches in the method for structured output classification are presented. As it was previously presented by authors, classification of single element in output structure can be performed by generalization of input attributes as well as already partially classified output elements [9]. The paper addresses crucial problem of how to order elements in the structured learning process to get greater final accuracy. The learning is performed by means of ensemble, boosting classification method adapted to structured prediction - AdaBoostSeq algorithm. Authors present several ordering heuristics for score function application in order to obtain better structured output classification accuracy.

In this paper the prototype of an e-learning system that incorporates learning style is described. This system with a personalized courseware was used to conduct an experiment. The e-learning system collected information about a student during the registration process. Next, a user was assigned to an experimental or a control group. Depending on the previous classification the student was offered a learning scenario suited to individual learner’s preferences or a universal learning scenario. The research was devoted to measure students’ learning results from both groups. Significantly higher results were obtained by learners whose user profiles were taken into consideration during the determination of the learning scenario.

In the age of rapidly increasing volumes of data, human experts have come to the urgent need to extract useful information from the huge amount of data. Knowldege discovery in databases has obtained much attention for researches and applications in business and in science. In this paper, we present a neuro-fuzzy approach using complex fuzzy sets (CFSs) for the problem of knowledge discovery. A CFS is an advanced fuzzy set, whose membership is complex-valued and characterized by an amplitude function and a phase function. The application of CFSs to the proposed complex neuro-fuzzy system (CNFS) can increase the functional mapping ability to find missing data for knowledge discovery. Moreover, we devise a hybrid learning algorithm to evolve the CNFS for modeling accuracy, combining the artificial bee colony algorithm and the recursive least squares estimator method. The proposed approach to knowledge discovery is tested through experimentation, whose results are compared with those by other approaches. The experimental results indicate that the proposed approach outperforms the compared approaches.

In the past, the MFFP-tree algorithm was proposed to handle the quantitative database for efficiently mining the complete fuzzy frequent itemsets. In this paper, we propose an integrated MFFP (called iMFFP)-tree algorithm for merging several individual MFFP trees into an integrated one. It can help derive global fuzzy rules among distributed databases, thus allowing managers to make more sophisticated decisions. Experimental results also showed the performance of the proposed approach.

We consider a special class of codes, namely

ω

-codes related to infinite word which had been studied by many authors. Until now, the best algorithm to test whether a regular language

X

is an

ω

-code has time complexity

${\cal O}(n^3)$

, where

n

is the size of the transition monoid of the minimal automaton recognizing

X

. In this paper, with any monoid

M

saturating

X

(the transition monoid above is only a special case), we propose a new test and establish a quadratic testing algorithm with time complexity

${\cal O}(n^2)$

to verify if

X

is an

ω

-code, where

n

is Card(

M

).

Attribute Oriented Induction method (short for AOI) is one of the most important methods of data mining. The input value of AOI contains a relational data table and attribute-related concept hierarchies. The output is a general feature inducted by the related data. Though it is useful in searching for general feature with traditional AOI method, it only can mine the feature from the single-value attribute data. If the data is of multiple-value attribute, the traditional AOI method is not able to find general knowledge from the data. In addition, the AOI algorithm is based on the way of induction to establish the concept hierarchies. Different principles of classification or different category values produce different concept trees, therefore, affecting the inductive conclusion. Based on the issue, this paper proposes a modified AOI algorithm combined with a simplified Boolean bit Karnaugh map. It does not need to establish the concept tree. It can handle data of multi value and find out the general features implied within the attributes.

This paper proposes a new approach by combining the Evolutionary Algorithm and Imperialist Competitive Algorithm. This approach tries to capture several people involved in community development characteristic. People live in different type of communities:

Monarchy

,

Republic

and

Autocracy

. People dominion is different in each community. Research work has been undertaken to deal with curse of dimensionality and to improve the convergence speed and accuracy of the basic ICA and EA algorithms. Common benchmark functions and large scale global optimization have been used to compare HEICA with ICA, EA, PSO, ABC, SDENS and jDElsgo. HEICA indeed has established superiority over the basic algorithms with respect to set of functions considered and it can be employed to solve other global optimization problems, easily. The results show the efficiency and capabilities of the new hybrid algorithm in finding the optimum. Amazingly, its performance is about 85% better than others. The performance achieved is quite satisfactory and promising.

