Conversational Collaborative Recommendation – An Experimental Analysis

Traditionally, collaborative recommender systems have been based on a single-shot model of recommendation where a single set of recommendations is generated based on a user’s (past) stored preferences. However, content-based recommender system research has begun to look towards more conversational models of recommendation, where the user is actively engaged in directing search at recommendation time. Such interactions can range from high-level dialogues with the user, possibly in natural language, to more simple interactions where the user is, for example, asked to indicate a preference for one of k suggested items. Importantly, the feedback attained from these interactions can help to differentiate between the user’s long-term stored preferences, and her current (short-term) requirements, which may be quite different. We argue that such interactions can also be beneficial to collaborative recommendation and provide experimental evidence to support this claim.     

Data mining: data analysis on a grand scale?

Modern data mining has evolved largely as a result of efforts by computer scientists to address the needs of 'data owners' in extracting useful information from massive observational data sets. Because of this historical context, data mining to date has largely focused on computational and algorithmic issues rather than the more traditional statistical aspects of data analysis. This paper provides a brief review of the origins of data mining as well as discussing some of the primary themes in current research in data mining, including scalable algorithms for massive data sets, discovering novel patterns in data, and analysis of text, web, and related multimedia data sets.     

Structured writing about stressful events: Exploring potential psychological mediators of positive health effects.

In a previous study, the authors found that structured writing about stressful events improved symptomatology in 112 patients with rheumatoid arthritis and asthma relative to patients who did not write (J. Smyth, A. Stone, A. Hurewitz, & A. Kaell, 1999). However, little is currently known about the pathways from the intervention to alterations in outcomes. In addition to measuring symptom outcomes after the intervention in the previous study, the authors monitored perceived stress, quality of sleep, affect, substance use, and medication use on a momentary basis for the 7 days prior to writing, during the 3 intervention days, and for the 14 days following the intervention (N?=?105). These variables were tested in a secondary data analysis to determine whether they mediated the effects observed in the J. Smyth, A. Stone, et al. study. No evidence was found supporting mediation, and the mechanism underlying structured writing about stressful events remains unknown. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Evidence that activation of Src family kinase is not required for fertilization-associated [Ca2+]i oscillations in mouse eggs

Recent evidence in marine invertebrate, frog, and zebrafish eggs suggests the involvement of a Src family kinase (SFK) in fertilization-induced Ca2+ release. In the present study, we have investigated whether activation of an SFK is required for initiation of intracellular Ca2+ ([Ca2+]i) oscillations in mouse fertilization. We detected a Hck-like protein and tyrosine-phosphorylated proteins in soluble and insoluble sperm fractions, respectively. However, the presence of these proteins did not correspond to the active fractions of porcine sperm extracts (pSE). Moreover, [Ca2+]i oscillations induced by pSE in mouse eggs were unaltered by pre-incubation of pSE with specific SFK inhibitors such as 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazol[3,4-d]-pyrimidine (PP2) or lavendustin A, despite the fact that the inhibitors were shown to be active both in vivo and in vitro. Another SFK inhibitor, peptide A, blocked oscillations when incubated with pSE prior to injection into eggs, but this inhibition required more than ten times the concentration reportedly required to inhibit SFK activity. In addition, pre-injection or pre-incubation of eggs with these inhibitors did not affect the ability of pSE to trigger [Ca2+]i oscillations in mouse eggs. Microinjection of a recombinant c-Src protein or mRNAs encoding constitutively active Src proteins did not induce [Ca2+]i release. Finally, when sperm and eggs, both of which were pre-treated with PP2, were fertilized, [Ca2+]i oscillations occurred normally. We can therefore conclude that activation of an SFK is neither necessary nor sufficient for triggering fertilization-induced [Ca2+]i oscillations.     

