Dataset Information

LOCO: The 88-million-word language of conspiracy corpus.

ABSTRACT: The spread of online conspiracy theories represents a serious threat to society. To understand the content of conspiracies, here we present the language of conspiracy (LOCO) corpus. LOCO is an 88-million-token corpus composed of topic-matched conspiracy (N = 23,937) and mainstream (N = 72,806) documents harvested from 150 websites. Mimicking internet user behavior, documents were identified using Google by crossing a set of seed phrases with a set of websites. LOCO is hierarchically structured, meaning that each document is cross-nested within websites (N = 150) and topics (N = 600, on three different resolutions). A rich set of linguistic features (N = 287) and metadata includes upload date, measures of social media engagement, measures of website popularity, size, and traffic, as well as political bias and factual reporting annotations. We explored LOCO's features from different perspectives showing that documents track important societal events through time (e.g., Princess Diana's death, Sandy Hook school shooting, coronavirus outbreaks), while patterns of lexical features (e.g., deception, power, dominance) overlap with those extracted from online social media communities dedicated to conspiracy theories. By computing within-subcorpus cosine similarity, we derived a subset of the most representative conspiracy documents (N = 4,227), which, compared to other conspiracy documents, display prototypical and exaggerated conspiratorial language and are more frequently shared on Facebook. We also show that conspiracy website users navigate to websites via more direct means than mainstream users, suggesting confirmation bias. LOCO and related datasets are freely available at https://osf.io/snpcg/ .

SUBMITTER: Miani A

PROVIDER: S-EPMC8545361 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundThere has been increasing interest in machine learning based natural language processing (NLP) methods in radiology; however, models have often used word embeddings trained on general web corpora due to lack of a radiology-specific corpus.PurposeWe examined the potential of Radiopaedia to serve as a general radiology corpus to produce radiology specific word embeddings that could be used to enhance performance on a NLP task on radiological text.Materials and methodsEmbeddings of dimension 50, 100, 200, and 300 were trained on articles collected from Radiopaedia using a GloVe algorithm and evaluated on analogy completion. A shallow neural network using input from either our trained embeddings or pre-trained Wikipedia 2014 + Gigaword 5 (WG) embeddings was used to label the Radiopaedia articles. Labeling performance was evaluated based on exact match accuracy and Hamming loss. The McNemar's test with continuity and the Benjamini-Hochberg correction and a 5×2 cross validation paired two-tailed t-test were used to assess statistical significance.ResultsFor accuracy in the analogy task, 50-dimensional (50-D) Radiopaedia embeddings outperformed WG embeddings on tumor origin analogies (p < 0.05) and organ adjectives (p < 0.01) whereas WG embeddings tended to outperform on inflammation location and bone vs. muscle analogies (p < 0.01). The two embeddings had comparable performance on other subcategories. In the labeling task, the Radiopaedia-based model outperformed the WG based model at 50, 100, 200, and 300-D for exact match accuracy (p < 0.001, p < 0.001, p < 0.01, and p < 0.05, respectively) and Hamming loss (p < 0.001, p < 0.001, p < 0.01, and p < 0.05, respectively).ConclusionWe have developed a set of word embeddings from Radiopaedia and shown that they can preserve relevant medical semantics and augment performance on a radiology NLP task. Our results suggest that the cultivation of a radiology-specific corpus can benefit radiology NLP models in the future.