Since the early 1990s, fMRI has come to dominate the brain mapping field due to its relatively low invasiveness, absence of radiation exposure, and relatively wide availability. It is widely used to get a 3-D map of brain activity, with a spatial resolution of few milliseconds. We try to employ various machine learning techniques to decode the cognitive states of a person, based on his brain fMRIs. This is particularly challenging because of the complex nature of brain and numerous interdependencies in the brain activity. We trained multiple classifiers for decoding cognitive states and analyzed the results. We also introduced a technique for considerably reducing the large dimensions of the fMRI data, thereby increasing the classification accuracy. We have compared our results with current state-of-the-art implementations, and a significant improvement in the performance was observed. We got 90% accuracy, which is significantly better than the state-of-the-art implementation. We ran our algorithm on a heterogeneous dataset containing fMRI scans from multiple persons, and still got an accuracy of 83%, which is significant since it shows our classifiers were able to identify some basic abstract underlying neural activity, which are subject-independent, corresponding to the each cognitive states.

Finding discriminative motifs has recently received much attention in biomedical field as such motifs allows us to characterize in distinguishing two different classes of sequences. Although the developed methods function on labeled data, it is common in biomedical applications that the quantity of labeled sequences is limited while a large number of unlabeled sequences is usually available. To overcome this obstacle, this paper presents a proposed semi-supervised learning method that enables the user to exploit unlabeled sequences to enlarge labeled sequence set, leading to improvement of the performance in finding discriminative motifs. The comparative experimental evaluation of the proposed semi-supervised learning shows that it can improve considerably the predictive accuracy of the found motifs.

In the paper we consider the two-stage binary classifier based on Bayes rule. Assuming that both the tree structure and the feature used at each non-terminal node have been specified, we present the expected total cost. This cost is considered for two types of loss function. First is the zero-one loss function and second is the node-dependent fuzzy loss function. The work focuses on the difference between the expected total costs for these two cases of loss function in the two-stage binary classifier. The obtained results are presented on the numerical example.

The rotation forest ensemble method using a genetic fuzzy rule-based system as a base learning algorithm was developed in Matlab environment. The method was applied to the real-world regression problem of predicting the prices of residential premises based on historical data of sales/purchase transactions. The computationally intensive experiments were conducted aimed to compare the accuracy of ensembles generated by our proposed method with bagging, repeated holdout, and repeated cross-validation models. The statistical analysis of results was made employing nonparametric Friedman and Wilcoxon statistical tests.

One of the serious problems of modern pattern recognition is concept drift i.e., model changing during exploitation of a given classifier. The paper proposes how to adapt a single classifier system to the new model without the knowledge of correct classes. The proposed simulated concept recurrence is implemented in the non-recurring concept shift scenario without the drift detection mechanism. We assume that the model could change slightly, what allows us to predict a set of possible models. Quality of the proposed algorithm was estimated on the basis of computer experiment which was carried out on the benchmark dataset.

LLE(Local linear embedding) and Isomap are widely used approaches for dimension reduction. For LLE, the neighborhood selection approach is an important research issue. For different types of datasets, we need different neighborhood selection approaches to have better chance for finding reasonable representation within the required number of dimensions. In this paper, the

ε

-distance approach and a modified version of

k

-nn method are introduced. For LLE and Isomap, the eigenvectors obtained from these methods are much more discussed, but there are more information hidden in the corresponding eigenvalues which can be used for finding embeddings contains more data information.

Rough set is a widespread concept in computer science and is applicable in many fields such as artificial intelligence, expert systems, data mining, pattern recognition and decision support systems. One of key problems of knowledge acquisition in theoretical study of rough sets is attribute reduction. Attribute reduction also called feature selection eliminates superfluous attributes in the information system and improves efficiency of data analysis process. But reducing attributes is a NP-hard problem. Recently, to overcome the technical difficulty, there are a lot of research on new approaches such as maximal tolerance classification (Fang Yang et al. 2010), genetic algorithm (N. Ravi Shankar et al. 2010), topology and measure of significance of attributes (P.G. JansiRani and R. Bhaskaran 2010), soft set (Tutut Herawan et al. 2010), positive approximation (Yuhua Qian et al. 2010), dynamic programming (Walid Moudani et al. 2010). However, there are still some challenging research issues that time consumption is still hard problem in attribute reduction. This paper introduces a new approach with a model presented with definitions, theorems, operations. Set of maximal random prior forms is put forward as an effective way for attribute reduction. The algorithm for seeking maximal random prior set are proposed with linear complexity, contributes to solve absolutely problems in attribute reduction and significantly improve the speed of calculation and data analysis.

