Tibetan Sentiment Classification Method Based on Semi-Supervised Recursive Autoencoders

We apply the semi-supervised recursive autoencoders (RAE) model for the sentiment classification task of Tibetan short text, and we obtain a better classification effect. The input of the semi-supervised RAE model is the word vector. We crawled a large amount of Tibetan text from the Internet, got Tibetan word vectors by using Word2vec, and verified its validity through simple experiments. The values of parameter α and word vector dimension are important to the model effect. The experiment results indicate that when α is 0.3 and the word vector dimension is 60, the model works best. Our experiment also shows the effectiveness of the semi-supervised RAE model for Tibetan sentiment classification task and suggests the validity of the Tibetan word vectors we trained.     

Management Selection, Operations Research and Relevance: A CRITIQUE OF “STATISTICAL CONTAMINATION AND THE SELECTION OF MANAGEMENT”

Editor's Note: The reviewers of the preceding paper by Martin and Horowitz expressed diametrically opposing viewpoints concerning their recommendations for publication. Both reviewers felt the paper was well written and technically sound; one reviewer endorsed the work and recommended publication as a valuable contribution to the literature. The other reviewer felt just as strongly that the authors' work is a misapplication of mathematical modeling. We made the decision to publish the paper and accompany it with a statement of rebuttal from the opposing reviewer, as it appears here.     

Inter-rater reliability and validity of peer reviews in an interdisciplinary field

Peer review is an integral part of science. Devised to ensure and enhance the quality of scientific work, it is a crucial step that influences the publication of papers, the provision of grants and, as a consequence, the career of scientists. In order to meet the challenges of this responsibility, a certain shared understanding of scientific quality seems necessary. Yet previous studies have shown that inter-rater reliability in peer reviews is relatively low. However, most of these studies did not take ill-structured measurement design of the data into account. Moreover, no prior (quantitative) study has analyzed inter-rater reliability in an interdisciplinary field. And finally, issues of validity have hardly ever been addressed. Therefore, the three major research goals of this paper are (1) to analyze inter-rater agreement of different rating dimensions (e.g., relevance and soundness) in an interdisciplinary field, (2) to account for ill-structured designs by applying state-of-the-art methods, and (3) to examine the construct and criterion validity of reviewers’ evaluations. A total of 443 reviews were analyzed. These reviews were provided by m = 130 reviewers for n = 145 submissions to an interdisciplinary conference. Our findings demonstrate the urgent need for improvement of scientific peer review. Inter-rater reliability was rather poor and there were no significant differences between evaluations from reviewers of the same scientific discipline as the papers they were reviewing versus reviewer evaluations of papers from disciplines other than their own. These findings extend beyond those of prior research. Furthermore, convergent and discriminant construct validity of the rating dimensions were low as well. Nevertheless, a multidimensional model yielded a better fit than a unidimensional model. Our study also shows that the citation rate of accepted papers was positively associated with the relevance ratings made by reviewers from the same discipline as the paper they were reviewing. In addition, high novelty ratings from same-discipline reviewers were negatively associated with citation rate.     

Improving Chinese Word Representation with Conceptual Semantics

The meaning of a word includes a conceptual meaning and a distributive meaning. Word embedding based on distribution suffers from insufficient conceptual semantic representation caused by data sparsity, especially for low-frequency words. In knowledge bases, manually annotated semantic knowledge is stable and the essential attributes of words are accurately denoted. In this paper, we propose a Conceptual Semantics Enhanced Word Representation (CEWR) model, computing the synset embedding and hypernym embedding of Chinese words based on the Tongyici Cilin thesaurus, and aggregating it with distributed word representation to have both distributed information and the conceptual meaning encoded in the representation of words. We evaluate the CEWR model on two tasks: word similarity computation and short text classification. The Spearman correlation between model results and human judgement are improved to 64.71%, 81.84%, and 85.16% on Wordsim297, MC30, and RG65, respectively. Moreover, CEWR improves the F1 score by 3% in the short text classification task. The experimental results show that CEWR can represent words in a more informative approach than distributed word embedding. This proves that conceptual semantics, especially hypernymous information, is a good complement to distributed word representation.     

