Short text classification with machine learning in the social sciences: The case of climate change on Twitter.

The News of the Week article that reports on Senator Kay Bailey Hutchison (R-TX) questioning the need to fund social science research at the National Science Foundation is alarming and shortsighted (“Senate panel chair asks why NSF funds social sciences,” 12 May, p. [829][1]). Social science research is at the fundamental core of basic research and has much to contribute to the economic viability of the United States. Twenty years of direct and jointly funded social and ecosystem science research at Colorado State University's Natural Resource Ecology Laboratory has produced deep insights into environmental and societal impacts of political upheaval, land use, and climate change in parts of Africa, Asia, and the Americas. Beyond greatly advancing our understanding of the coupled human-environmental system, the partnership of social and ecosystem science has brought scientists and decision-makers together to begin to develop solutions to difficult problems. Insights from the social sciences about the vulnerability, sustainability, and resilience of social and environmental systems have greatly increased our understanding of the complex interactions of the world on which we depend. The contributions of social and ecological science for data analyses associated with spatial relationships, transfer of goods and services, valuation, and decision-making processes have improved our understanding of how human activities are altering environmental systems worldwide. These findings have become the core of policy and development decisions used throughout the world. It is critical that the NSF support social science funding and that, as budget decisions are made, the social science allocation should move forward equally with increases as in other basic and applied research. Failure to do so will further hinder U.S. competitiveness in the future and will slow transfer of knowledge and usefulness to the public. [1]: /lookup/doi/10.1126/science.312.5775.829a

This article examines the value of "eclecticism" as the foundation of meta-theoretical, mixed methods and interdisciplinary research in social sciences. On the basis of the analysis of the historical background of the concept, it is first suggested that eclecticism-based theoretical scholarship in social sciences could benefit from the more systematic research method that has been developed for synthesizing theoretical works under the name metatheorizing. Second, it is suggested that the mixed methods community could base its research approach on philosophical eclecticism instead of pragmatism because the basic idea of eclecticism is much more in sync with the nature of the combined research tradition. Finally, the Kuhnian frame is used to support the argument for interdisciplinary research and, hence, eclecticism in social sciences (rather than making an argument against multiple paradigms). More particularly, it is suggested that integrating the different (inter)disciplinary traditions and schools into one is not necessarily desirable at all in social sciences because of the complexity and openness of the research field. If it is nevertheless attempted, experience in economics suggests that paradigmatic unification comes at a high price.

Mild cognitive impairment (MCI) is a clinical condition characterized by a decline in cognitive ability and progression of cognitive impairment. It is often considered a transitional stage between normal aging and Alzheimer's disease (AD). This study aimed to compare deep learning (DL) and traditional machine learning (ML) methods in predicting MCI using plasma proteomic biomarkers. A total of 239 adults were selected from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort along with a pool of 146 plasma proteomic biomarkers. We evaluated seven traditional ML models (support vector machines (SVMs), logistic regression (LR), naïve Bayes (NB), random forest (RF), k-nearest neighbor (KNN), gradient boosting machine (GBM), and extreme gradient boosting (XGBoost)) and six variations of a deep neural network (DNN) model-the DL model in the H2O package. Least Absolute Shrinkage and Selection Operator (LASSO) selected 35 proteomic biomarkers from the pool. Based on grid search, the DNN model with an activation function of "Rectifier With Dropout" with 2 layers and 32 of 35 selected proteomic biomarkers revealed the best model with the highest accuracy of 0.995 and an F1 Score of 0.996, while among seven traditional ML methods, XGBoost was the best with an accuracy of 0.986 and an F1 Score of 0.985. Several biomarkers were correlated with the APOE-ε4 genotype, polygenic hazard score (PHS), and three clinical cerebrospinal fluid biomarkers (Aβ42, tTau, and pTau). Bioinformatics analysis using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) revealed several molecular functions and pathways associated with the selected biomarkers, including cytokine-cytokine receptor interaction, cholesterol metabolism, and regulation of lipid localization. The results showed that the DL model may represent a promising tool in the prediction of MCI. These plasma proteomic biomarkers may help with early diagnosis, prognostic risk stratification, and early treatment interventions for individuals at risk for MCI.

The purportedly irreconcilable aims of “area studies” versus formal disciplines are a long-standing concern. In reality, their objectives are often inseparable, and approaches that start from and center a region have strongly contributed to theory building within disciplines. Few social scientists have been so productive in building bridges between these competing frames as James C. Scott, as evidenced by his celebrated body of work, his election to the presidency of the Association for Asian Studies, and his receipt of meritorious citations such as the Social Science Research Council's 2020 Albert O. Hirschman Prize, awarded to “scholars who have made outstanding contributions to international, interdisciplinary social science research, theory, and public communication.”

