Management Intelligent Systems

Management intelligent systems (MIS) is a new paradigm integrating management with intelligent systems. What are the core components of MIS? What are intelligent systems for management? How can management integrate with intelligent systems? All these questions remain open in an MIS context. This article addresses these issues by examining MIS from a strategic perspective. More specifically, this article first examines management in information systems, management for intelligent systems and intelligent systems for management. Then this article provides a strategic model for MIS encompassing core components of MIS, through integrating main management functions with intelligent systems taking into account decision making of managers in organizations. The approach proposed in this article will facilitate the research and development of MIS, management and intelligent systems and information systems.

Many new technical systems are distributed systems that involve complex interaction between humans and machines, which notably reduces their usability. The properties of Agent Based Simulation make it especially suitable for simulating this kind of system. However, it is necessary to define new middleware solutions that allow the connection of simulation and visualization software. This paper describes the results achieved from a multiagent-based middleware for the behavior simulation and visualization of agents. The middleware modules presented in this study allow a complete integration of technologies for the development of Multiagent Systems and Agent Based Simulation, the construction of virtual organizations of agents, and the connection to external modules that represent the entities of the agents.

In this research a three staged hybrid genetic-fuzzy systems modeling methodology is developed and applied to an empirical data set in order to determine the hidden fuzzy if-then rules. The empirical data was collected in an earlier study in order to establish the relations among human capital, organizational support and innovativeness. The results demonstrate that the model based on the fuzzy if-then rules outperforms more traditional techniques. Furthermore, the proposed methodology is a valuable tool for successful knowledge management.

New communication technologies integrated into modern vehicles offer an opportunity for better assistance to people injured in traffic accidents. To improve the overall rescue process, a fast and accurate estimation of the severity of the accident represents a key point to help the emergency services to better determine the amount of required resources. This paper proposes a novel intelligent system which is able to automatically estimate the severity of traffic accidents based on the concept of datamining and knowledge inference.Our system considers the most relevant variables that can characterize the severity of the accidents (variables such as the vehicle speed, the type of vehicles involved, and the airbag status). Results show that data mining classification algorithms, combined with an adequate selection of relevant features and a prior division of collisions based on the impact direction, allows generating estimation models able to predict the severity of new accidents.

In the future, Ambient Intelligence (AmI) technology could assist people autonomously and interpret their intentions. Current technology can already be used to recognize the presence of a person in a private or public space and trigger an automatic response or reaction depending on the user activity. This work describes ContextCare, an extension for an video surveillance system in a health care scenarios based on activity recognition using sensor smartphones. Both systems are coordinated using ECA parading.

The increasing of public transportation or bike use became an important issue in addressing economic, energy and environmental challenges. With this regard one of the most main tasks is to find and analyze the factors influencing car dependency and the attitudes of people in terms of preferred transport mode. In this paper Fuzzy Cognitive Maps (FCM) are explored to show how travelers make decisions based on their knowledge of different transport modes properties. The results of this study will help transportation policy decision makers in better understanding of people’s needs and actualizing different policy formulations and implementations.

Many companies are making considerable investments in tracking technology, such as GPS and RFID. Although tracking technology captures vast amounts of information about the ongoing operations, companies struggle to effectively apply this captured information for enhancing their operational control. In order to contribute in solving this problem, this paper presents a generic system for enhancing operational control, which applies the captured information in a more effective way. The proposed system is based on the approach of intelligent products. The intelligent products represent physical objects, and are capable of autonomously performing some of the repetitive tasks required for operational control. The usefulness of the system is demonstrated by presenting the results of several applications of the system.

Due to the diversity and complexity of its projects, the Civil Engineering domain has historically encompassed very heterogeneous disciplines. From the beginning, any Civil Infrastructure project is systematically divided into smaller subprojects in order to reduce or isolate the overall complexity. However, as a parallel design work, these subdesigns may experience divergences which often lead to design conflicts when they are merged back to the global design. If a high-quality design is desired, these conflicts need to be detected and solved.We present a Multiagent system able to manage these design conflicts by detecting them, by assisting the engineers in the negotiation of solutions, and finally by learning how to solve future similar problems. The advantage of the system is that what is learned is not one individual’s knowledge but the project’s global distributed cognition.

The product life cycle of cars is becoming shorter and carmakers constantly introduce new or revised models in their lines, tailored to their customer needs. At the same time, new car model design decisions may have a substantial effect on the cost and revenue drivers. For example, although a new car model configuration with component commonality may lower manufacturing cost, it also hinders increased revenues that could have been achieved through product differentiation. This paper applies a state of the art, nature inspired approach to design car lines that optimize the degree of differentiation vs commonality among models in the line. Our swarm intelligence mechanism is applied to stated preference data derived from a large-scale conjoint experiment that measures consumer preferences for passenger cars in a sample of 1,164 individuals. Our approach provides interesting insights on how new and existing car models can be combined in a product line and suggests that differentiation among models within a product line elevates customer satisfaction.

