LkeRec: Toward Lightweight End-to-End Joint Representation Learning for Building Accurate and Effective Recommendation

Explicit and implicit knowledge about users and items have been used to describe complex and heterogeneous side information for recommender systems (RSs). Many existing methods use knowledge graph embedding (KGE) to learn the representation of a user-item knowledge graph (KG) in low-dimensional space. In this article, we propose a lightweight end-to-end joint learning framework for fusing the tasks of KGE and RSs at the model level. Our method proposes a lightweight KG embedding method by using bidirectional bijection relation-type modeling to enable scalability for large graphs while using self-adaptive negative sampling to optimize negative sample generating. Our method further generates the integrated views for users and items based on relation-types to explicitly model users’ preferences and items’ features, respectively. Finally, we add virtual “recommendation” relations between the integrated views of users and items to model the preferences of users on items, seamlessly integrating RS with user-item KG over a unified graph. Experimental results on multiple datasets and benchmarks show that our method can achieve a better accuracy of recommendation compared with existing state-of-the-art methods. Complexity and runtime analysis suggests that our method can gain a lower time and space complexity than most of existing methods and improve scalability.

The multi-dimensional space of the futures of work

Purpose The authors present nine dimensions to provide structure for the many Futures of Work (FoW). This is done to advance a more sociotechnical and nuanced approach to the FoW, which is too-often articulated as singular and unidimensional. Futurists emphasize they do not predict the future, but rather, build a number of possible futures – in plural – often in the form of scenarios constructed based on key dimensions. Such scenarios help decision-makers consider alternative actions by providing structured frames for careful analyses. It is useful that the dimensions be dichotomous. Here, the authors focus specifically on the futures of knowledge work.Design/methodology/approach Building from a sustained review of the FoW literature, from a variety of disciplines, this study derives the nine dimensions.Findings The nine FoW dimensions are: Locus of Place, Locus of Decision-making, Structure of Work, Technologies’ Roles, Work–Life, Worker Expectations, Leadership Model, Firm’s Value Creation and Labor Market Structure. Use of the dimensions is illustrated by constructing sample scenarios.Originality/value While FoW is multi-dimensional, most FoW writing has focused on one or two dimensions, often highlighting positive or negative possibilities. Empirical papers, by their nature, are focused on just one dimension that is supported by data. However, future-oriented policy reports tend are more often multi-faceted analyses and serve here as the model for what we present.

Meaning and structural dynamics in poetry: a computational perspective

This work addresses Sri Aurobindo’s mantric poem, Savitri, with a computational linguistics approach. This is one of the longest poems ever written in English. We build the connectivity matrix between all main word pairs and analyse its structure. Concepts emerge as directions that better explain the variance of the data in the hyperspace of words. When projected to the low dimensional space of concepts, the vector of attention as the reader moves through the text shows a large correlation across sections of the poem, thus acting the future and the past over again. These findings suggest that the mathematical structure of Savitri is and reflects a substrate for the author’s main ideas, facilitating the reader’s understanding of the poem’s meaning via its long-range dynamical correlations. Acknowledging an irreducible essence to poetry, future studies on the relationship between words and sounds, and sounds and ideas may provide invaluable hints of the origin of language and its intimate relationship with the evolution of human consciousness.

METHOD OF FINDING LOSSES IN THE THROTTLE WINDING AS A RESULT OF THE SKIN EFFECT AND THE PROXIMITY EFFECT

Представлен метод использования расчетных методик и моделирования магнитных полей в двухмерном пространстве для нахождения высокочастотных потерь в обмотке моточных изделий, таких как дроссель или трансформатор. Была проведена работа по анализу литературы по данной теме, а также поднят вопрос оптимизации и адаптации аналитических формул для случая использования проводников круглого сечения и намотки, имеющей неоднородное распределение слоев в окне сердечника. Был также поднят вопрос об аналитическом нахождении длины обмоточного провода намотки с различным количеством слоев и переменного количества используемых при этом витков. Для проведения автоматизации расчета с помощью формул был написан скрипт, строящий зависимость сопротивления переменному току относительно частоты, используя аналитические формулы. Была написана программа для автоматической постановки начальных условий и граничных значений параметров моделирования, процесса самого моделирования электромагнитных полей, анализа полученных данных, а также формирования массива для построения графика полученной при этом зависимости сопротивления от частоты. В данном методе используется свободно распространяемое программное обеспечение как для математических расчетов, так и моделирования электромагнитных полей. Итогом данной работы стало сравнение полученных результатов, которые показали хорошую сходимость и преемственность этапов данного метода Here we present a method for using computational methods and modeling magnetic fields in two-dimensional space to find high-frequency losses in the winding of winding products, such as a choke or transformer. We analyzed the literature on this topic, as well as the issue of optimization and adaptation of analytical formulas for the case of using round-section conductors and winding having a non-uniform distribution of layers in the core window. We discussed the analytical finding of the length of the winding wire of the winding with a different number of layers and a variable number of turns used in this case. To automate the calculation using formulas, we wrote a script that builds the dependence of the resistance to alternating current relative to the frequency using analytical formulas. In addition, we wrote a program for automatically setting the initial conditions and boundary values of the modeling parameters, the process of modeling electromagnetic fields itself, analyzing the data obtained, as well as forming an array for plotting the resulting dependence of resistance on frequency. This method uses freely distributed software for both mathematical calculations and modeling of electromagnetic fields. The result of this work was a comparison of the results obtained, which showed good convergence and continuity of the stages of this method

Reply to Comment on 'Inverse problem from the discrete spectrum in the D = 2 dimensional space'

We want to thank our colleague F. Fernandez for his interest and his careful reading of our paper "The inverse problem from discrete spectrum in the D = 2 dimensional space". We are confused to have left a number of mistakes in the manuscript.

EZ Clear for simple, rapid, and robust mouse whole organ clearing

Tissue clearing for whole organ cell profiling has revolutionized biology and imaging for exploration of organs in three-dimensional space without compromising tissue architecture. But complicated, laborious procedures, or expensive equipment, as well as the use of hazardous, organic solvents prevents the widespread adoption of these methods. Here we report a simple and rapid tissue clearing method, EZ Clear, that can clear whole adult mouse organs in 48 hours in just three simple steps. Samples stay at room temperature and remain hydrated throughout the clearing process, preserving endogenous and synthetic fluorescence, without altering sample size. After wholemount clearing and imaging, EZ Cleared samples can be further processed for downstream embedding and cryosectioning followed by standard histology or immunostaining, without loss of endogenous or synthetic fluorescence signal. Overall, the simplicity, speed, and flexibility of EZ Clear make it easy to adopt and apply to diverse approaches in biomedical research.

Comment on Inverse problem from discrete spectrum in the D = 2 dimensional space (2014 Phys. Scr. 989 085201)

We analyse a method for the construction of the potential-energy function from the moments of the ground-state density. The sum rule on which some expressions are based appear to be wrong, as well as the moments and potential-energy functions derived for some illustrative examples.