Double-codified gold nanolabels for enhanced immunoanalysis

A novel double-codified nanolabel (DC-AuNP) based on gold nanoparticle (AuNP) modified with anti-human IgG peroxidase (HRP)-conjugated antibody is reported. It represents a simple assay that allows enhanced spectrophotometric and electrochemical detection of antigen human IgG as a model protein. The method takes advantage of two properties of the DC-AuNP label: first, the HRP label activity toward the OPD chromogen that can be related to the analyte concentration and measured spectrophotometrically; second, the intrinsic electrochemical properties of the gold nanoparticle labels that being proportional to the protein concentration can be directly quantified by stripping voltammetry. Beside these two main direct determinations of human IgG, a secondary indirect detection was also applicable to this system, exploiting the high molar absorptivity of gold colloids, by which, the color intensity of their solution was proportional to the concentration of the antigen used in the assay. Paramagnetic beads were used as supporting material to immobilize the sandwich-type immunocomplexes resulting in incubation and washing times shorter than those typically needed in classical ELISA tests by means of a rapid magnetic separation of the unbound components. A built-in magnet graphite-epoxy-composite electrode allowed a sensibly enhanced adsorption and electrochemical quantification of the specifically captured AuNPs. The used DC-AuNP label showed an excellent specificity/selectivity, as a matter of fact using a different antigen (goat IgG) a minimal nonspecific electrochemical or spectrophotometric signal was measured. The detection limits for this novel double-codified nanoparticle-based assay were 52 and 260 pg of human IgG/mL for the spectrophotometric (HRP-based) and electrochemical (AuNP-based) detections, respectively, much lower than those typically achieved by ELISA tests. The developed label and method is versatile, offers enhanced performances, and can be easily extended to other protein detection schemes as well as in DNA analysis.     

Training Neural Networks by Adaptive Random Search Techniques

A relatively simple stochastic optimization procedure based on the adaptive random search algorithm is presented to train artificial neural networks of the type encountered in applied mechanics applications. After discussing some essential features of the algorithm that influence its search efficiency, a procedure is outlined for replacing the back-propagation training approach by the new method in order to train networks involving high-dimensional parameter vectors. The method is successfully used in conjunction with a multilayer network involving a parameter vector of very high dimension. It is shown that the adaptive random search approach shifts the training effort from the user to the computer by exchanging additional computer search effort for easier training tasks on the part of the user. Extensive simulation studies are presented to provide statistically significant results related to the characteristics of the stochastic training approach. Guidelines are provided for applying the method to generic neural network training episodes.     

Statistical topic models for multi-label document classification

Machine learning approaches to multi-label document classification have to date largely relied on discriminative modeling techniques such as support vector machines. A?drawback of these approaches is that performance rapidly drops off as the total number of labels and the number of labels per document increase. This problem is amplified when the label frequencies exhibit the type of highly skewed distributions that are often observed in real-world datasets. In this paper we investigate a class of generative statistical topic models for multi-label documents that associate individual word tokens with different labels. We investigate the advantages of this approach relative to discriminative models, particularly with respect to classification problems involving large numbers of relatively rare labels. We compare the performance of generative and discriminative approaches on document labeling tasks ranging from datasets with several thousand labels to datasets with tens of labels. The experimental results indicate that probabilistic generative models can achieve competitive multi-label classification performance compared to discriminative methods, and have advantages for datasets with many labels and skewed label frequencies.     

Measures for investigating the contextual modulation of information transmission

The aim of this paper is to show how information theoretic measures can be used to analyse and interpret the results of psychophysical experiments designed to search for conditions under which information from one source modulates the transmission of information from another source. We therefore use measures of mutual and conditional information to analyse systems with two inputs. The information transmitted by such a system can be split into three components depending on whether it is shared between the two inputs or is specific to each. We are concerned here with distinguishing systems that use one input to modulate transmission of information about the other from systems that simply add both inputs, and show how the three components provide evidence for distinguishing between additive and modulatory effects. We also report numerical simulations of the sampling biasses and variances of these measures as a function of the sample size and propose minimum sample sizes that should be used to overcome the bias.     

SedLog: A shareware program for drawing graphic logs and log data manipulation

SedLog is a free multi-platform software package for creating graphic sediment logs providing an intuitive graphical user interface. The graphic sediment logs generated by SedLog can be exported as PDF, Scalable Vector Graphics (SVG), or JPEG for use by other drawing applications or for publications. Log data can be imported and exported in Comma Separated Values (CSV) format. The logs can also be printed to any paper size the user wants. Zoom In, Zoom Out, Fit page, Fit Height and Fit Width facilities are also provided to enable the user to customise the workspace size.     

A near‐real time automatic MODIS data processing system

The Moderate Resolution Imaging Spectroradiometer (MODIS) on‐board the Aqua and Terra platforms was designed to improve understanding of global dynamics and processes occurring on the land, in the oceans, and in the lower atmosphere. The UK Dundee Satellite Receiving Station has two X‐band receiving systems capable of capturing direct broadcast data from these spacecraft with a range covering the European shelf‐areas, north‐east Atlantic ocean and the western Mediterranean Sea. Raw data are transferred to the Plymouth Marine Laboratory (PML) and processed in near‐real time into ocean colour and sea‐surface temperature products for the academic community. Data can be used operationally and are made available through the web within 1.5 hours of the satellite overpass time. To our knowledge this is the only such developed system in Europe producing near‐real time MODIS ocean colour products.