DrugExBERT for Pharmacovigilance – A Novel Approach for Detecting Drug Experiences from User-Generated Content

2023 | International Conference on Information Systems | Citations: 0

Authors: Bohnen, Eva; Erlebach, Stefanie; Zimmermann, Steffen

Abstract: Pharmaceutical companies have to maintain drug safety through pharmacovigilance ... Expand

Abstract: Pharmaceutical companies have to maintain drug safety through pharmacovigilance systems by monitoring various sources of information about adverse drug experiences. Recently, user-generated content (UGC) has emerged as a valuable source of real-world drug experiences, posing new challenges due to its high volume and variety. We present DrugExBERT, a novel approach to extract adverse drug experiences (adverse reaction, lack of effect) and supportive drug experiences (effectiveness, intervention, indication, and off-label use) from UGC. To be able to verify the extracted drug experiences, DrugExBERT additionally provides explications in the form of UGC phrases that were critical for the extraction. In our evaluation, we demonstrate that DrugExBERT outperforms state-of-the-art pharmacovigilance approaches as well as ChatGPT on several performance measures and that DrugExBERT is data-and drug-agnostic. Thus, our novel approach can help pharmaceutical companies meet their legal obligations and ethical responsibility while ensuring patient safety and monitoring drug effectiveness. Collapse

Topics: health information system application programming interface generative AI database system

Methods: chatGPT natural language processing RoBERTa BERT machine learning

Stand for Something or Fall for Everything: Predict Misinformation Spread with Stance-Aware Graph Neural Networks

2023 | International Conference on Information Systems | Citations: 0

Authors: Chen, Zihan; Sun, Jingyi; Liu, Rong Emily; Mai, Feng

Abstract: Although pervasive spread of misinformation on social media platforms has becom ... Expand

Abstract: Although pervasive spread of misinformation on social media platforms has become a pressing challenge, existing platform interventions have shown limited success in curbing its dissemination. In this study, we propose a stance-aware graph neural network (stance-aware GNN) that leverages users' stances to proactively predict misinformation spread. As different user stances can form unique echo chambers, we customize four information passing paths in stance-aware GNN, while the trainable attention weights provide explainability by highlighting each structure's importance. Evaluated on a realworld dataset, stance-aware GNN outperforms benchmarks by 32.65% and exceeds advanced GNNs without user stance by over 4.69%. Furthermore, the attention weights indicate that users' opposition stances have a higher impact on their neighbors' behaviors than supportive ones, which function as social correction to halt misinformation propagation. Overall, our study provides an effective predictive model for platforms to combat misinformation, and highlights the impact of user stances in the misinformation propagation. Collapse

Topics: misinformation Twitter filter bubble social media social influence

Methods: machine learning RoBERTa experiment deep learning graph neural network

Text Analytics

2023 | HICSS | Citations: 0

Authors: Cogburn, D.; Hine, M.; Yoon, V.

Abstract: This virtual minitrack for HICSS-56 recognizes that most global collaboration s ... Expand

Abstract: This virtual minitrack for HICSS-56 recognizes that most global collaboration systems, social media, and information systems of all types, generate enormous amounts of textual data, including system logs, email archives, websites, blog posts, meeting transcripts, speeches, annual reports, published material, and social media posts.While this structured, semistructured, and unstructured textual data is readily available, it presents tremendous challenges to researchers trying to analyze these large bodies of text with traditional methods. Text mining in big data analytics is an increasingly important technique for an interdisciplinary group of scholars, practitioners, government officials, and international organizations. This minitrack explores the tools, techniques, and insights generated from the analysis of text data. Collapse

Topics: application programming interface

Methods: natural language processing RoBERTa experiment question answering classification method

Automated Deductive Content Analysis of Text: A Deep Contrastive and Active Learning Based Approach

2023 | International Conference on Information Systems | Citations: 0

Authors: De Oliveira Silveira, Alysson; Ray, Arindam; Ebrahimi, Mohammadreza; Bhattacherjee, Anol

Abstract: Content analysis traditionally involves human coders manually combing through t ... Expand