A multi-view method of scientific paper classification via heterogeneous graph embeddings

The classification task of scientific papers can be implemented based on contents or citations. In order to improve the performance on this task, we express papers as nodes and integrate scientific papers’ contents and citations into a heterogeneous graph. It has two types of edges. One type represents the semantic similarity between papers, derived from papers’ titles and abstracts. The other type represents the citation relationship between papers and the journals or proceedings of conferences of their references. We utilize a contrastive learning method to embed the nodes in the heterogeneous graph into a vector space. Then, we feed the paper node vectors into classifiers, such as the decision tree, multilayer perceptron, and so on. We conduct experiments on three datasets of scientific papers: the Microsoft Academic Graph with 63,211 scientific papers in 20 classes, the Proceedings of the National Academy of Sciences with 38,243 scientific papers in 18 classes, and the American Physical Society with 443,845 scientific papers in 5 classes. The experimental results on the multi-class task show that our multi-view method scores the classification accuracy up to 98%, outperforming state-of-the-arts.

     

Hashtag Recommendation Using LSTM Networks with Self-Attention

On Twitter, people often use hashtags to mark the subject of a tweet. Tweets have specific themes or content that are easy for people to manage. With the increase in the number of tweets, how to automatically recommend hashtags for tweets has received wide attention. The previous hashtag recommendation methods were to convert the task into a multi-class classification problem. However, these methods can only recommend hashtags that appeared in historical information, and cannot recommend the new ones. In this work, we extend the self-attention mechanism to turn the hashtag recommendation task into a sequence labeling task. To train and evaluate the proposed method, we used the real tweet data which is collected from Twitter. Experimental results show that the proposed method can be significantly better than the most advanced method. Compared with the state-of-the-art methods, the accuracy of our method has been increased 4%.     

A Test Statistic in Canonical Correlation Analysis for Categorical Variables

In canonical correlation analysis (CCA), it is important to estimate the number of nonzero canonical correlations in the population. One way to estimate the number is to consider the dimensionality testing problem. In CCA for continuous variables, some test statistics for the problem have been derived under the normality assumption. However, there are only a few papers on test statistics in CCA for categorical variables.

In this article, a test statistic in CCA for categorical variables is suggested. The test statistic is derived from the rational used for continuous variables. Some properties of the test statistic are examined through mathematical investigations and numerical simulations.

     

Model-based identification of damage from sparse sensor measurements using Neumann series expansion

This paper presents a novel and effective modal data-based methodology for structural damage localization and quantification when the structure is equipped with a limited number of sensors. Damage detection problem is defined as an inverse model-based problem and a new damage-sensitive cost function is introduced using calculated Generalised Flexibility Matrix and Modal Assurance Criterion. The second-order approximation of Neumann Series Expansion-based Model Reduction approach is employed for numerically simulation of sparse sensor installation. Finally, a hybrid version of two different evolutionary optimization algorithms, named Particle Swarm Optimization–Colonial Competitive Algorithm (PSO–CCA), is suggested and utilized for solving optimization problem. This hybridization, not only can pick the positive points of the PSO and CCA for searching complex solution domain, but also can lead to achieving a powerful, fast speed optimization strategy. The efficiency of the presented method is demonstrated by studying three numerical examples under different damage patterns. Various challenges, such as the robustness of the method in the presence of random noises in the input data, are investigated. The obtained results introduce the presented method as a viable and practical strategy for structural damage identification, especially when a limited number of sensors are installed on the structure.     

On peer review in computer science: analysis of its effectiveness and suggestions for improvement

In this paper we focus on the analysis of peer reviews and reviewers behaviour in a number of different review processes. More specifically, we report on the development, definition and rationale of a theoretical model for peer review processes to support the identification of appropriate metrics to assess the processes main characteristics in order to render peer review more transparent and understandable. Together with known metrics and techniques we introduce new ones to assess the overall quality (i.e. ,reliability, fairness, validity) and efficiency of peer review processes e.g. the robustness of the process, the degree of agreement/disagreement among reviewers, or positive/negative bias in the reviewers’ decision making process. We also check the ability of peer review to assess the impact of papers in subsequent years. We apply the proposed model and analysis framework to a large reviews data set from ten different conferences in computer science for a total of ca. 9,000 reviews on ca. 2,800 submitted contributions. We discuss the implications of the results and their potential use toward improving the analysed peer review processes. A number of interesting results were found, in particular: (1) a low correlation between peer review outcome and impact in time of the accepted contributions; (2) the influence of the assessment scale on the way how reviewers gave marks; (3) the effect and impact of rating bias, i.e. reviewers who constantly give lower/higher marks w.r.t. all other reviewers; (4) the effectiveness of statistical approaches to optimize some process parameters (e.g. ,number of papers per reviewer) to improve the process overall quality while maintaining the overall effort under control. Based on the lessons learned, we suggest ways to improve the overall quality of peer-review through procedures that can be easily implemented in current editorial management systems.     

