TEG—a hybrid approach to information extraction

This paper describes a hybrid statistical and knowledge-based information extraction model, able to extract entities and relations at the sentence level. The model attempts to retain and improve the high accuracy levels of knowledge-based systems while drastically reducing the amount of manual labour by relying on statistics drawn from a training corpus. The implementation of the model, called TEG (trainable extraction grammar), can be adapted to any IE domain by writing a suitable set of rules in a SCFG (stochastic context-free grammar)-based extraction language and training them using an annotated corpus. The system does not contain any purely linguistic components, such as PoS tagger or shallow parser, but allows to using external linguistic components if necessary. We demonstrate the performance of the system on several named entity extraction and relation extraction tasks. The experiments show that our hybrid approach outperforms both purely statistical and purely knowledge-based systems, while requiring orders of magnitude less manual rule writing and smaller amounts of training data. We also demonstrate the robustness of our system under conditions of poor training-data quality. Ronen Feldman is a senior lecturer at the Mathematics and Computer Science Department of Bar-Ilan University in Israel, and the Director of the Data Mining Laboratory. He received his B.Sc. in Math, Physics and Computer Science from the Hebrew University, M.Sc. in Computer Science from Bar-Ilan University, and his Ph.D. in Computer Science from Cornell University in NY. He was an Adjunct Professor at NYU Stern Business School. He is the founder of ClearForest Corporation, a Boston based company specializing in development of text mining tools and applications. He has given more than 30 tutorials on next mining and information extraction and authored numerous papers on these topics. He is currently finishing his book “The Text Mining Handbook” to the published by Cambridge University Press. Benjamin Rosenfeld is a research scientist at ClearForest Corporation. He received his B.Sc. in Mathematics and Computer Science from Bar-Ilan University. He is the co-inventor of the DIAL information extraction language. Moshe Fresko is finalizing his Ph.D. in Computer Science Department at Bar-Ilan University in Israel. He received his B.Sc. in Computer Engineering from Bogazici University, Istanbul/Turkey on 1991, and M.Sc. on 1994. He is also an adjunct lecturer at the Computer Science Department of Bar-Ilan University and functions as the Information-Extraction Group Leader in the Data Mining Laboratory.     

A method of detecting and tracking irises and eyelids in video

We locate the eye corners, eyelids, and irises in every frame of an image sequence, and analyze the movements of the irises and eyelids to determine changes in gaze direction and blinking, respectively. Using simple models for the motions of the head and eyes, we determine the head-independent motions of the irises and eyelids by stabilizing for the head motion. The head-independent motions of the irises can be used to determine behaviors like saccades and smooth pursuit. Tracking the upper eyelid and using the distance between its apex and the center of the iris, we detect instances of eye closure during blinking. In experiments on two short image sequences, in one of which the subject was wearing glasses, we successfully located the irises in every frame in which the eyes were fully or partially open, and successfully located the eyelids 80% of the time. When motion information in the form of normal flow was used, the irises were successfully tracked in every frame in which the eyes were fully or partially open, and the eyelids were successfully located and tracked 90% of the time.     

Observation of transient disorder during myosin subfragment-1 binding to actin by stopped-flow fluorescence and millisecond time resolution electron cryomicroscopy: evidence that the start of the crossbridge power stroke in muscle has variable geometry

The mechanism of binding of myosin subfragment-1 (S1) to actin in the absence of nucleotides was studied by a combination of stopped-flow fluorescence and ms time resolution electron microscopy. The fluorescence data were obtained by using pyrene-labeled actin and exhibit a lag phase. This demonstrates the presence of a transient intermediate after the collision complex and before the formation of the stable "rigor" complex. The transient intermediate predominates 2-15 ms after mixing, whereas the rigor complex predominates at time >50 ms. Electron microscopy of acto-S1 frozen 10 ms after mixing revealed disordered binding. Acto-S1 frozen at 50 ms or longer showed the "arrowhead" appearance characteristic of rigor. The most likely explanation of the disorder of the transient intermediate is that the binding is through one or more flexible loops on the surfaces of the proteins. The transition from disordered to ordered binding is likely to be part of the force-generating step in muscle.     

Insulation lead failure: is it a matter of insulation coating, venous approach, or both?

Gag gene mutants of human immunodeficiency virus type 1 (HIV-1) were analyzed for their potentials of inhibiting the replication of wild-type (wt) HIV-2, the second AIDS virus, in a single-round of viral replication. Of twenty-two HIV-1 gag mutants examined, seven were found to efficiently interfere with the replication of wt HIV-2. Some mutants, which can suppress the replication of wt HIV-1, did not show this inhibitory effect. These mutants were defective at the late phase of viral replication. A mutant designated NL-C1a was demonstrated to be very effective against the replication of HIV-1 and HIV-2 in monocytic cells as well as in lymphocytic cells.     

Insulin-like growth factor binding protein production in human follicular thyroid carcinoma cells

OBJECTIVE: The study investigated the value of using national or regional data bases to examine care in a specific hospital. DATA SOURCES: The following data sources were included: (1) the results of the 1992 HCFA analysis of the index hospital for patients hospitalized in fiscal year 1990; (2) the 1989 Medicare Provider Analysis and Review (MEDPAR) file; and (3) clinical information from bypass surgery patients in Wisconsin and from the index hospital. PRINCIPAL FINDINGS: The assessment of the mortality rates in the index hospital for all conditions combined and for CABG patients differed depending on what data base was used and how the data were analysed. The national data were most useful in establishing that the coding practices for all patients and the mortality rate for intra-aortic balloon patients differed between the index hospital and other hospitals. The regional clinical data base for bypass surgery patients was used to establish that the high mortality rates for intra-aortic balloon patients were due to patient selection. CONCLUSIONS: National claims data must be analysed carefully before applying results to an individual hospital. Even a careful analysis is more for raising questions about care at a specific hospital rather than for reaching definitive conclusions.     

Agonists and antagonists for lipopolysaccharide-induced cytokines

Agonistic and antagonistic properties of LPS and partial structures in the induction of cytokines are reviewed. Studies on structure-activity relationships of LPS and lipid A with human mononuclear cells reveal that S- and notably R-form LPS are very potent cytokine inducers. Synthetic E. coli lipid A is somewhat less active, whereas synthetic S. minnesota-type lipid A is significantly less active. Pentaacylated forms of lipid A are less potent than hexaacylated forms, and tetraacylated synthetic precursor Ia and bisacylated disaccharides and monosaccharides are completely inactive, indicating that a structure-dependent hierarchy of LPS and lipid A partial structures determines the monokine-inducing capacity in human mononuclear cells. Precursor Ia is a potent LPS antagonist. The mechanism of its inhibitory activity is shown to be due to competitive binding to cellular binding sites (receptors). Proinflammatory and antiinflammatory cytokines, receptor antagonists, and soluble cytokine receptors influence the cytokine-inducing activity of LPS, suggesting a complex regulatory network.