A current strategy for improving sales as well as customer satisfaction in the e-commerce field is to provide product recommendation to users. The increasing acceptance of web recommender systems is mainly due to the advances achieved in the intensive research carried out for several years. However, in spite of these improvements, recommender systems still present some important drawbacks that prevent from satisfying entirely their users. In this work, a methodology that combines an association rule mining method with the definition of a domain-specific ontology is proposed in order make efficient book recommendations.

Due to the ever-growing amount of information available on Web shops, it has become increasingly difficult to get an overview of Web-based product information. There are clear indications that better search capabilities, such as the exploitation of annotated data, are needed to keep online shopping transparent for the user. For example, annotations can help present information from multiple sources in a uniform manner. This paper proposes an algorithm that can autonomously map heterogeneous product taxonomies forWeb shop data integration purposes. The proposed approach uses word sense disambiguation techniques, approximate lexical matching, and a mechanism that deals with composite categories. Our algorithm’s performance on three real-life datasets was compared favourably against two other state-of-the-art taxonomy mapping algorithms. The experiments show that our algorithm performs at least twice as good compared to the other algorithms w.r.t. precision and F-measure.

Wireless Sensor Networks is a key technology for gathering relevant information from different sources. In this sense, Multi-Agent Systems can facilitate the integration of heterogeneous sensor networks and expand the sensors’ capabilities changing their behavior dynamically and personalizing their reactions. Both Wireless Sensor Networks and Multi-Agent Systems can be successfully applied to different management scenarios, such as logistics, supply chain or production. The Hardware-Embedded Reactive Agents (HERA) platform allows developing applications where agents are directly embedded in heterogeneous wireless sensor nodes with reduced computational resources. This paper presents the reasoning mechanism included in HERA to provide HERA Agents with Case-Based Planning features that allow solving problems considering past experiences.

We propose a Semantic Web Service Discovery framework for finding semantically annotated Web services by using natural language processing techniques. The framework searches through a set of annotated Web services for matches with a user query, which consists of keywords, so that knowledge about semantic languages is not required. For matching keywords with Semantic Web service descriptions given in Web Service Modeling Ontology (WSMO), techniques like part-of-speech tagging, lemmatization, and word sense disambiguation are used. Three different matching algorithms are defined and evaluated for their ability to do exact matching and approximate matching between the user query and Web Service descriptions.

The potential fraud problems, international economic crisis and the crisis of confidence in markets have affected financial institutions, which have tried to maintain customer trust in many different ways. To maintain the trust level in financial institutions, the implementation of electronic banking for customers has been considered a successful strategy. However, the parameters that define user trust have not been analysed in detail due to the lack of experience and the recent use of e-banking. This paper aims to determine which variables are relevant to user trust by applying machine learning techniques as multi-objective genetic algorithms for the preparation of business strategies to improve confidence and profitability. The algorithms have been tuned following the indications given by experts and their results have been validated by them, setting a level of reliability. There is also a comparison among different fitness functions used in the evolution process that are able to rank the subset of variables encoded by the individuals.

Recommender systems have typically been used in tourism applications to filter out irrelevant information and to provide personalized recommendations to the users. With the advent of mobile devices and ubiquitous computing, RSs have begun to incorporate Location Based Services (LBS) into mobile tourism guides to provide users with interesting points of interest (POIs) according to their contextual information, mainly physical location. In this paper, we propose a context-aware system for mobile devices that incorporates some implicit contextual information that is scarcely used in the literature: the user’s speed and his trajectory. This system has been specifically crafted to assist travelling users by providing them with smart and personalized POIs along their route taking into account their current location and driving speed.

Mark

i

S is a software platform that uses intelligent systems for knowledge extraction from large marketing databases. Mark

i

S allows the marketing expert to model, learn and analyze marketing models using two different genetic algorithms for learning fuzzy systems with multi-item variables. Using these intelligent systems the expert can obtain more valuable information about the model and improve his/her decisions.MarkiS functioning is divided into five steps: the creation of the model, the presentation of the dataset, the edition of the variables and items of the model, the learning of the fuzzy rule-based systems that explain the model, and the analysis of these fuzzy systems.

