Improving document clustering in a learned concept space

Most document clustering algorithms operate in a high dimensional bag-of-words space. The inherent presence of noise in such representation obviously degrades the performance of most of these approaches. In this paper we investigate an unsupervised dimensionality reduction technique for document clustering. This technique is based upon the assumption that terms co-occurring in the same context with the same frequencies are semantically related. On the basis of this assumption we first find term clusters using a classification version of the EM algorithm. Documents are then represented in the space of these term clusters and a multinomial mixture model (MM) is used to build document clusters. We empirically show on four document collections, Reuters-21578, Reuters RCV2-French, 20Newsgroups and WebKB, that this new text representation noticeably increases the performance of the MM model. By relating the proposed approach to the Probabilistic Latent Semantic Analysis (PLSA) model we further propose an extension of the latter in which an extra latent variable allows the model to co-cluster documents and terms simultaneously. We show on these four datasets that the proposed extended version of the PLSA model produces statistically significant improvements with respect to two clustering measures over all variants of the original PLSA and the MM models.     

Evidence for data missing at random in youth physical activity monitoring research

This study examined whether or not activity monitor data collected as part of a typical 7-day physical activity (PA) measurement protocol can be expected to be missing at random. A total of 315 participants (9–18 years) each wore a SenseWear Armband monitor for 7 consecutive days. Participants were classified as “compliant” (86 boys and 124 girls) if they had recorded accelerometer data during 70% or more of the predefined awake time (7 AM–10 PM) on four different days; and “non-compliant” (44 boys and 51 girls) when not meeting these criteria. Linear mixed models were used to examine differences in energy expenditure (EE) levels by compliance across 10 different time periods. The results indicated that non-compliant girls were older (13.4 ± 2.9 vs. 12.2 ± 2.5) and taller (156.8 ± 10.3 vs. 152.8 ± 11.3) than their same gender compliant peers (P < .05). Comparisons of EE rates at segmented portions of the day revealed no differences between compliant and non-compliant groups (P ≥ .05). Differences in EE ranged from ?0.32 kcal · kg^?1 · h^?1 (before school time) to 0.62 kcal · kg^?1 · h^?1 (physical education class) in boys and ?0.39 kcal · kg^?1 · h^?1 (transportation from school) to 0.37 kcal · kg^?1 · hour^?1 (recess) in girls. The results showed that compliant and non-compliant individuals differed in a few demographic characteristics but exhibited similar activity patterns. This suggests that data were considered to be missing at random, but additional work is needed to confirm this observation in a representative sample of children using other types of activity monitors and protocols.     

Bromo-oxidation reaction in enzyme-entrapped alginate hollow microfibers

In this article, the authors present the fabrication of an enzyme-entrapped alginate hollow fiber using a microfluidic device. Further use of enzyme-entrapped alginate hollow fibers as a biocatalytic microchemical reactor for chemical synthesis is also deliberated in this article. To ensure that there is no enzyme leaching from the fiber, fiber surfaces were coated with chitosan. To confine the mobility of reactants and products within the porous hollow fibers the entire fibers were embedded into a transparent polydimethylsiloxane (PDMS) matrix which also works as a support matrix. A vanadium-containing bromoperoxidase enzyme isolated from Corallina confusa was used as a model enzyme to demonstrate the use of these alginate hollow-fiber reactors in bromo-oxidation of phenol red to bromophenol blue at different dye flow rates. Stability of the entrapped enzyme at different temperatures and the effect of the chitosan coating on the reaction conversion were also studied. It was observed that molecules as big as 27 kDa can be retained in the matrix after coating with chitosan while molecules with molecular-weight of around 378 Da can still diffuse in and out of the matrix. The kinetic conversion rate in this microfluidic bioreactor was more than 41-fold faster when compared with the standard test-tube procedure.     

XMage: An image retrieval method based on partial similarity

XMage is introduced in this paper as a method for partial similarity searching in image databases. Region-based image retrieval is a method of retrieving partially similar images. It has been proposed as a way to accurately process queries in an image database. In region-based image retrieval, region matching is indispensable for computing the partial similarity between two images because the query processing is based upon regions instead of the entire image. A naive method of region matching is a sequential comparison between regions, which causes severe overhead and deteriorates the performance of query processing. In this paper, a new image contents representation, called Condensed eXtended Histogram (CXHistogram), is presented in conjunction with a well-defined distance function CXSim() on the CX-Histogram. The CXSim() is a new image-to-image similarity measure to compute the partial similarity between two images. It achieves the effect of comparing regions of two images by simply comparing the two images. The CXSim() reduces query space by pruning irrelevant images, and it is used as a filtering function before sequential scanning. Extensive experiments were performed on real image data to evaluate XMage. It provides a significant pruning of irrelevant images with no false dismissals. As a consequence, it achieves up to 5.9-fold speed-up in search over the R*-tree search followed by sequential scanning.     