Abstract: Content analysis traditionally involves human coders manually combing through text documents to search for relevant concepts and categories. However, this approach is time-intensive and not scalable, particularly for secondary data like social media content, news articles, or corporate reports. To address this problem, the paper presents an automated framework called Automated Deductive Content Analysis of Text (ADCAT) that uses deep learning-based semantic techniques, an ontology of validated construct measures, a large language model, human-in-the-loop disambiguation, and a novel augmentation-based weighted contrastive learning approach for improved language representations, to build a scalable approach for deductive content analysis. We demonstrate the effectiveness of the proposed approach to identify firm innovation strategies from their 10-K reports to obtain inferences reasonably close to human coding. Collapse

Topics: knowledge representation human in the loop human computer interaction innovation management organizational innovation

Methods: ontological modelling machine learning language model large language model qualitative content analysis

Incorporating Multiple Knowledge Sources for Targeted Aspect-based Financial Sentiment Analysis

2023 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Du, Kelvin; Xing, Frank; Cambria, Erik

Abstract: Combining symbolic and subsymbolic methods has become a promising strategy as re ... Expand

Abstract: Combining symbolic and subsymbolic methods has become a promising strategy as research tasks in AI grow increasingly complicated and require higher levels of understanding. Targeted Aspect-based Financial Sentiment Analysis (TABFSA) is an example of such complicated tasks, as it involves processes like information extraction, information specification, and domain adaptation. However, little is known about the design principles of such hybrid models leveraging external lexical knowledge. To fill this gap, we define anterior, parallel, and posterior knowledge integration and propose incorporating multiple lexical knowledge sources strategically into the fine-tuning process of pre-trained transformer models for TABFSA. Experiments on the Financial Opinion mining and Question Answering challenge (FiQA) Task 1 and SemEval 2017 Task 5 datasets show that the knowledge-enabled models systematically improve upon their plain deep learning counterparts, and some outperform state-of-the-art results reported in terms of aspect sentiment analysis error. We discover that parallel knowledge integration is the most effective and domain-specific lexical knowledge is more important according to our ablation analysis. Collapse

Topics: microblog domain-specific language Twitter social media knowledge graph

Methods: sentiment analysis RoBERTa transformer model BERT XLNet

Dissecting AI-Generated Fake Reviews: Detection and Analysis of GPT-Based Restaurant Reviews on Social Media

2023 | International Conference on Information Systems | Citations: 0

Authors: Gambetti, Alessandro; Han, Qiwei

Abstract: Recent advances in generative models such as GPT may be used to fabricate indis ... Expand

Abstract: Recent advances in generative models such as GPT may be used to fabricate indistinguishable fake customer reviews at a much lower cost, posing challenges for social media platforms to detect this kind of content. This study addresses two research questions: (1) the effective detection of AI-generated restaurant reviews generated from high-quality elite authentic reviews, and (2) the comparison of out-of-sample predicted AI-generated reviews and authentic reviews across multiple dimensions of review, user, restaurant, and content characteristics. We fine-tuned a GPT text detector to predict fake reviews, significantly outperforming existing solutions. We applied the model to predict non-elite reviews that already passed the Yelp filtering system, revealing that AI-generated reviews typically score higher ratings, users posting such content have less established Yelp reputations and AI-generated reviews are more comprehensible and less linguistically complex than human-generated reviews. Notably, machine-generated reviews are more prevalent in low-traffic restaurants in terms of customer visits. Collapse

Topics: fake review artificial intelligence application programming interface open source pandemic

Methods: machine learning survey computational algorithm sensitivity analysis ANOVA

“Maybe You Should Talk to Someone”: The Role of Online Communities on Mental Healthcare

2023 | International Conference on Information Systems | Citations: 0

Authors: Ghosh, Soham; Chakraborty, Soumyakanti; Sharma, Megha; Susarla, Anjana

Abstract: Online Health Communities like YouTube offer mental health patients an alternat ... Expand

