With the rising prominence of social media, users frequently supplement their written content with images. This trend has brought about new challenges in automatic processing of social media messages. In order to fully understand the meaning of a post, it is necessary to capture the relationship between the image and the text. In this work we address the two main objectives of the ImageArg shared task. Firstly, we aim to determine the stance of a multi-modal tweet toward a particular issue. We propose a strong baseline, fine-tuning transformer based models on concatenation of tweet text and image text. The second goal is to predict the impact of an image on the persuasiveness of the text in a multi-modal tweet. To capture the persuasiveness of an image, we train vision and language models on the data and explore other sets of features merged with the model, to enhance prediction power. Ultimately, both of these goals contribute toward the broader aim of understanding multi-modal messages on social media and how images and texts relate to each other.
With the increase of deception and misinformation especially in social media, it has become crucial to be able to develop machine learning methods to automatically identify deceptive language. In this proposal, we identify key challenges underlying deception detection in cross-domain, cross-lingual and multi-modal settings. To improve cross-domain deception classification, we propose to use inter-domain distance to identify a suitable source domain for a given target domain. We propose to study the efficacy of multilingual classification models vs translation for cross-lingual deception classification. Finally, we propose to better understand multi-modal deception detection and explore methods to weight and combine information from multiple modalities to improve multi-modal deception classification.
We present machine learning classifiers to automatically identify COVID-19 misinformation on social media in three languages: English, Bulgarian, and Arabic. We compared 4 multitask learning models for this task and found that a model trained with English BERT achieves the best results for English, and multilingual BERT achieves the best results for Bulgarian and Arabic. We experimented with zero shot, few shot, and target-only conditions to evaluate the impact of target-language training data on classifier performance, and to understand the capabilities of different models to generalize across languages in detecting misinformation online. This work was performed as a submission to the shared task, NLP4IF 2021: Fighting the COVID-19 Infodemic. Our best models achieved the second best evaluation test results for Bulgarian and Arabic among all the participating teams and obtained competitive scores for English.
In this paper we investigate the efficacy of using contextual embeddings from multilingual BERT and German BERT in identifying fact-claiming comments in German on social media. Additionally, we examine the impact of formulating the classification problem as a multi-task learning problem, where the model identifies toxicity and engagement of the comment in addition to identifying whether it is fact-claiming. We provide a thorough comparison of the two BERT based models compared with a logistic regression baseline and show that German BERT features trained using a multi-task objective achieves the best F1 score on the test set. This work was done as part of a submission to GermEval 2021 shared task on the identification of fact-claiming comments.
We address the problem of predicting psychiatric hospitalizations using linguistic features drawn from social media posts. We formulate this novel task and develop an approach to automatically extract time spans of self-reported psychiatric hospitalizations. Using this dataset, we build predictive models of psychiatric hospitalization, comparing feature sets, user vs. post classification, and comparing model performance using a varying time window of posts. Our best model achieves an F1 of .718 using 7 days of posts. Our results suggest that this is a useful framework for collecting hospitalization data, and that social media data can be leveraged to predict acute psychiatric crises before they occur, potentially saving lives and improving outcomes for individuals with mental illness.
We address the problem of automatic detection of psychiatric disorders from the linguistic content of social media posts. We build a large scale dataset of Reddit posts from users with eight disorders and a control user group. We extract and analyze linguistic characteristics of posts and identify differences between diagnostic groups. We build strong classification models based on deep contextualized word representations and show that they outperform previously applied statistical models with simple linguistic features by large margins. We compare user-level and post-level classification performance, as well as an ensembled multiclass model.
Humans rarely perform better than chance at lie detection. To better understand human perception of deception, we created a game framework, LieCatcher, to collect ratings of perceived deception using a large corpus of deceptive and truthful interviews. We analyzed the acoustic-prosodic and linguistic characteristics of language trusted and mistrusted by raters and compared these to characteristics of actual truthful and deceptive language to understand how perception aligns with reality. With this data we built classifiers to automatically distinguish trusted from mistrusted speech, achieving an F1 of 66.1%. We next evaluated whether the strategies raters said they used to discriminate between truthful and deceptive responses were in fact useful. Our results show that, although several prosodic and lexical features were consistently perceived as trustworthy, they were not reliable cues. Also, the strategies that judges reported using in deception detection were not helpful for the task. Our work sheds light on the nature of trusted language and provides insight into the challenging problem of human deception detection.
Most efforts at identifying abusive speech online rely on public corpora that have been scraped from websites using keyword-based queries or released by site or platform owners for research purposes. These are typically labeled by crowd-sourced annotators – not the targets of the abuse themselves. While this method of data collection supports fast development of machine learning classifiers, the models built on them often fail in the context of real-world harassment and abuse, which contain nuances less easily identified by non-targets. Here, we present a mixed-methods approach to create classifiers for abuse and harassment which leverages direct engagement with the target group in order to achieve high quality and ecological validity of data sets and labels, and to generate deeper insights into the key tactics of bad actors. We use women journalists’ experience on Twitter as an initial community of focus. We identify several structural mechanisms of abuse that we believe will generalize to other target communities.
We explore linguistic indicators of schizophrenia in Reddit discussion forums. Schizophrenia (SZ) is a chronic mental disorder that affects a person’s thoughts and behaviors. Identifying and detecting signs of SZ is difficult given that SZ is relatively uncommon, affecting approximately 1% of the US population, and people suffering with SZ often believe that they do not have the disorder. Linguistic abnormalities are a hallmark of SZ and many of the illness’s symptoms are manifested through language. In this paper we leverage the vast amount of data available from social media and use statistical and machine learning approaches to study linguistic characteristics of SZ. We collected and analyzed a large corpus of Reddit posts from users claiming to have received a formal diagnosis of SZ and identified several linguistic features that differentiated these users from a control (CTL) group. We compared these results to other findings on social media linguistic analysis and SZ. We also developed a machine learning classifier to automatically identify self-identified users with SZ on Reddit.
We explore deception detection in interview dialogues. We analyze a set of linguistic features in both truthful and deceptive responses to interview questions. We also study the perception of deception, identifying characteristics of statements that are perceived as truthful or deceptive by interviewers. Our analysis show significant differences between truthful and deceptive question responses, as well as variations in deception patterns across gender and native language. This analysis motivated our selection of features for machine learning experiments aimed at classifying globally deceptive speech. Our best classification performance is 72.74% F1-Score (about 17% better than human performance), which is achieved using a combination of linguistic features and individual traits.
Collecting spontaneous speech corpora that are open-ended, yet topically constrained, is increasingly popular for research in spoken dialogue systems and speaker state, inter alia. Typically, these corpora are labeled by human annotators, either in the lab or through crowd-sourcing; however, this is cumbersome and time-consuming for large corpora. We present four different approaches to automatically tagging a corpus when general topics of the conversations are known. We develop these approaches on the Columbia X-Cultural Deception corpus and find accuracy that significantly exceeds the baseline. Finally, we conduct a cross-corpus evaluation by testing the best performing approach on the Columbia/SRI/Colorado corpus.