Search

During the COVID-19 pandemic, the novel coronavirus had an impact not only on public health but also on the mental health of the population. Public sentiment on mental health and depression is often captured only in small, survey-based studies, while work based on Twitter data often only looks at the period during the pandemic and does not make comparisons with the pre-pandemic situation. We collected tweets that included the hashtags #MentalHealth and #Depression from before and during the pandemic (8.5 months each). We used LDA (Latent Dirichlet Allocation) for topic modeling and LIWC, VADER, and NRC for sentiment analysis. We used three machine-learning classifiers to seek evidence regarding an automatically detectable change in tweets before vs. during the pandemic: (1) based on TF-IDF values, (2) based on the values from the sentiment libraries, (3) based on tweet content (deep-learning BERT classifier). Topic modeling revealed that Twitter users who explicitly used the hashtags #Depression and especially #MentalHealth did so to raise awareness. We observed an overall positive sentiment, and in tough times such as during the COVID-19 pandemic, tweets with #MentalHealth were often associated with gratitude. Among the three classification approaches, the BERT classifier showed the best performance, with an accuracy of 81% for #MentalHealth and 79% for #Depression. Although the data may have come from users familiar with mental health, these findings can help gauge public sentiment on the topic. The combination of (1) sentiment analysis, (2) topic modeling, and (3) tweet classification with machine learning proved useful in gaining comprehensive insight into public sentiment and could be applied to other data sources and topics. [ABSTRACT FROM AUTHOR]

Presents a mathematical definition of the probability-weighted amount of information (PWI), a measure of term specificity in documents that is based on an information-theoretic view of retrieval events. Corresponds with the term frequency-inverse document frequency (tf-idf) measures that are used in information retrieval systems. (Author/LRW)

In this paper we present new methods for the automated learning of heuristics in knowledge-lean applications and for finding heuristics that can be generalized to unlearned domains. These applications lack domain knowledge for credit assignment; hence, operators for composing new heuristics are generally model-free, domain independent, and syntactic in nature. The operators we have used are genetics-based; examples of which include mutation and cross-over. Learning is based on a generate-and-test paradigm that maintains a pool of competing heuristics, tests them to a limited extent, creates new ones from those that perform well in the past, and prunes poor ones from the pool. We have studied three important issues in learning better heuristics: 1) anomalies in performance evaluation, 2) rational scheduling of limited computational resources in testing candidate heuristics in single-objective as well as multiobjective learning, and 3) finding heuristics that can be generalized to unlearned domains. We show experimental results in learning better heuristics for 1) process placement for distributed-memory multicomputers, 2) node decomposition in a branch-and-bound search, 3) generation of test patterns in VLSI circuit testing, and 4) VLSI cell placement and routing. Index Terms - Branch-and-bound search, generalization, genetics-based learning, heuristics, knowledge-lean applications, performance evaluation, process mapping, resource scheduling, VLSI circuit testing, VLSI placement and routing.

Scientific documents generally contain multiple mathematical expressions in them. Detecting inline mathematical expressions are one of the most important and challenging tasks in scientific text mining. Recent works that detect inline mathematical expressions in scientific documents have looked at the problem from an image processing perspective. There is little work that has targeted the problem from NLP perspective. Towards this, we define a few features and applied Conditional Random Fields (CRF) to detect inline mathematical expressions in scientific documents. Apart from this feature based approach, we also propose a hybrid algorithm that combines Bidirectional Long Short Term Memory networks (Bi‐LSTM) and feature‐based approach for this task. Experimental results suggest that this proposed hybrid method outperforms several baselines in the literature and also individual methods in the hybrid approach. [ABSTRACT FROM AUTHOR]

Face hallucination is a technique that reconstructs high-resolution (HR) faces from low-resolution (LR) faces, by using the prior knowledge learned from HR/LR face pairs. Most state-of-the-arts leverage position-patch prior knowledge of the human face to estimate the optimal representation coefficients for each image patch. However, they focus only the position information and usually ignore the context information of the image patch. In addition, when they are confronted with misalignment or the small sample size (SSS) problem, the hallucination performance is very poor. To this end, this paper incorporates the contextual information of the image patch and proposes a powerful and efficient context-patch-based face hallucination approach, namely, thresholding locality-constrained representation and reproducing learning (TLcR-RL). Under the context-patch-based framework, we advance a thresholding-based representation method to enhance the reconstruction accuracy and reduce the computational complexity. To further improve the performance of the proposed algorithm, we propose a promotion strategy called reproducing learning. By adding the estimated HR face to the training set, which can simulate the case that the HR version of the input LR face is present in the training set, it thus iteratively enhances the final hallucination result. Experiments demonstrate that the proposed TLcR-RL method achieves a substantial increase in the hallucinated results, both subjectively and objectively. In addition, the proposed framework is more robust to face misalignment and the SSS problem, and its hallucinated HR face is still very good when the LR test face is from the real world. The MATLAB source code is available at https://github.com/junjun-jiang/TLcR-RL. [ABSTRACT FROM AUTHOR]