Modern logistics companies today rely on advanced ICT solutions for information processing and sharing. Access to data and information about the demand for logistics services and supply opportunities are becoming a key competitive factor. Unfortunately, only the largest companies can afford advanced systems. Small and medium logistics companies have limited or no IT-competence. Tools are therefore needed to facilitate cooperation between smaller logistics companies, which in turn will reduce transaction costs. The paper proposes an idea of the SMART model, which is based on agent technology and cloud computing. It will allow easier collection and flow of information as well as better and cheaper access to logistics management systems.

The article presents investigations upon the effectiveness of computer assistance in the transformation of explicit and tactic knowledge in the course of supplier selection process in an enterprise. A computer system is oriented towards achieving greater process efficiency than in a traditional procedure based upon expert knowledge of employees. Authors of the article have modeled a multi-variant process of supplier selection providing two options, one with computer assistance and the other without this assistance. The obtained results are used to identify the restrictions and formulate the conclusions as for the requirements that should be met by the systems assisting the decision-making in this area.

Companies are facing problems regarding reduction of their logistics costs. Good organization of transport processes can bring a lot of profits both economical and environmental. Companies participating in supply process more and more often prefer to create temporary relations and form virtual cooperation networks than to keep traditional long-term contracts. The aim of the paper is to present the tool that supports the coordination of transport in virtual logistics clusters. The main problems and requirements are identified. Moreover authors present the results of questionnaire conducted among transport users and transport providers regarding the communications standards.

The supply chain configuration has recently been one of the key elements of supply chain management. The complexity of the relations and variety of the aims of their particular members cause it to be very difficult to build a supply chain effectively. Therefore, finding a feasible configuration in which both the business network and the company can achieve the highest possible level of performance constitutes a problem. Authors proposed the SCtechNet model based on graph theory, business network concept and the competitiveness indicator that helps to solve this problem by dynamic configuration of supply chains. The simulation results based on proposed model are presented and discussed.

An article presents identification methodologies and estimation of factors which influence a logistic process safety in a network context through the use of a network thinking methodology. A proposed version of a network thinking methodology for the use of a safety analysis of logistic processes realisation, unlike an originalProbst and Urlich concept, only uses its modified stages. A significant element which differentiates a solution presented by the authors is the use of one of grey systems theory’s methods so called Grey Relational Analysis in order to quantitatively formulate a common experts’ opinion concerning an impact mutual force of identified factors. Knowledge of a correlation force between factors is a basis to classifying factors into groups which are used for taking appropriate and optimal managing actions.

A solution presented in the article will be tested with reference to real data for a buying decision process in a company of a chemical branch using a computer support that simplifies computation processes.

Due to increasingly competitive markets, firms constantly search for new ways to lower their operational costs. One of many areas on which companies focus the most, is inventory management. More and more enterprises lay their faith in complex technical solutions which, in their opinion, will give them competitive advantage over their competitors. This paper is dedicated to demonstrate an innovative way to improve a traditional concept in inventory management (Vendor Managed Inventory), by using agent technologies. We propose an approach to utilize intelligent yet highly mobile software agents which, from an economical point of view, cold be both – cheaper and more effective than traditional information systems.

Nowadays, one of the fundamental challenges which exists in applying RFID technology is optimal managing of a large amount of data which are produced and gathered to exploit RFID. The notion leads to the decrease of information systems productivity and also reduces the information efficiency. Recently, some efforts have been made in order to introduce different data warehouse models for RFID to utilize information. Our research aims at introducing an improved model for RFID data warehouse in supply chain. The research has manifested a new framework for managing a large amount of RFID data. Firstly, an effective model of coding data path in supply chain is initiated and then in order to retrieve time according to coding path, a numerical model of XML environment has been used. Finally, by utilizing an index technique, RFID data is aggregated which caused a considerable reduction in the volume of data and also made a dramatic fall in the response time of queries on RFID data.