Transportation of AIE-visualized nanoliposomes is dominated by the protein corona

Liposomes, especially cationic liposomes, are the most common and well-investigated nanocarriers for biomedical applications, such as drug and gene delivery. Like other types of nanomaterials, once liposomes are incubated in a biological milieu, their surface can be immediately cloaked by biological components to form a protein corona, which confers a new ‘biological identity’ and modulates downstream interactions with cells. However, it remains unclear how the protein corona affects the transportation mechanism after liposomes interact with cells. Here, we employed home-made aggregation-induced-emission-visualized nanoliposomes TR4@Lipo as a model to investigate transportation with or without the protein corona by optical imaging techniques. The results show that the protein corona can change the cellular transportation mechanism of TR4@Lipo from energy-independent membrane fusion to energy-dependent endocytosis. The protein corona also modulates the intracellular distribution of loaded cargoes. This knowledge furthers our understanding of bio-nano interactions and is important for the efficient use of cationic liposomes.     

Mentoring Undergraduate Interdisciplinary Mathematics Research Students: Junior Faculty Experiences

Abstract

To be successful, junior faculty must properly manage their time in the face of expanding responsibilities. One such responsibility is supervising undergraduate research projects. Student research projects (either single or multi-student) can be undertaken as a full-time summer experience, or as a part-time academic year commitment. With many potential undergraduate research formats, and with different types of students, junior faculty may find challenges in forming their research group, establishing a structure that promotes student productivity, picking an appropriate project, or in effectively mentoring their students. This article draws from the authors’ experiences to help junior faculty navigate these complexities so that all parties reap the benefits of undergraduate research in interdisciplinary mathematical disciplines.     

Antimelanogenic effects of <Emphasis Type="Italic">Inula britannica</Emphasis> flower petal extract fermented by <Emphasis Type="Italic">Lactobacillus plantarum</Emphasis> KCCM 11613P

The inhibitory effects of Lactobacillus plantarum-fermented and non-fermented Inula britannica extracts on the tyrosinase activity were comparatively investigated to examine whether and how they improve the whitening activity, and the contents of total flavonoids and polyphenolics as bioactive compounds were determined. The skin whitening activity using in vitro or ex vivo tyrosinase and L-3,4-dihydroxyphenylalanine (L-DOPA) staining was examined. The total flavonoid content (TFC) was increased by 13.4% after 72 h-fermentation. The viabilities of the B16F10 cells treated with the fermented and non-fermented control extracts were 100.26% and 92.15% at 500 μg/ml, respectively. In addition, the inhibition of tyrosinase activity was increased by the fermented samples from 29.33% to 41.74% following fermentation for up to 72 h. The tyrosinase activity of the untreated control group was increased to 145.69% in B16F10 cells. The results showed that I. britannica fermented by L. plantarum dose-dependently inhibited tyrosinase activity, which was stimulated by α-melanocyte stimulating hormone. These results suggest that lactic fermented I. britannica extracts can be used as effective skin-whitening materials.     

Dual-mode on-demand droplet routing in multiple microchannels using a magnetic fluid as carrier phase

We present dual-mode, on-demand droplet routing in a multiple-outlet microfluidic device using an oil-based magnetic fluid. Magnetite (Fe₃O₄) nanoparticle-contained oleic acid (MNOA) was used as a carrier phase for droplet generation and manipulation. The water-in-MNOA droplets were selectively distributed in a curved microchannel with three branches by utilizing both a hydrodynamic laminar flow pattern and an external magnetic field. Without the applied magnetic field, the droplets travelled along a hydrodynamic centerline that was displaced at each bifurcating junction. However, in the presence of a permanent magnet, they were repelled from the centerline and diverted into the desired channel when the repelled distance exceeded the minimum offset allocated to the channel. The repelled distance, which is proportional to the magnetic field gradient, was manipulated by controlling the magnet''s distance from the device. To evaluate routing performance, three different sizes of droplets with diameters of 63, 88, and 102 μm were directed into designated outlets with the magnet positioned at varying distances. The result demonstrated that the 102-μm droplets were sorted with an accuracy of ∼93%. Our technique enables on-demand droplet routing in multiple outlet channels by simply manipulating magnet positions (active mode) as well as size-based droplet separation with a fixed magnet position (passive mode).     

David F. Treagust: congenial soul, science educator, and international research leader

For almost a half century David F. Treagust has been an exemplary science educator who has contributed through his dedication and commitments to students, curriculum development and collaboration with teachers, and cutting edge research in science education that has impacted the field globally, nationally and locally. A hallmark of his outstanding career is his collaborative style that inspires others to produce their best work.     

A Corpus-Based Learning Method of Compound Noun Indexing Rules for Korean

In Korean information retrieval, compound nouns play an important role in improving precision in search experiments. There are two major approaches to compound noun indexing in Korean: statistical and linguistic. Each method, however, has its own shortcomings, such as limitations when indexing diverse types of compound nouns, over-generation of compound nouns, and data sparseness in training. In this paper, we propose a corpus-based learning method, which can index diverse types of compound nouns using rules automatically extracted from a large corpus. The automatic learning method is more portable and requires less human effort, although it exhibits a performance level similar to the manual-linguistic approach. We also present a new filtering method to solve the problems of compound noun over-generation and data sparseness.