Digital Narratives of Grief and Resilience: Insights from the Integrating Emotional Stories Online ( IESO) Platform

doi:10.21203/rs.3.rs-4959421/v1

Download PDF

Article

Digital Narratives of Grief and Resilience: Insights from the Integrating Emotional Stories Online ( IESO) Platform

https://doi.org/10.21203/rs.3.rs-4959421/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Social media has altered how humans express emotions, among them, grief. Features unique to social media platforms, such as persistence, replicability, scalability, and searchability have afforded social media users new spaces and larger communities in which to express deep, complex pain and find support. This interdisciplinary study, leveraging insights from social work, computer science, and linguistics, explores how grief and daily stressors are expressed, on a digital platform, among Black Harlem residents and students attending a large university in New York City. Particular attention is given to posts with features that reflect African American English and social media-specific language, seeking to understand challenges associated with interpreting these expressions; misinterpretation can lead to differential support and treatment for those seeking help. To examine the experience, we created the Integrating Emotional Stories Online (IESO) platform, to collect expressions of grief and everyday stress. The platform allowed participants to anonymously post and self-label their emotions; we analyzed posts using qualitative and computational topic modeling methods to identify core themes. We identified four primary themes: expressions of interpersonal grief, systemic grief, everyday stressors and self-care & joy. The computational topic modeling approach validated those themes. The study was conducted in the context of significant events such as the COVID-19 pandemic, protests against police brutality, and mass shootings. Findings reveal the complex ways grief and everyday stress may be communicated in digital space, emphasizing the importance of deep analysis and reflection when examining hyper-local dialects. Findings also underscore the importance of digital platforms in understanding and supporting individuals who experience grief and other forms of stress, providing valuable insights for mental health practitioners.

Health sciences/Medical research

Scientific community and society/Social sciences

Grief

Digital Platforms

Social Media

Black Harlem Community

Qualitative Methods

Computational Topic Modeling

COVID-19

Police Violence

Grief is a universally experienced, yet often misunderstood, emotion (Hall, 2014). Although many aspects of grieving are private, those that are grieving often desire to share the news of loss with others. In modern American history, this has occurred through death announcements in newspaper obituaries and church notices, on websites, and through word-of-mouth (de Vries & Rutherford, 2004). In contemporary society, however, much of how individuals communicate their experiences with grief has moved online (Moore et al., 2019).

Digital platforms, like social media, have reconfigured the experience of expressing grief, allowing for persistence, replicability, scalability, and searchability. These features, created by engineers and product designers, offer a relatively new forum for expressing experiences with deep pain, grief, and daily stressors, while also providing an opportunity to find and build support to cope. It is not always clear, however, when posts on social media are expressing grief and distress, and this makes it difficult to study what people are saying about their pain, particularly if they are speaking in hyper-local dialects and phrases.

The difficulty in interpreting underlying intentions in posts creates additional challenges for studying expressions of grief and daily stressors. For example, a teenager comes to a therapy session and presents their psychiatrist with a TikTok post that has triggered anxiety and depression. That post contains written text and audio that contains a mix of African American Language (AAL) and language artifacts specific to social media. The psychiatrist misunderstands the post and characterizes the content as violent or aggressive, when it actually intends to express grief and distress. Could this misunderstanding result in a different treatment choice?

To address these questions, we designed an interdisciplinary study using qualitative methods and computational topic modeling to identify categories of expressed grief and everyday stressors, following triggering, often traumatic, events—in particular, posts from Black Harlem community members and students attending a university in New York City. We created an interface, Integrating Emotional Stories Online (IESO), to allow participants to write anonymously about their reactions to daily events. IESO users actively used the site from February to December 2022, and posted 949 times in total. Current events during this period included the COVID-19 pandemic and ubiquitous losses, responses to George Floyd’s murder and protests against police brutality, and multiple mass shootings. IESO enabled users to self-label their posts, indicating their emotions and providing a level of interpretability not typically available on social media platforms. In this study, self-labels enable analysis of how people feel when they’re expressing themselves.

We present four themes related to how users express grief, within the context of Covid-19 and ongoing police violence, on the IESO platform: interpersonal grief, institutional & systemic grief, everyday stressors, and self-care & joy. In addition, we use the computational approach topic modeling (Deerwester et al., 1990; Blei et al., 2003) to validate our qualitative themes and possibly discover new themes.

2.1 Qualitative Results

In analyzing the 949 IESO posts in this dataset, social work graduate student annotators identified four key themes defined and described below: Interpersonal Grief (115 posts), Institutional and Systemic Losses (48 posts), Everyday Stressors and Mental Health (462 posts), and Self Care and Joy (278 posts). All themes show strong or almost perfect agreement amongst coders, except for Institutional and Systemic Losses, which shows moderate agreement (Cohen, 1960). The average agreement across all thematic categories is strong. In terms of demographic data, a majority of examples¹ chosen (44.4%) were written by Black or African-American IESO users. Hispanic or Latino users wrote 33.3%, while white users wrote 11.1%. Asian users wrote 5.6%, and users who did not wish to disclose their identity wrote another 5.6%. For the examples of IESO posts, different participants are labeled with numbers, i.e. ‘Participant 1’ and ‘Participant 2’ to maintain anonymity. Some participants are referenced multiple times; they are referred to by the initial number.