Despite the increased quality of Machine Translation output, human interaction will remain a crucial activity to guarantee the quality of the final translation products. Human-computer interaction in translation will likely be the more successful the more we understand the properties and complementarities of both partners. This paper traces cognitive approaches in machine translation back to the mid-1980s and argues that we now have the technologies available that will allow us to eventually arrive at an in-depth understanding of the human translation processes. It illustrates some of the research methods in empirical translation process research and suggests ngrams of Activity Units for measuring the translation process. [ABSTRACT FROM AUTHOR]

This study is a mixed method, i.e. qualitative as well as quantitative, analysis of the stylistic characteristics of texts submitted to Japanese Q & A communities. Along with the development of social media, Q & A communities are attracting much scholarly attention as important resources for analysing online communication. In Q & A communities, people freely submit questions and answers; questions are classified into subject categories; and the best answers are selected. In this study, we analyse the stylistic characteristics of three types of submission, i.e. questions, best answers, and normal answers, in two different subject categories, i.e. "personal computers and related devices" and "love and human relations advice". The results show that the textual styles clearly distinguished these six classes of texts and clarified their respective characters. Our findings provide useful knowledge about how people differ in their communication styles regarding subject categories and on how people select communication styles. This study will contribute to research into discovering current online communication styles. [ABSTRACT FROM AUTHOR]

Along with its methodological development, authorship analysis has expanded in scope to new application areas like authorship profiling and computational sociolinguistics as well as conventional ones like authorship attribution. For these new applications, providing a new interpretation of text through the textual characteristics is as important as improving the classification performance between the authors, which was the aim in conventional applications. Lexical indicators were one of the most frequently used characteristics in conventional applications as they were effective at discriminating between authors, but most of the previously used indicators were based on the frequencies of morphemes, and reflected only limited aspects of the writing styles of the authors. In order to use these types of characteristics for new applications, we need to develop indicators reflecting other various aspects of the authors writing styles that are useful for interpretation as well as classification. As such, we propose the use of two types of co-occurrence-based indicators, namely network indicators and a co-occurrence-based concentration indicator in this field. Experimental results using the Aozora Bunko corpora, along with qualitative analyses, showed that our indicators were very effective at capturing aspects of the styles of the authors as well as for improving the classification performance. We concluded that our indicators successfully supplement previously used indicators and are useful for various new applications in authorship analysis. [ABSTRACT FROM AUTHOR]

In this paper, the authors evaluate a method to calculate association between specialized terms using co-occurrence information in author keywords from academic papers. When author keywords found in academic paper databases are used, co-occurrence information in units of separate compound words can easily be obtained. However, because only a few words co-occur in one paper, the problem of data sparseness arises. Thus, in this paper, the authors take into consideration indirect co-occurrence relationships when computing relatedness. They create a large-scale terminology graph which connects sets of keywords for a paper using co-occurrence relationship links, then define the distance between two sets of terms using the average path length. In addition, the authors evaluate the validity of their proposed method for calculating association by applying the terminology graph created using a real, large-scale academic paper database to the problem of automatic classification of texts, then compare the results to when direct co-occurrence and context vectors are used. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(3): 85–95, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.1197 [ABSTRACT FROM AUTHOR]

In this paper, genetic algorithms are interpreted as a combination of probabilistic solution-space decomposition and randomized search. We study a method for characterization of the solution space from this point of view. Initially, a statistical measure called the variance coefficient is defined as an index to characterize the solution space. Next, three parameters that are commonly used to characterize a solution space are expressed in terms of the defined variance coefficients. The three parameters are the Walsh coefficient, epistasis variance, and correlation coefficient between generations. In particular, the generation correlation, which used to be known only empirically as an effective performance measure for genetic algorithms, is clearly expressed in terms of the variance coefficients. Based on the definition, the theoretical values of the generation correlations are compared for representative crossover operator; namely uniform crossover and one-point crossover. In addition, the correspondence between the theory and the performance of actual genetic algorithms are demonstrated by simple simulation experiments. © 1998 Scripta Technica, Syst Comp Jpn, 29(5): 1–10, 1998 [ABSTRACT FROM AUTHOR]

A new population-oriented search algorithm called Stochastic Schemata Exploiter (SSE) is proposed here. Similarly to genetic algorithms (GAs), SSE also performs search in the space of solutions by a schemata processing mechanism; however, SSE is characterized by an emphasis on local search. So far, the adaptive global search ability of GAs has been viewed positively. In contrast, SSE takes the point of view that, when applied to actual optimization problems, the GA adaptive global search ability is not necessarily effective; instead, the global search is started by simplifying the population-oriented search of GAs and reducing the number of control parameters. After reviewing herein the schemata processing in GAs, the property of schemata exploitation is defined, a schemata exploitation search algorithm is devised and the proposed method is evaluated and compared with a simple GA using GA-easy and GA-hard test problems. [ABSTRACT FROM AUTHOR]