Mining Skewed and Sparse Transaction Data for Personalized Shopping Recommendation

A good shopping recommender system can boost sales in a retailer store. To provide accurate recommendation, the recommender needs to accurately predict a customer's preference, an ability difficult to acquire. Conventional data mining techniques, such as association rule mining and collaborative filtering, can generally be applied to this problem, but rarely produce satisfying results due to the skewness and sparsity of transaction data. In this paper, we report the lessons that we learned in two real-world data mining applications for personalized shopping recommendation. We learned that extending a collaborative filtering method based on ratings (e.g., GroupLens) to perform personalized shopping recommendation is not trivial and that it is not appropriate to apply association-rule based methods (e.g., the IBM SmartPad system) for large scale prediction of customers' shopping preferences. Instead, a probabilistic graphical model can be more effective in handling skewed and sparse data. By casting collaborative filtering algorithms in a probabilistic framework, we derived HyPAM (Hybrid Poisson Aspect Modelling), a novel probabilistic graphical model for personalized shopping recommendation. Experimental results show that HyPAM outperforms GroupLens and the IBM method by generating much more accurate predictions of what items a customer will actually purchase in the unseen test data. The data sets and the results are made available for download at http://chunnan.iis.sinica.edu.tw/hypam/HyPAM.html.     

Lubrication performance of finite journal bearings considering effects of couple stresses and surface roughness

To inspect the performance characteristics of finite journal-bearing systems, the combined effects of couple stress due to a Newtonian lubricant blended with additives and the presence of roughness on journal-bearing surfaces are studied in this article. Basing on the Stokes theory and Christensen’s stochastic model, the stochastic generalized Reynolds equation is deduced. The film pressure distribution equation is numerically solved by using the conjugate gradient method of iterations. According to the results, the couple stress effects can raise the film pressure of the lubricant fluid, improve the load-carrying capacity and reduce the friction parameter, especially at high eccentricity ratio. The surface roughness effect is dominant in long bearing approximation and the influence of transverse or longitudinal roughness to the journal bearing is in reverse trend. In general, the critical value of length-to-diameter is 1.1.     

Three-dimensional tracking and temporal analysis of liposomal transport in live cells using bright-field imaging

Gold nanoparticles (AuNPs) confined in liposomes of diameters around 200 nm produce strong scattering signal owing to surface plasmon resonance, and therefore bright-field optical tracking of the AuNP-encapsulating liposomes can be conducted in living cells. Using an optical profiling technique called noninterferometric wide-field optical profilometry and a bright-field tracking algorithm, the polynomial-fit Gaussian weight method, we analyze three-dimensional (3D) motion of such liposomes in living fibroblasts. The positioning accuracy in three dimensions is nearly 20 nm. We tag the liposome membranes with fibroblast growth factor-1 and reveal the intracellular transportation processes toward or away from the nucleus. On the basis of a temporal analysis of the intracellular 3D trajectories of AuNP-encapsulating liposomes, we identify directed and diffusive motions in the transportation processes.     

An efficient methodology for the structural synthesis of geared kinematic chains

Through the use of acyclic graphs, an efficient methodology has been developed for the synthesis of the kinematic structure of geared kinematic chains with any number of links. First, a systematic approach is proposed for the enumeration of N-vertex acyclic graphs. For each N-vertex acyclic graph, all the N-vertex geared graphs are generated by adding (N — F — 1) geared edges to the acyclic graphs, eliminating geared graphs which violate the fundamental rules, and non-isomorphic geared graphs are obtained by comparing the structural codes. Finally, the catalog of the graphs of one-degree-of-freedom (DOF) geared kinematic chains with N links are synthesized from the catalog of acyclic graphs with N vertices. As a result, the structural synthesis of one-DOF geared kinematic chains with up to eight links has been successfully constructed.     

Comparison of friction measurements using the atomic force microscope and the surface forces apparatus: the issue of scale

Results are presented of lateral force measurements using the atomic force microscope (AFM) and the surface forces apparatus (SFA). Two different probes are used in the AFM measurements; a sharp silicon nitride tip (radius R20 nm) and a glass ball (R15 m). The lateral force is measured between the (silicon nitride or glass) probe and a mica surface which has been coated by a thin lubricant film. In the SFA, a thin lubricant film separates two molecularly smooth mica surfaces (R1 cm) which are slid relative to each other. Perfluoropolyether (PFPE) and polydimethylsiloxane (PDMS) were used as the lubricant films. In the SFA where the contact diameter is largest, the PFPE film shows much lower friction than PDMS. As the size of the probe decreases, the difference in the measured friction decreases. For sharp AFM tips, no clear distinction between the tribological properties of the films can be made. Hence, the measured coefficient of friction varies according to the length scale probed, at least for small dimensions.     