Social science research has been attacked by neoliberal thinkers who allege that such research lacks economic objectives. In the face of neoliberal and positivist criteria for evaluating the social impact of social science inquiry, social science researchers are developing qualitative evaluation methodologies through which we can have direct contact with citizens. These qualitative methodologies declare our social responsibility as social researchers in addressing relevant problems, especially those affecting the most vulnerable people. From these qualitative methodologies, the most vulnerable groups are included in the assessment of the social impacts of social research. Some examples of people who have participated in this qualitative evaluation include women, youth, immigrants, and Roma organizations. Participants perceived social science researchers as being far from their social reality, but in this research, they began to overcome their skepticism that social science research can help to solve those problems affecting their everyday lives.

Modelling short text is challenging due to the small number of word co-occurrence and insufficient semantic information that affects downstream Natural Language Processing (NLP) tasks, for example, text classification. Gathering information from external sources is expensive and may increase noise. For efficient short text classification without depending on external knowledge sources, we propose Expressive Short text Classification (EStC). EStC consists of a novel document context-aware semantically enriched topic model called the Short text Topic Model (StTM) that captures words, topics and documents semantics in a joint learning framework. In StTM, the probability of predicting a context word involves the topic distribution of word embeddings and the document vector as the global context, which obtains by weighted averaging of word embeddings on the fly simultaneously with the topic distribution of words without requiring an additional inference method for the document embedding. EStC represents documents in an expressive (number of topics × number of word embedding features) embedding space and uses a linear support vector machine (SVM) classifier for their classification. Experimental results demonstrate that EStC outperforms many state-of-the-art language models in short text classification using several publicly available short text data sets.

Social science has an important role in aviation biofuels research, yet social science methods and approaches tend to be underdeveloped and under-utilized in the broader aviation biofuels literature and biofuels overall. Over the last 5 years, social science approaches in aviation biofuels research, particularly site-selection, have made several advances. Where early site-selection models either entirely excluded social science concepts or included only a few measurements using poor proxies, current models more accurately, and more comprehensively capture key social science concepts to better examine and predict project implementation success and long-term sustainability. Despite several studies published within the last 20 years noting the need for more empirical studies of social sustainability and improvement in incorporation of social criteria, progress has remained rather stagnant in several areas. To help move the field forward, we conduct a review of the current state of social science research in aviation biofuels with a focus on sustainability, site-selection, and public acceptance research, identifying key approaches, important developments, and research gaps and weaknesses of current approaches. While several review studies already exist, they tend to focus on a single area of biofuels such as public acceptance. By broadening our review to several areas, we are able to identify several common limitations across these areas that contribute to the continued underutilization of social science approaches in aviation biofuels. This includes the preference for practical and reliable indicators for social criteria that prioritize quantitative methods over other approaches. Based on these limitations, we make several recommendations to improve social science research in aviation biofuels, including ensuring that social scientists are key members of the research team, the adoption of a mixed-methods research designs that combines quantitative and qualitative approaches that better measure some criteria and local-level impacts, and adequate resources for social science research throughout biofuel development projects as these methods are often more time-consuming and costly to implement. We argue that implementing these recommendations in future aviation biofuel development projects will improve social science approaches utilized in aviation biofuels research and address a long-acknowledged gap in the field.

Addressing the policy challenges and opportunities of “Big data”

Short text is usually expressed in refined slightly, insufficient information, which makes text classification difficult. But we can try to introduce some information from the existing knowledge base to strengthen the performance of short text classification. Wikipedia [2, 13, 15] is now the largest human-edited knowledge base of high quality. It would benefit to short text classification if we can make full use of Wikipedia information in short text classification. This paper presents a new concept based [22] on Wikipedia short text representation method, by identifying the concept of Wikipedia mentioned in short text, and then expand the concept of wiki correlation and short text messages to the feature vector representation.

This review seeks to understand the implications of empiricism, interpretivism, phenomenology, hermeneutics, pluralism, and solipsism in social science inquiry within social reality, using scoping review method. An in-depth review of the literature was conducted to understand the various philosophical alignments or paradigms applied in social sciences research. The study is aimed at understanding and applying these paradigms with the view of having a deep comprehension of the turbulences plaguing society and proffering solutions to them. From the review, it was discovered that the various philosophical alignments in social science inquiry provide the scientist or researcher with a worldview of the different perspectives and multiple levels of analysis of the social world. Thus, creating a systematic lens from which individuals (at several levels), groups, and organizations are examined to know the; “when”, “how”, and “why” they behave the way and manner that they do. It was concluded that an adequate understanding of these paradigms would better shape the methodology to be adopted in conducting research studies within the social and/or behavioural sciences. This will help determine its objectivity, rigor, or the extent to which scientific methodology is applied within social science researches. Lastly, a combination of these paradigms creates a mixed-method, which demonstrates knowledge validity and objectivity in investigator triangulation (which involves; observation, questionnaire administration, and interview), data triangulation (which encompasses collecting data at different times from different people in different places as a cross-check for validity and to check the interpretation and conclusion arrived at), methodological triangulation (which includes within-method triangulation and between method triangulation: the former entails the application of various techniques within the same method, while the latter applies a combination of research methods), generalization, verification, explanation, and deductions. This will allow for data gathering and/or fact-finding, in search of new knowledge, and subsuming new valid knowledge, enhance research results and findings efficacy.

