Quantifying Intimacy in Language

Pei, Jiaxin; Jurgens, David

doi:10.18653/v1/2020.emnlp-main.428

Cited by 19 publications

(17 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Phi coefficient is 0.77 for the SINGLE-TWEET setting, showing a very strong positive correlation between profanities and abusive labels, while it is only moderate (0.39) in the CONTEXT configuration. This finding is in line with past works, showing that the offensiveness of swearing is context-dependent (Pamungkas et al, 2020) and that profanities are frequently used among people with a strong social relationship without any offensive goal (Bak et al, 2012), but rather to signal an informal attitude among the members of a community or even intimacy (Pei and Jurgens, 2020).…”

Section: Manual (Re)annotationsupporting

confidence: 91%

Abuse is Contextual, What about NLP? The Role of Context in Abusive Language Annotation and Detection

Menini¹,

Aprosio²,

Tonelli³

2021

Preprint

View full text Add to dashboard Cite

The datasets most widely used for abusive language detection contain lists of messages, usually tweets, that have been manually judged as abusive or not by one or more annotators, with the annotation performed at message level. In this paper, we investigate what happens when the hateful content of a message is judged also based on the context, given that messages are often ambiguous and need to be interpreted in the context of occurrence. We first re-annotate part of a widely used dataset for abusive language detection in English in two conditions, i.e. with and without context. Then, we compare the performance of three classification algorithms obtained on these two types of dataset, arguing that a context-aware classification is more challenging but also more similar to a real application scenario.'Abuse is contextual' is one of the key claims reported in (Prabhakaran et al., 2020), where the authors describe the content of a panel with NLP practitioners and human rights experts at the Human Rights Conference RightsCon 2020. A similar remark is made in (Jurgens et al., 2019) in a recent position paper on current NLP methods to fight online abuse, arguing that NLP research tends to consider a narrow scope of what constitutes abuse, without respecting, for instance, community norms in classification decisions. Similarly, (Vidgen et al., 2019) claim that one of the main research challenges concerning abusive language detection is accounting for context, although they focus mainly on user-level variables and network representations.In this work, we present an analysis aimed at better understanding what context is in abusive language detection, and what its effects are on data annotation and classification. We focus on discourse context, i.e. the messages preceding a given tweet, because it has been relatively understudied in abusive language detection, while user-level and network-level features have already been extensively discussed in past works (Fehn Unsvåg and

show abstract

Section: Manual (Re)annotationsupporting

confidence: 91%

Abuse is Contextual, What about NLP? The Role of Context in Abusive Language Annotation and Detection

Menini¹,

Aprosio²,

Tonelli³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Birth control users might analyze the risk and benefit of making these disclosures in certain settings and to certain audiences. While social penetration theory (Altman and Taylor 1973) posits that more disclosures are possible as social bonds deepen, prior work has also found that intimate language can be frequent among both close connections and strangers (but not in between) (Pei and Jurgens 2020). The variations we observe across methods, side effects, and sensemaking practices could be indicative of platform affordances for privacy, audience size, and anonymity, each of which can affect decisions to self-disclose.…”

Section: Discussionmentioning

confidence: 65%

Sensemaking About Contraceptive Methods Across Online Platforms

McDowall¹,

Antoniak²,

Mimno³

2023

Preprint

View full text Add to dashboard Cite

Selecting a birth control method is a complex healthcare decision. While birth control methods provide important benefits, they can also cause unpredictable side effects and be stigmatized, leading many people to seek additional information online, where they can find reviews, advice, hypotheses, and experiences of other birth control users. However, the relationships between their healthcare concerns, sensemaking activities, and online settings are not well understood. We gather texts about birth control shared on Twitter, Reddit, and WebMD-platforms with different affordances, moderation, and audiences-to study where and how birth control is discussed online. Using a combination of topic modeling and hand annotation, we identify and characterize the dominant sensemaking practices across these platforms, and we create lexicons to draw comparisons across birth control methods and side effects. We use these to measure variations from survey reports of side effect experiences and method usage. Our findings characterize how online platforms are used to make sense of difficult healthcare choices and highlight unmet needs of birth control users.

show abstract

“…We choose this method as it can produce more stable and fined-grained scores than directly scoring (Kiritchenko and Mohammad, 2017). We note similar methods have been applied to various tasks, including measuring offensiveness (Hada et al, 2021) and intimacy (Pei and Jurgens, 2020) in the computational linguistic literature.…”

Section: Data Collectionmentioning

confidence: 99%

Analyzing the Intensity of Complaints on Social Media

Fang¹,

Zong²,

Li³

et al. 2022

Preprint

View full text Add to dashboard Cite

Complaining is a speech act that expresses a negative inconsistency between reality and human expectations. While prior studies mostly focus on identifying the existence or the type of complaints, in this work, we present the first study in computational linguistics of measuring the intensity of complaints from text. Analyzing complaints from such perspective is particularly useful, as complaints of certain degrees may cause severe consequences for companies or organizations. We create the first Chinese dataset containing 3,103 posts about complaints from Weibo, a popular Chinese social media platform. These posts are then annotated with complaints intensity scores using Best-Worst Scaling (BWS) method. We show that complaints intensity can be accurately estimated by computational models with the best mean square error achieving 0.11. Furthermore, we conduct a comprehensive linguistic analysis around complaints, including the connections between complaints and sentiment, and a cross-lingual comparison for complaints expressions used by Chinese and English speakers. We finally show that our complaints intensity scores can be incorporated for better estimating the popularity of posts on social media. 1

show abstract

Quantifying Intimacy in Language

Cited by 19 publications

References 58 publications

Abuse is Contextual, What about NLP? The Role of Context in Abusive Language Annotation and Detection

Abuse is Contextual, What about NLP? The Role of Context in Abusive Language Annotation and Detection

Sensemaking About Contraceptive Methods Across Online Platforms

Analyzing the Intensity of Complaints on Social Media

Contact Info

Product

Resources

About