Project description:BackgroundWord embeddings have been prevalently used in biomedical Natural Language Processing (NLP) applications due to the ability of the vector representations being able to capture useful semantic properties and linguistic relationships between words. Different textual resources (e.g., Wikipedia and biomedical literature corpus) have been utilized in biomedical NLP to train word embeddings and these word embeddings have been commonly leveraged as feature input to downstream machine learning models. However, there has been little work on evaluating the word embeddings trained from different textual resources.MethodsIn this study, we empirically evaluated word embeddings trained from four different corpora, namely clinical notes, biomedical publications, Wikipedia, and news. For the former two resources, we trained word embeddings using unstructured electronic health record (EHR) data available at Mayo Clinic and articles (MedLit) from PubMed Central, respectively. For the latter two resources, we used publicly available pre-trained word embeddings, GloVe and Google News. The evaluation was done qualitatively and quantitatively. For the qualitative evaluation, we randomly selected medical terms from three categories (i.e., disorder, symptom, and drug), and manually inspected the five most similar words computed by embeddings for each term. We also analyzed the word embeddings through a 2-dimensional visualization plot of 377 medical terms. For the quantitative evaluation, we conducted both intrinsic and extrinsic evaluation. For the intrinsic evaluation, we evaluated the word embeddings' ability to capture medical semantics by measruing the semantic similarity between medical terms using four published datasets: Pedersen's dataset, Hliaoutakis's dataset, MayoSRS, and UMNSRS. For the extrinsic evaluation, we applied word embeddings to multiple downstream biomedical NLP applications, including clinical information extraction (IE), biomedical information retrieval (IR), and relation extraction (RE), with data from shared tasks.ResultsThe qualitative evaluation shows that the word embeddings trained from EHR and MedLit can find more similar medical terms than those trained from GloVe and Google News. The intrinsic quantitative evaluation verifies that the semantic similarity captured by the word embeddings trained from EHR is closer to human experts' judgments on all four tested datasets. The extrinsic quantitative evaluation shows that the word embeddings trained on EHR achieved the best F1 score of 0.900 for the clinical IE task; no word embeddings improved the performance for the biomedical IR task; and the word embeddings trained on Google News had the best overall F1 score of 0.790 for the RE task.ConclusionBased on the evaluation results, we can draw the following conclusions. First, the word embeddings trained from EHR and MedLit can capture the semantics of medical terms better, and find semantically relevant medical terms closer to human experts' judgments than those trained from GloVe and Google News. Second, there does not exist a consistent global ranking of word embeddings for all downstream biomedical NLP applications. However, adding word embeddings as extra features will improve results on most downstream tasks. Finally, the word embeddings trained from the biomedical domain corpora do not necessarily have better performance than those trained from the general domain corpora for any downstream biomedical NLP task.

Project description:Purpose Word learning difficulties have been documented in multiple studies involving children with dyslexia and developmental language disorder (DLD; see also specific language impairment). However, no previous studies have directly contrasted word learning in these two frequently co-occurring disorders. We examined word learning in second-grade students with DLD-only and dyslexia-only as compared to each other, peers with both disorders (DLD + dyslexia), and peers with typical development. We hypothesized that children with dyslexia-only and DLD-only would show differences in word learning due to differences in their core language strengths and weaknesses. Method Children (N = 244) were taught eight novel pseudowords paired with unfamiliar objects. The teaching script included multiple exposures to the phonological form, the pictured object, a verbal semantic description of the object, and spaced retrieval practice opportunities. Word learning was assessed immediately after instruction with tasks requiring recall or recognition of the phonological and semantic information. Results Children with dyslexia-only performed significantly better on existing vocabulary measures than their peers with DLD-only. On experimental word learning measures, children in the dyslexia-only and DLD + dyslexia groups showed significantly poorer performance than typically developing children on all word learning tasks. Children with DLD-only differed significantly from the TD group on a single word learning task assessing verbal semantic recall. Conclusions Overall, results indicated that children with dyslexia display broad word learning difficulties extending beyond the phonological domain; however, this contrasted with their relatively strong performance on measures of existing vocabulary knowledge. More research is needed to understand relations between word learning abilities and overall vocabulary knowledge and how to close vocabulary gaps for children with both disorders. Supplemental Material https://doi.org/10.23641/asha.14832717.

Dataset Information

LOCO: The 88-million-word language of conspiracy corpus.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets