Amitava Bagchi

Improving Linkage of Web Pages

Organizations maintain informational websites for wired devices. The information content of such websites tends to change slowly with time, so a steady pattern of usage is soon established. User preferences, both at the individual and at the aggregate level, can then be gauged from user access log files. We propose a heuristic scheme based on simulated annealing that makes use of the aggregate user preference data to re-link the pages to improve navigability. This scheme is also applicable to the initial design of websites for wireless devices. Using the aggregate user preference data obtained from a parallel wired website, and given an upper bound on the number of links per page, our methodology links the pages in the wireless website in a manner that is likely to enable the “typical” wireless user to navigate the site efficiently. Later, when a log file for the wireless website becomes available, the same approach can be used to refine the design further.

The multiple resource constrained project scheduling problem: A breadth-first approach

We describe a simple breadth-first tree search scheme for minimizing the makespan of a project consisting of a partially ordered network of activities under multiple resource constraints. The method compares quite favourably with existing techniques that employ depth-first or best-first search; in particular, it is able to solve optimally on a Pentium PC running SCO UNIX the entire set of 680 benchmark problems (First Lot: 480, Second Lot: 200) generated by Kolisch et al., 1995. The new algorithm has also been checked out experimentally on additional random test problems at graded levels of difficulty that were generated using two parameters: the threshold, which determined the predecessors of an activity, and the total resource availability of each resource type. The breadth-first scheme can be modified quite readily to do best-first search or to minimize measures other than makespan such as mean flow time and maximum tardiness.

Graph search methods for non-order-preserving evaluation functions: applications to job sequencing problems

Abstract Graph search with A ∗ is frequently faster than tree search. But A ∗ graph search operates correctly only when the evaluation function is order-preserving. In the non-orderpreserving case, no paths can be discarded and the entire explicit graph must be stored in memory. Such situations arise in one-machine minimum penalty job sequencing problems when setup times are sequence dependent. GREC, the unlimited memory version of a memory-constrained search algorithm of the authors called MREC, has a clear advantage over A ∗ in that it is able to find optimal solutions to such problems. At the same time, it is as efficient as A ∗ in solving graph search problems with order-preserving evaluation functions. Experimental results indicate that in the non-order-preserving case, GREC is faster than both best-first and depth-first tree search, and can solve problem instances of larger size than best-first tree search.

On Detecting Data Flow Errors in Workflows

When designing a business workflow, it is customary practice to create the control flow structure first and to ensure its correctness. Information about the flow of data is introduced subsequently into the workflow and its correctness is independently verified. Improper specification of data requirements of tasks and XOR splits can cause problems such as wrong branching at XOR splits and the failure of tasks to execute. Here we present a graph traversal algorithm called GTforDF for detecting data flow errors in both nested and unstructured workflows, and illustrate its operation on realistic examples. Two of these have interconnected loops and are free of control flow errors, and the third one is an unstructured loop-free workflow. Our approach extends and generalizes data flow verification methods that have been recently proposed. It also makes use of the concept of corresponding pairs lately introduced in control flow verification. It thus has the potential for development into a unified algorithmic procedure for the concurrent detection of control flow and data flow errors. The correctness of the algorithm has been proved theoretically. It has also been tested experimentally on many examples.

Detecting Data Flow Errors in Workflows: A Systematic Graph Traversal Approach

When designing a workflow, it is customary practice to create the control flow structure first and to ensure its correctness. Information about the flow of data is introduced subsequently into the workflow and its correctness is independently verified. Improper specification of data requirements of tasks and XOR splits can cause problems such as wrong branching at XOR splits and the failure of tasks to execute. Here we present a graph traversal algorithm called GTforDF for detecting data flow errors in a workflow that is free of control flow errors, and illustrate its operation on two realistic workflows with interconnected loops. Our approach extends and generalizes data flow verification methods that have been recently proposed. It also makes use of the concept of corresponding pairs lately introduced in control-flow verification. It thus has the potential for development into a unified algorithmic procedure for the concurrent detection of control flow and data flow errors.