Heuristics for single-pass welding task sequencing

Welding task sequencing is a prerequisite in the offline programming of robot arc welding. Single-pass welding task sequencing can be modelled as a modified travelling salesman problem. Owing to the difficulty of the resulting arc-routing problems, effective local search heuristics are developed. Computational speed becomes important because robot arc welding is often part of an automated process-planning procedure. Generating a reasonable solution in an acceptable time is necessary for effective automated process planning. Several different heuristics are proposed for solving the welding task-sequencing problem considering both productivity and the potential for welding distortion. Constructive heuristics based on the nearest neighbour concept and tabu search heuristics are developed and enhanced using improvement procedures. The effectiveness of the heuristics developed is tested and verified on actual welded structure problems and random problems.     

A survey on the resource-constrained project scheduling problem

In this paper, research on the resource-constrained project scheduling problem is classified according to specified objectives and constraints. Each classified area is extensively surveyed, and special emphasis is given to trends in recent research. Specific papers involving nonrenewable resource constraints and time/cost-based objectives are discussed in detail because they present models that are close representations of real-world problems. The difficulty of solving such complex models by optimization techniques is noted. For the purposes of this survey, a set of 78 optimally solved test problems from the literature and a second set of 110 benchmark problems have been subjected to analysis with some well-known dispatching rules and a scheduling algorithm that consists of a decision-making process utilizing the problem constraints as a base of selection. The computational results are reported and discussed in the text. Constructive scheduling algorithms that are directly based on the problem constraints and whose performances are independent of problem characteristics are identified as a promising area for future research.     

Comparison of multi-objective evolutionary approaches for task scheduling in distributed computing systems

Parallel and distributed systems play an important part in the improvement of high performance computing. In these type of systems task scheduling is a key issue in achieving high performance of the system. In general, task scheduling problems have been shown to be NP-hard. As deterministic techniques consume much time in solving the problem, several heuristic methods are attempted in obtaining optimal solutions. This paper presents an application of Elitist Non-dominated Sorting Genetic Algorithm (NSGA-II) and a Non-dominated Sorting Particle Swarm Optimization Algorithm (NSPSO) to schedule independent tasks in a distributed system comprising of heterogeneous processors. The problem is formulated as a multi-objective optimization problem, aiming to obtain schedules achieving minimum makespan and flowtime. The applied algorithms generate Pareto set of global optimal solutions for the considered multi-objective scheduling problem. The algorithms are validated against a set of benchmark instances and the performance of the algorithms evaluated using standard metrics. Experimental results and performance measures infer that NSGA-II produces quality schedules compared to NSPSO.     

The Bias of Sighted Reviewers in Research Proposal Evaluation: A Comparative Analysis of Blind and Open Review in Korea

This article compares empirically the major factors affecting blinded and sighted reviewers in the selection of research proposals to be funded in a "scientifically small" country. Fisher's Z-test shows that the applicant characteristics (rank of undergraduate school where the applicant studied, professional age of the applicant, and academic recognition of the applicant) are the major factors leading to the significantly different evaluation scores between blinded and sighted reviewers. This means that "open" evaluation of research proposals is obviously biased. Policy implications of the findings and future research directions are discussed.     

A Multicategory Kernel Distance Weighted Discrimination Method for Multiclass Classification

Abstract

Distance weighted discrimination (DWD) is an interesting large margin classifier that has been shown to enjoy nice properties and empirical successes. The original DWD only handles binary classification with a linear classification boundary. Multiclass classification problems naturally appear in various fields, such as speech recognition, satellite imagery classification, and self-driving vehicles, to name a few. For such complex classification problems, it is desirable to have a flexible multicategory kernel extension of the binary DWD when the optimal decision boundary is highly nonlinear. To this end, we propose a new multicategory kernel DWD, that is, defined as a margin-vector optimization problem in a reproducing kernel Hilbert space. This formulation is shown to enjoy Fisher consistency. We develop an accelerated projected gradient descent algorithm to fit the multicategory kernel DWD. Simulations and benchmark data applications are used to demonstrate the highly competitive performance of our method, as compared with some popular state-of-the-art multiclass classifiers.