Itemset trees for targeted association querying

Association mining techniques search for groups of frequently co-occurring items in a market-basket type of data and turn these groups into business-oriented rules. Previous research has focused predominantly on how to obtain exhaustive lists of such associations. However, users often prefer a quick response to targeted queries. For instance, they may want to learn about the buying habits of customers that frequently purchase cereals and fruits. To expedite the processing of such queries, we propose an approach that converts the market-basket database into an itemset tree. Experiments indicate that the targeted queries are answered in a time that is roughly linear in the number of market baskets, N. Also, the construction of the itemset tree has O(N) space and time requirements. Some useful theoretical properties are proven.     

Predicting Missing Items in Shopping Carts

Existing research in association mining has focused mainly on how to expedite the search for frequently co-occurring groups of items in “shopping cart” type of transactions; less attention has been paid to methods that exploit these “frequent itemsets” for prediction purposes. This paper contributes to the latter task by proposing a technique that uses partial information about the contents of a shopping cart for the prediction of what else the customer is likely to buy. Using the recently proposed data structure of itemset trees (IT-trees), we obtain, in a computationally efficient manner, all rules whose antecedents contain at least one item from the incomplete shopping cart. Then, we combine these rules by uncertainty processing techniques, including the classical Bayesian decision theory and a new algorithm based on the Dempster-Shafer (DS) theory of evidence combination.     

A MIP based heuristic for capacitated MRP systems

Although Material Requirements Planning (MRP) is the most widely used production planning tool in today’s manufacturing companies, its inability to perform an exhaustive capacity planning, lack of a comprehensive and integrated shop floor extension and using constant and inflated lead times necessitate intelligent methods for developing cost effective production plans. A single optimization model might be employed to overcome these limitations, but it would be intractable to use it in large manufacturing systems. Hence, in this paper, we propose a heuristic method called Capacity Allocater and Scheduler, CAS, to eliminate drawbacks of MRP systems and provide solutions for large-scale instances. The CAS procedure, based on iteratively solving relaxed Mixed Integer Programming (MIP) models, is built on a lot sizing and scheduling framework, which considers both supply alternatives and lot size restrictions simultaneously. Finally, we give a detailed numerical example to demonstrate how CAS may be used in practice, and provide our concluding remarks.     

Comparison of OR and AI methods in discrete manufacturing using fuzzy logic

Since 1950s the techniques of Operations Research (OR) and Optimization have been utilized to increase the efficiency of the production systems. With the widespread use of computers, it has even become easier to deal with industrial problems. However the complexity of the problems still reveals the difficulty in providing solutions. The use of artificial intelligence (AI) seems to attract the attention of the researcher to overcome to the difficulties. This has already been realized with several successful applications. In this study, the use of AI and OR techniques is compared using fuzzy logic. The progress of manufacturing systems, characteristics of production processes, system managements and system behavior are taken into account. The study is focussed on only discrete manufacturing.     

Combining Subclassifiers in Text Categorization: A DST-Based Solution and a Case Study

Text categorization systems often use machine learning techniques to induce document classifiers from preclassified examples. The fact that each example document belongs to many classes often leads to very high computational costs that sometimes grow exponentially in the number of features. Seeking to reduce these costs, we explored the possibility of running a "baseline induction algorithm" separately for subsets of features, obtaining a set of classifiers to be combined. For the specific case of classifiers that return not only class labels but also confidences in these labels, we investigate here a few alternative fusion techniques, including our own mechanism that was inspired by the Dempster-Shafer Theory. The paper describes the algorithm and, in our specific case study, compares its performance to that of more traditional mechanisms.     

Reliable video broadcasts via protected Steiner trees

The introduction of the latest generation of ROADMs in the communication long-haul transport networks allows network planners to consider some new cost-effective design alternatives. Specifically, for video broadcast services ROADM wavelength drop-and-continue technology enables simple wavelength connections at each node via a tree-like topology, and the intelligent control plane permits the use of various shared protection schemes (with failure restoration switching times comparable to SONET BLSR). In this article we formulate models for reliable TV/video broadcast. We consider the network topologies based on minimum spanning trees. The objective is to minimize the total network cost while ensuring that the broadcast, originating in one (or two) source node(s), is delivered to a set of destination nodes and the network will tolerate at least one single link failure. The resulting protected tree networks are illustrated, and the cost of protection strategies is analyzed.     

Reliability Analysis for Integrated Networks with Application to Burst Switching

System performance and reliability are jointly assessed for highly reliable telecommunication or computer networks. The underlying model assumes that at most a small number of components can be "down" at a time, and that the average repair/replacement time of a failed component is small when compared to the average uptimes of network components. At steady state the system is assumed to follow a regenerative stochastic process. This methodology is used to evaluate highly distributed voice/data integrated networks. The performance measure selected for this application is the traffic loss rate.     

Quick and simple procedures to assess software reliability and facilitate project management

A software reliability model is considered that is easy to implement, use, and interpret. The model works extremely well in the latter stages of testing. A complete history of failures does not need to be stored in a data base or maintained. This reduces the cost of assessing software reliability. Furthermore, it is possible to use the model to estimate software reliability when failure statistics have not been extensively collected. Various estimation procedures are discussed that can aid in project planning. The use of these estimation procedures is illustrated through two sets of actual failure data.     

Pragmatic Testing Protocols to Measure Software Reliability

This paper investigates four different test protocols based upon the Jelinski-Moranda model of software reliability growth. The main task is to estimate the number of remaining errors in a system at the time of system release. These four protocols are: 1) testing is discontinued after some number of errors has been observed; 2) testing is discontinued at the end of a prescribed period of time; 3) testing is discontinued when a specified amount of time is observed to be completely error free; 4) error checking and debugging is done only by a prespecified time schedule. Two examples with actual failure data illustrate the estimation procedures under different protocols.