A faster alternative to SSS * with extension to variable memory

The two best known and most frequently referenced minimax search methods for game trees are Alpha-Beta [2] and SSS* [6]. The algorithms are quite dissimilar in structure and properties. Alpha-Beta is a depth-first recursive procedure that needs little memory to execute. Although it evaluates more terminal nodes than SSS*, it generally runs considerably faster owing to its low overhead. In contrast, SSS* is a non-recursive procedure similar in many ways to the best-first search algorithm A*. Although SSS* is superior to Alpha-Beta in number of terminals evaluated, SSS* is seldom used in practice. This is because the improved pruning power of SSS” is more than offset by (see [4]) (a> the enormous storage requirement for the global OPEN list, which is brd/‘l units for a (6, d) uniform tree; (b) the overhead of maintaining OPEN sorted on h-values; (c) the overhead of occasionally purging nodes from OPEN belonging to provably suboptimal solution trees.

Searching graphs with A*: applications to job sequencing

The best-first search algorithm A* allows search graphs that are trees, directed acyclic graphs or directed graphs with cycles. In real life applications of A* the search graph is generally implemented as a tree. It is shown here that for certain well known one-machine job sequencing problems that arise in job shops, graph search is much faster than best-first tree search when problem instances are of small and medium size. Moreover, graph search uses less memory and so is able to solve larger problems. Depth-first search needs little memory, and is therefore capable in principle of solving problems of arbitrary size, but is slower than graph search by orders of magnitude for the examples that were studied.

Average-case analysis of heuristic search in tree-like networks

A search graph has the form of an m-ary tree with bi-directional arcs of unit cost. There is a goal node at a distance N from the root, and there may be other goal nodes at distances ≥ N from the root. It is assumed that the heuristic estimates of nongoal nodes, after being appropriately normalized, are independent and identically distributed random variables. The heuristic is not required to be admissible. Under what conditions is the expected number of node expansions E(Z) polynomial in N? Earlier efforts by Pearl and others at answering this question have considered search trees with only one goal node. An attempt is made here to develop a general and unified method of analysis applicable to situations with more than one goal node. It is shown that, for most probability distributions on the heuristic estimates, E(Z) is exponential in N; the one major exception being the case when the number of goal nodes is polynomial in N and the normalizing function for the error is logarithmic. Pearl’s contention that the average-case analysis of weighted heuristic search is not too attractive is also verified. It is hoped that the general approach described here will encourage similar studies on search graphs other than trees.

Job sequencing with quadratic penalties: An A*-based graph search approach

The authors tried to establish that A* is much better than traditional branch-and-bound procedures at solving certain types of minimum-penalty job-sequencing problems. Townsends algorithm (1976) for minimum-penalty job sequencing on one machine with quadratic penalties was implemented both as tree search and as a graph-based A* formation that uses Townsends lower bounds at nodes as heuristic estimates. The graph implementation took much less time to run, and problem instances of much larger size could be solved. The approach seems to be particularly suitable for sequencing and other optimization problems where lower bounds at nodes can be determined without excessive computational effort.<<ETX>>

Tuning the quality parameters of a firewall to maximize net benefit

A firewall protects the informational assets of an organization from intruders. Incoming message packets are filtered by the firewall before being forwarded to their destinations inside the organization. In the process, a fraction q1 of benign (i.e., desirable or harmless) packets and a fraction q2 of intrusive (i.e., undesirable or harmful) packets get blocked. Ideally, we should have q1 = 0 and q2 = 1, but in practice q1 and q2 are functionally related. Since the firewall has a non-zero service time, it also causes a delay because packets get queued for service. Thus by using a firewall an organization incurs a cost, but there is also a corresponding benefit. This study considers the simple case when a single firewall is in use. We do an economic analysis and derive a mathematical expression for the net benefit. We then maximize it by tuning the quality parameters q1 and q2 appropriately.