Prosaico is an intelligent system used for personalised sport advising in sports facilities. Personalised sport advising consists of advising the users of a sports facility when choosing the most appropriate type of exercise given their condition and their circumstances. To introduce personalised sport advising in publicly-funded facilities we have to face several problems, for instance, low instructor/user ratio, consistency in the recommendations between instructors and/or the attendance of users with health issues. Our solution is to improve and automate the management of information in a way which enhances the quality of the advising service. In order to do that it is necessary to develop an intelligent system for the management of the sports facility, focusing on the automation of the generation of work plans. This intelligent system is defined as a constraint-satisfaction problem.

Nowadays the globalization process is a reality. As a result, the creation of new transnational corporations has increased. Any corporation of this magnitude requires a very complex staff hierarchy to properly function. As a result, typical problems in bureaucracy management have developed that usually occurred in highly developed economies. These problems have been researched at a theoretical level for a very long time. Certain results have been obtained by using enterprise resource planning systems that ensure transparency in any point of the process where the document processing flows. There are moments when a tight control must be exercised as in reducing personnel schema. This must be done without decreasing the organizational efficiency. Anyway, it is very hard to control a structure when one also uses a part of it to generate the executive reports necessary. In this paper a solution that may help to increase the control over bureaucracy is presented. The solution is used to generate software based on agents in order to find and extract the real organization structure by doing automatic analysis of all document workflows. Probably the most important advantage of the solution is that it is not under the control of the audited structure and consequently the results will not be modified because the lower staff will always try to protect themselves.

This article presents a proximity detection prototype that uses ZigBee technology developed using the agent’s platform PANGEA (Platform for Automatic coNstruction of orGanizations of intElligent Agents). PANGEA is an agent platform to develop open multiagent systems, specifically those including organizational aspects such as virtual agent organizations. The platform allows the complete management of organizations and offers tools to the end user. Due to the specific characteristics of this prototype, PANGEA is the perfect candidate to develop the prototype that will be included in the future in an integral system primarily oriented to facilitate the integration of people with disabilities into the workplace.

A new general multi-objective optimization heuristic algorithm, suitable for being applied to continuous multi-objective optimization problems is proposed. It deals with the problem at hand in a fast and efficient way. It combines ideas from different multi-objective and single-objective optimization evolutionary algorithms, although it also incorporates new devices which help to reduce the computational requirements, and also to improve the quality of the provided solutions. To show its applicability, a bi-objective competitive facility location and design problem is considered. This problem has been previously tackled through exact general methods, but they require high computational effort. A comprehensive computational study shows that the heuristic method is competitive, being able to reduce, in average, the computing time of the exact method by approximately 98%, and offering good quality in the final solutions.

The purpose of this paper is to determine the added value of pictorial stimulus-choice data in customer churn prediction. Using Random Forests and 5 times 2 fold cross-validation, this study analyzes how much pictorial stimulus – choice data and survey data increase the AUC of a churn model over and above administrative, operational and complaints data. The finding is that pictorial-stimulus choice data significantly increases AUC of models with administrative and operational data. The practical implication of this finding is that companies should start considering mining pictorial data from social media sites (e.g. Pinterest), in order to augment their internal customer database. This study is original in that it is the first that assesses the added value of pictorial stimulus-choice data in predictive models. This is important because more and more social media websites are focusing on pictures.

A multiple agent-based system is intended to capture complex behavioral patterns by utilizing a collection of autonomous computer systems (called agents) that can interact with decision makers and then learn, perform, and delegate tasks on their behalf. With its ability to handle a large amount of information from heterogeneous sources in dynamically changing environments, a multiple agent-based system can significantly improve the company’s business intelligence and operational efficiency. Though rarely used in demand planning, this paper proposes a multiple agent-based system for demand forecasting of ice cream which poses unique challenges due to volatility and seasonality of ice cream consumption. To validate the usefulness of the proposed system for demand planning, the forecasting outcomes of the proposed system was compared to those of traditional forecasting techniques. Our experiments showed that the proposed multiple agent-based system outperformed its traditional forecasting counterparts in terms of its accuracy and consistency.

Lack of adaptability within WorkFlow Management Systems (WFMS) has been early identified as one of their limitations. WFMS suffer from disadvantages such as not supporting the dynamic incorporation/modification of process models and poor adaptability of process models at runtime. The static workflow definition and its passive interpretation does not allow WFMS to demonstrate flexible behavior and to deal with real-life situations, such as fast changing customer requirements and enterprise goal shifts. In this work we propose the design and development of two features (manual intervention and statefulness), which are expected to tackle this limitation. Our work considers and agent-based environment for the WFMS implementation.