Abstract: Online Health Communities like YouTube offer mental health patients an alternative channel to learn about mental illnesses, the treatment path to follow, and to share their experiences. For many patients who are reluctant to seek professional help, a video on mental health uploaded by a content creator may serve as a substitute for a counsellor. Our work aims to develop an understanding of the relationship between language formality and social support and provide normative guidelines for content creators on social media platforms. Using two transformer-based deep learning classification models, we determine the degree of language formality or informality present in the content, and three dimensions of social support in the comments. We then utilize propensity score estimation to establish the causal effect of (in)formality on the dimensions of social support for 994 videos and 3,10,157 comments. Our findings indicate that informal speech increases emotional support, leading to better health outcomes. Collapse

Topics: YouTube social media electronic mail blog

Methods: deep learning RoBERTa BERT propensity score matching propensity score method

Theories: speech act theory

FINFLUENCERS: OPINION MAKERS OR OPINION FOLLOWERS?

2023 | European Conference On Information Systems | Citations: 0

Authors: Haase, Frederic; Rath, Oliver; Kurka, Marco; Schoder, Detlef

Abstract: This paper explores the concept of Finfluencers: financial social network actor ... Expand

Abstract: This paper explores the concept of Finfluencers: financial social network actors with high potential social influence. Our research aims to clarify whether Finfluencers drive or are influenced by the broader social network sentiment, thereby establishing their role as either opinion makers or opinion followers. Using a dataset of 71 million tweets focusing on stocks and cryptocurrencies, we grouped actors by their social networking potential (SNP). Next, we derived sentiment time series using state-ofthe-art sentiment models and applied the technique of Granger causality. Our findings suggest that the sentiment of Finfluencer actors on Twitter has short-term predictive power for the sentiment of the larger group of actors. We found stronger support for cryptocurrencies in comparison to stocks. From the perspective of financial market regulation, this study emphasizes the relevance of understanding sentiment on social networks and high social influence actors to anticipate scams and fraud. Collapse

Topics: Twitter social network social influence cryptocurrency social media

Methods: longitudinal research time series analysis Granger causality descriptive statistic sentiment analysis

Detection of Stock Manipulation Influencer Content using Supervised Learning

2023 | International Conference on Information Systems | Citations: 0

Authors: Haase, Frederic; Rath, Oliver; Lauten, Julia; Schoder, Detlef

Abstract: In recent years, social media influencers have emerged as key players in stock ... Expand

Abstract: In recent years, social media influencers have emerged as key players in stock manipulation schemes. Despite their growing impact, methods to detect such activities remain scarcely explored. In this study, we examine the social media content of stock manipulation influencers (SMIs) implicated in a $100 million fraud case by the U.S. Securities and Exchange Commission (SEC) in 2022. Leveraging natural language processing (NLP) techniques, we first investigate the linguistic characteristics present in the social media content published by SMIs. Next, we develop and evaluate supervised learning models to detect manipulative content. Our results have significant implications for investors, regulators, and the broader financial community. They reveal the unique linguistic characteristics of SMI content and demonstrate the potential of machinelearning and deep-learning-based techniques in advancing fraud detection systems. Collapse

Topics: Twitter social media fraud detection database system social media analytics

Methods: machine learning RoBERTa DistilBERT computational algorithm natural language processing

Catching the Viewer's Eye: Examining Exploration and Exploitation Strategies in the Live Streaming Market

2023 | International Conference on Information Systems | Citations: 0

Authors: Kwon, Sung Hyun; Hann, Il-Horn

Abstract: Live streaming has become an important feature on social media platforms like F ... Expand

Abstract: Live streaming has become an important feature on social media platforms like Facebook and Instagram. More and more entrepreneurs (content creators) are competing in live streaming platforms like Twitch to maximize the attention they receive from consumers (viewers). In this competitive landscape, it is crucial for entrepreneurs to develop and provide new compelling content that can maximize the consumers' attention and aid the discovery of their content. We adopt an exploration-exploitation framework and assess the four strategies these new entrepreneurs could use to attract viewership and position themselves on Twitch: exploration, exploitation, learning from viewers, and their neighbor streamers. We combined the natural language processing techniques with theory-driven measures to accomplish this. Using our proposed measures, we estimate the utility of consumers from these different strategies using the discrete choice demand model. Collapse

Topics: Twitch database system decision making wiki online game

Methods: natural language processing RoBERTa transformer model BERT language model