Molecular basis of lubrication

The advent of micro-electromechanical devices (MEMs), sensors, actuators, microsystems, and nanotechnology have called to attention the effect of friction on moving parts in nano/micro devices. To take full advantage of the opportunity to sense, compute, and actuate in real time, fast-moving parts are often necessary or desirable. As the scales of the components shrink, adhesion, stiction, friction, and wear become a significant technological barrier for the successful deployment of durable devices. Most current devices in production avoid such contacts.The nature of the surface contacts, as component scale moves from macro to micro to nano, is dominated by surface forces that normally are dwarfed by mechanical loading. Therefore nanolubrication needs to take into account different factors than conventional lubrication concepts. This paper compares traditional lubrication concepts and those necessary for nanolubrication and proposes various nanometer scale thick lubricating film designs as a means to control the surface properties of surfaces at nano/micro scales.Many of the concepts derive their origin from studies and observations from the magnetic hard disk technology where a “monolayer” of lubricant protects the system and has proven to be robust and safe. Examples from magnetic hard disks will be used to illustrate some of the concepts.     

Adaptable Scheduling Algorithm for Grids with Resource Redeployment Capability

Two distinct characteristics of grid computing systems are resource heterogeneity and availability variation. There are many well-designed scheduling algorithms proposed for heterogeneous computing systems. However, the availability variation is seldom considered in developing scheduling ongoing applications on a grid. In this paper, two scheduling algorithms called AMOF and AMOSF are proposed. Both of them consider availability variation as well as resource heterogeneity while scheduling an ongoing workflow application on the grid. An experiment has been conducted to demonstrate that AMOF and AMOSF algorithms outperform the well-known scheduling algorithms: GS and HEFT in most of the cases.     

A novel approach for optimizing the optical performance of the broadband tap coupler

An integrated approach using neural networks, exponential desirability functions and genetic algorithms to optimize parameter design problems with multiple responses is presented. The proposed approach aims to identify the input parameter settings to maximize the overall minimal satisfaction level with respect to all the responses. The proposed approach is illustrated by optimizing the fused process parameters created during fused biconic taper coupler development to improve the performance and reliability of a 1% (1199) single-window broadband tap coupler. The proposed solution procedure was implemented on a Taiwanese manufacturer of fibreoptic passive components. The implementation results demonstrate the practicability of the method. Comparison analysis revealed that the proposed procedure outperformed the traditional Taguchi method in resolving multi-response parameter design problems.     

Composite hydrogeological investigation and characterisation methods applied at a nuclear power plant site in Taiwan

Drawing from the lessons learnt from the Fukushima nuclear disaster in Japan, the assessment of subsurface environmental hazard around nuclear power plants has emerged as a critical task. Consequently, aiming to better understand the possible environmental impact of radiation leaks into the groundwater, a prescreening programme was initiated in 2011 by the Taiwanese government. As part of this programme, this study conducted various borehole prospecting techniques to identify in situ hydrogeological characteristics at Chin‐Shan Nuclear Power Plant. Borehole electrical log, sonic log and temperature/conductivity log were conducted to explain the regional lithologic conditions and permeability of the formation. In conjunction with this, the interwell tracer and pumping test was carried out to simultaneously determine the hydraulic parameters. In our opinion, the implementation of such in situ end‐to‐end investigations is essential in interpreting in situ fluid and solute transport dynamics prior to programming any numerical scheme for early warning, vulnerability assessment and regular monitoring of a nuclear power plant site.     

An ergonomic study of visual optimization by light color managements

Light‐emitting diodes (LEDs) have the advantage of controllable light colors and provide the optimal lighting quality against various color attributes required in different environments. However, from the humanized design point of view, beside satisfying physical demands of people, it is also important to achieve satisfaction. With a good match between the LED lighting equipment and ambient colors, the space imagery that complies with psychological needs of human beings can be found. This is expected to become a new market direction for future lighting. In this study, the LED light source feature of adjustable correlated color temperature is used to match wall colors of typical commercial spaces and for the investigation and analysis of the impact of colors of light sources and environments on human beings. The measures of mental and subjective feelings of a group of people under various circumstances have also been investigated. Multivariate analysis of variance has been conducted on three color temperatures of light sources, six wall colors, and measures of five emotion adjectives, and the correlation between these three variables has been further investigated. © 2014 Wiley Periodicals, Inc. Col Res Appl, 41, 72–84, 2016