We investigate a number of simple methods for improving the word-alignment accuracy of IBM Model 1. We demonstrate reduction in alignment error rate of approximately 30% resulting from (1) giving extra weight to the probability of alignment to the null word, (2) smoothing probability estimates for rare words, and (3) using a simple heuristic estimation method to initialize, or replace, EM train...

The connections between the E(5)−models (the original E(5) using an infinite square well, E(5)−β 4, E(5)−β 6 and E(5)−β 8), based on particular solutions of the geometrical Bohr Hamiltonian with γ-unstable potentials, and the interacting boson model (IBM) are explored. For that purpose, the general IBM Hamiltonian for the U(5)− O(6) transition line is used and a numerical fit to the different E...

As companies race to transform their businesses into e-businesses, they are discovering that the transformation process is not always straightforward. The IBM Application Framework for e-business (the Framework) is a means for achieving business transformation and a foundation framework for developing and extending e-business processes and applications. This paper describes the Framework, showi...

Model transformation is at the core of the OMG’s Model Driven ArchitectureTM (MDA). The current submission process on the OMG MOF 2.0 Query/Views/Transformation (QVT) Request for Proposals (RFP) [1] elicited eight initial submissions and currently, five revised submissions are still competing. IBM has formulated, in response to the eight first submissions, a set of recommendations for the final...

Two decades after their invention, the IBM word-based translation models, widely available in the GIZA++ toolkit, remain the dominant approach to word alignment and an integral part of many statistical translation systems. Although many models have surpassed them in accuracy, none have supplanted them in practice. In this paper, we propose a simple extension to the IBM models: an `0 prior to en...

Lake Tanganyika has a very rich pelagic fishery, which consists mainly of endemic clupeid species (Stolothrissa tanganicae and Limnothrissa miodon) and their predators, endemic centropomids (Lates spp.). Stolothrissa tanganicae is the most important species in the fishery. Coenen (1995) estimated a total overall fish catch for Lake Tanganyika from 1992 to be 167,000 metric tones, of which 80,00...

We describe a Bayesian projection model to produce country-specific projections of the total fertility rate (TFR) for all countries. The model decomposes the evolution of TFR into three phases: pre-transition high fertility, the fertility transition, and post-transition low fertility. The model for the fertility decline builds on the United Nations Population Division's current deterministic pr...