PurposeThe purpose of this paper is to describe and discuss a few principal and crucial steps of “antecedents” and “postcedents” in relation to structural equation modeling (SEM) in social science research.Design/methodology/approachThis paper is based upon a conceptual and reflective discussion of SEM in social sciences research.FindingsSEM is not science per se, but just a tool that is used to extract meaning from the previous steps of research processes in terms of a research phenomenon in focus.Research limitations/implicationsBased upon the steps of validity and reliability discussed in relation to predetermined measurement and structural models in SEM, application implications are provided.Practical implicationsLike many statistical tools, SEM can be misused to manipulate findings and fit of measurement and structural models. Researchers may become more interested in the tool than in the subject matter.Originality/valueUsing SEM does not elevate research to be science or make the researcher a scientist! An area of concern for further debate is whether the widespread and variable applications of SEM in social science (in particular, marketing) research are all science.

With growing attention to societal issues and implications of synthetic biology, we investigate sources of social science publication knowledge in synthetic biology and probe what might be learned by comparison with earlier rounds of social science research in nanotechnology. “Social science” research is broadly defined to include publications in conventional social science as well as humanities, law, ethics, business, and policy fields. We examine the knowledge clusters underpinning social science publications in nanotechnology and synthetic biology using a methodology based on the analysis of cited references. Our analysis finds that social science research in synthetic biology already has traction and direction, rooted in an ethical, legal, and social implications framework. However, compared with nanotechnology, social science research in synthetic biology could further explore opportunities and openings for engagement, anticipatory, and downstream application perspectives that will help to build a wider platform for insights into current and future societal impacts.

The Society of Behavioral Medicine (SBM) Leadership Institute provides leadership training for mid-career behavioral medicine scientists. As an institute fellow, I explored the experiences of social scientists working in healthcare settings. Exploratory research to describe the experiences, barriers, and facilitators social scientists face working in a healthcare delivery and research system. Our research questions were exploratory, focused on understanding (a) the experiences of social scientists working in a healthcare delivery and research system, (b) barriers that social scientists face working in a healthcare delivery and research system, (c) perceived value of sociological and anthropologic methods to clinicians and other health researchers, and (4) to understand the landscape of organizational models that feature social scientists as part of a larger scientific enterprise. We developed a plan for needs assessment activities. We conducted a design workshop of social scientists (n = 11). Using human-centered design methods, we elicited experiences and asked participants to develop a prototype model where social scientists could find career fulfillment with minimal structural barriers. We found that social scientists feel a strong sense of pride in their social science training and passion for the work they do. They reported occupying multiple disciplines at once, including multiple promotion and reputational expectations. Participants commonly reported that they themselves, as well as collaborators, viewed social science work, particularly qualitative research, as supplemental to quantitative work. Participants described perceptions that social science research is less rigorous or requires less training compared with other disciplines. Participants identified several contextual factors that influenced the perceived value of social science research in healthcare, including the focus of external funding priorities on larger clinical questions, organizational structure, and tangible leadership support of social science. The SBM Leadership Institute created a space for preliminary research into an area that could improve the experiences and research impact of social and behavioral scientists.

Nowadays, an abundance of short text is being generated that uses nonstandard writing styles influenced by regional languages. Such informal and code-switched content are under-resourced in terms of labeled datasets and language models even for popular tasks like sentiment classification. In this work, we (1) present a labeled dataset called MultiSenti for sentiment classification of code-switched informal short text, (2) explore the feasibility of adapting resources from a resource-rich language for an informal one, and (3) propose a deep learning-based model for sentiment classification of code-switched informal short text. We aim to achieve this without any lexical normalization, language translation, or code-switching indication. The performance of the proposed models is compared with three existing multilingual sentiment classification models. The results show that the proposed model performs better in general and adapting character-based embeddings yield equivalent performance while being computationally more efficient than training word-based domain-specific embeddings.

The massive expansion of scientific literature on climate change1 poses challenges for global environmental assessments and our understanding of how these assessments work. Big data and machine learning can help us deal with large collections of scientific text, making the production of assessments more tractable, and giving us better insights about how past assessments have engaged with the literature. We use topic modelling to draw a topic map, or topography, of over 400,000 publications from the Web of Science on climate change. We update current knowledge on the IPCC, showing that compared with the baseline of the literature identified, the social sciences are in fact over-represented in recent assessment reports. Technical, solutions-relevant knowledge—especially in agriculture and engineering—is under-represented. We suggest a variety of other applications of such maps, and our findings have direct implications for addressing growing demands for more solution-oriented climate change assessments that are also more firmly rooted in the social sciences2,3. The perceived lack of social science knowledge in assessment reports does not necessarily imply an IPCC bias, but rather suggests a need for more social science research with a focus on technical topics on climate solutions. The rapid growth of climate change research presents challenges for IPCC assessments and their stated aim of being comprehensive, objective and transparent. Here the authors use topic modelling to map the climate change literature, and assess how well it is represented in IPCC reports.