There are several approaches that model information extraction as a token classification task, using various tagging strategies to combine multiple tokens. We describe the tagging strategies that can be found in the literature and evaluate their relative performances. We also introduce a new strategy, called Begin/After tagging or BIA, and show that it is competitive to the best other strategies.

We explore methods to implement Conditional Random Fields (CRF) for Chinese Part-Of-Speech Tagging. We focus on the task of POS tagging without pre-segmentation, and propose a hierarchical Conditional Random Fields to do Segmenta-tion and POS Tagging at one time step. Experiments are going to be done for my method to compare it with existent methods on this task.

Automatic image tagging is becoming increasingly important to organize large amounts of image data. To identify concepts in images, these tagging systems rely on large sets of annotated image training sets. In this work we analyze image sets taken from online community-driven image databases, such as Flickr, for use in concept identification. Realworld performance is measured using our flexible...

OBJECTIVE This preliminarily study was designed to determine and to compare the efficacy of two commercially available barium-based fecal tagging agents for CT colonography (CTC) (high-density [40% w/v] and low-density [4.6% w/v] barium suspensions) in a population in Korea. MATERIALS AND METHODS In a population with an identified with an average-risk for colorectal cancer, 15 adults were adm...

Although epitope tagging has been widely used for analyzing protein function in many organisms, there are few genetic tools for epitope tagging in Tetrahymena. In this study, we describe several C-terminal epitope tagging modules that can be used to express tagged proteins in Tetrahymena cells by both plasmid- and PCR-based strategies.

We compare heavy quark tagging and anti-tagging efficiencies for vertex detectors with different inner raddi using the topological vertex technique developed at the SLC/SLD experiment. Charm tagging benefits by going to very small inner radii. Talk presented at 5th International Linear Collider Workshop (LCWS 2000) Fermilab, Batavia, Illinois 24-28 Oct 2000 ∗Work supported by Department of Ener...

BACKGROUND Adenosine stress perfusion is very sensitive for detection of coronary artery disease (CAD), and yields good specificity. Standard adenosine cine imaging lacks high sensitivity, but is very specific. Myocardial tagging improves detection of wall motion abnormalities (WMAs). Perfusion and tagging cardiovascular magnetic resonance (CMR) both benefit from high field imaging (improved co...

This paper focuses on the task of tagging text with their parts of speech. The methodology chosen for this task is the Maximum Entropy based Model and although complex will only be explained briefly. More importantly, the focus will center on the differences in performance of the maxent model with varying feature sets compared to the baseline model. One problem highlighted in part-of-speech tag...

This paper separates conditional parameter estimation, which consistently raises test set accuracy on statistical NLP tasks, from conditional model structures, such as the conditional Markov model used for maximum-entropy tagging, which tend to lower accuracy. Error analysis on the POS tagging task shows that the actual tagging errors made by the conditionally structured model derive not only f...