Automated analysis of the ever-increasing amount of reviews available through the Web can enable businesses to identify why people like or dislike (aspects of) products or brands, yet to this end, a reliable indication of the intended sentiment of reviews is of crucial importance. This sentiment is typically quantified in universal star ratings, which are not always available. We propose and compare the performance of several statistical methods of automatically classifying star ratings of reviews represented by means of a binary vector representation, with features signaling the presence of sentiment-carrying words. A nearest neighbor classifier maximizes recall, whereas a naïve Bayes classifier excels in terms of precision, accuracy, and the root mean squared error of the assigned number of stars.

Various machine learning techniques have been explored for credit scoring and management, but no consistent conclusions have been drawn on which method shows the best behaviour. This paper presents an experimental analysis involving five real-world databases with several credit scoring models, including logistic regression, neural networks, support vector machines, decision trees, rule induction algorithms, Bayesian models,

k

nearest neighbours decision rule, and classifier ensembles. Particularly, we analyse the performance of this set of algorithms by means of a non-parametric statistical test and two post-hoc procedures for making pairwise comparisons.

The abundance of available event data, originating from process-aware information systems, creates opportunities for enterprise risk management applications at the intersection of the business & management, artificial intelligence and knowledge representation research fields. This paper proposes a rule-based process mining approach for dealing with uncertainty and risk. The applicability of the approach is demonstrated using the updating and debugging process of a social security service provider.

Within the field of finance, Value-at-Risk (VaR) is a widely adopted tool to assess portfolio risk. When calculating VaR based on historical stock return data, the data could be sensitive to outliers caused by seldom occurring news events in the sampled period. Using a data set of news events, of which the irregular events are identified using a Poisson distribution, we research whether the VaR accuracy can be improved by considering news events as additional input in the calculation. Our experiments show that when a rare event occurs, removing the event-generated noise from the stock prices for a small, optimized time window can improve VaR predictions.

Adaptive scheduling strategies are about considering the state of computational grids to obtain efficient and reliable schedules and to prevent the system performance deterioration. In this work, emerging adaptive strategies in grid computing, namely Fuzzy Rule-Based Systems (FRBS) -based strategies and a new adaptive scheduling approach, gaussian scheduling founded on Gaussian Mixture Models (GMMs) are compared. Both types of strategies focus on modeling the state of resources and select the most convenient site of the grid at every scheduling step given the current conditions. FRBSs provide a fuzzy characterization of the grid state and the inference of a suitability index based on their own knowledge given in the form of fuzzy IF-THEN rules. Besides, a GMM can be trained to model a complex probability density distribution indicating the suitability of every site in the grid to be the target of the schedule with the current conditions of its resources. This way the GMM scheduler assigns a probability to every state of the site where a higher probability is associated to a higher suitability of selection. Simulations based on real grid facilities are conducted to test the FRBS and GMM-based models and results are analyzed in terms of accuracy and convergence behaviour of their associated learning processes.

Minimal Enclosing Ball (MEB) is a spherically shaped boundary around a normal dataset, it is used to separate this set from abnormal data. MEB has a limitation for dealing with a large dataset in which computational load drastically increases as training data size becomes large. To handle this problem in huge dataset used in different domains, we propose two approaches using Fuzzy C-mean clustering method. These approaches find the concentric balls with minimum volume of data description to reduce the chance of accepting abnormal data that contain most of the training samples. Our method uses a divide-and-conquer strategy; trains each decomposed sub-problems to get support vectors and retrains with the support vectors to find a global data description of a whole target class. Our study is experimented on speech information to eliminate all noise data and reducing time training. For this, the training data, learned by Support Vector Machines (SVMs), is partitioned among several data sources. Computation of such SVMs can be achieved by finding a core-set for the image of the data. Numerical experiments on some real-world datasets verify the usefulness of our approaches for data mining.

The algorithms are the most common form of problem solving in many science fields. Algorithms include parameters that need to be tuned with the objective of optimizing its processes. This work uses Hoeffding race techniques, with the objective to obtain the best initial combination of variables to use it as an input configuration. Hoeffding race quickly discard less promising candidates as soon as there are evidences enough to remove them from the competition. These evidences are based on the use of any statistical test that, at a given confidence level, would set a range of expected performance for configuration. All the experiment was applied in AdaNAS (Adaptive Neighboring-Ant Search), an algorithm that was developed to route queries through the Internet. Our results show that there is a significant gain in efficiency of the AdaNAS algorithm by using the simple, but powerful, technique of initial setting of parameters presented in this paper. In our experiments, the average efficiency was improved 50% by using a good initial configuration.

The huge amount of spam messages has limited the benefits introduced by e-mail communications. Therefore, spam filters are indispensable to fight against spam deliveries. However, the development of spam filters is very expensive whereas the usage of external filtering services can damage communications privacy. In such situation, we introduce an automatic procedure to integrate knowledge extracted by using rough-sets theory into spam filters to develop a low-cost filtering infrastructure.