2.1.1 Interpersonal Grief

The posts coded as interpersonal grief consist of three subthemes: adaptive grief, expressions of longing, and avoidance of emotions. All posts that were coded as Interpersonal grief discuss grief due to the loss of a loved one. Posts coded adaptive grief consist of users discussing grief from an adaptive perspective, i.e. learning to live with loss and integrate it into their lives. Posts coded expressions of longing consist of posts that express yearning and longing for lost loved ones, sometimes manifesting in feelings of helplessness or distress. Posts coded avoidance of emotions discussed emotions, such as helplessness, longing, and sadness, that users attempted to avoid.

2.1.1.1 Adaptive Grief

Adaptive responses to grief typically reflect users figuring out ways to manage their emotions and attempt to make meaning of their losses. We define adaptive grief as the ways people who have lost someone attempt to integrate that loss into their lives, and adapt to it. Responses can be positive or gratifying, as well as upsetting or troubling, but often involve a sense of acceptance of a loss. Participant 1 writes:

I’ve had better days. This is one of the days I’m not feeling my best. Still lost for words. At this point I’m just existing. But I suppose I’m stronger than I realize, I’m still continuing my everyday life working, working out, going to school. But time heals everything, I suppose I just give it time.

While acknowledging the pain they are currently feeling, as well as their state of “just existing,” Participant 1 still acknowledges their resilience, and the faith that time will heal their grief. This user also discusses continuing their everyday routines and maintaining a sense of normalcy even in the face of loss.

A different user, Participant 2, finds purpose through their grief:

Recently I lost a friend and during his funeral the pastor express the importance of living life. He said, God not oh put you on this earth for you but for you to reach and make a change in everyone’s lives you crossed path with. Through living you are fulfilling your purpose on earth so when your no longer here the people you met are left with lifetime memories and lessons that will stick with you forever. put you This very part of the sermon stuck out to me to most. It has given me deeper understanding of life and has giving me some sort of comfort and help with coming to reality of his death. This will help me with comforting my brother in law who lost his best-friend.

In the post above, the user notes that they were able to not only accept the reality of their friend’s death, but see greater purpose in their friend’s life. They cite the importance of the sermon given by the pastor at their friend’s funeral service, providing them with a sense of comfort. They integrated lessons about coming to terms with reality of a loved one’s death and used them to comfort someone else who is now grieving.

Significantly, these adaptive responses to grief demonstrate how people learn to persevere, do not let grief overtake them, and use coping skills to aid others through inevitable future losses.

2.1.1.2 Expressions of Longing

Expressions of longing reflect an individual's sense of yearning for loved ones they have lost. We include in this definition yearning-adjacent emotions that may accompany grief. These emotions include feelings of disorganization and despair that may be highlighted by feelings of anger, sadness, or hopelessness, in addition to guilt, high levels of stress, and a lack of control.

The following are three separate expressions of longing:

Participant 3: “My dads passing was probably the worst pain I’ve ever felt in my entire life. And sometimes I feel okay because I think about all the good times we shared. We were really close. And ever since he passed my life just been so wierd.”

Participant 4: “Overwhelmed. When death comes without warning. This month has been so uncertain day to day emotion. Feeling like I am in a closed box. Mentally afraid of the future. Questioning my self and the relationships that I have with those that is alive. I am doing enough to show that I care or love them? Knowing that death is irreversible. Searching for inner peace and mediating is helping me cope with all these thoughts.”

Participant 5: “I feel sad and guilty because my grandmother passed away and I did not get a chance to say goodbye to her…My parents were on their way to pick me up when she passed, and they couldn't say goodbye either. I'm worried that my parents will hold this against me for the rest of my life…I'm devastated that I won't be able to see her again and there was nothing I could do in lifetime to help her.”

These posts demonstrate the emotional intensity, including overwhelm and helplessness, that IESO users experienced when navigating death. As Participant 3 noted, losing their father was the “worst pain” they have ever felt, highlighting their intense loss, hopelessness, and despair. Moreover, grief is a catalyst for people to confront their own mortality, as noted by Participant 4,“[Am I] doing enough to show that I care or love them?,” regarding others in the user’s life. This reflection illustrates how death can catalyze people to examine their relationships and actions and think critically about what they might want to change.

When analyzing posts surrounding interpersonal grief, which comprised only of posts regarding the loss of a loved one, it was clear that there were other forms of mourning and deep pain expressed on IESO that is not culturally or psychologically recognized as traditional grief. Some grief expressed on IESO has a less clearly culturally recognized path for mourning, including topics such as miscarriages, the loss of a pet, heartbreak, as well as systemic losses. Participant 7 discusses the lack of societal understanding for ambiguous losses: “I lost a baby couple years ago and I refuse to forget. I feel since miscarriages are common. People dont understand the loss a mother will forever feel” By highlighting the grief caused by miscarriage, this poster not only discusses their own yearning and pain, but also argues that society does not understand the true ramifications of miscarriages. This platform was a space for users to discuss the lack of societal understanding surrounding these ambiguous or non-traditional losses as well.

2.1.1.3 Avoidance of Emotions

IESO users cope with grief in various ways, including avoiding the difficult feelings surrounding their loss. We define avoidance of emotions as participants portraying the need to evade, distract from, or escape their intense and often negative feelings surrounding a loss. Participant 1 writes about averting intense emotionality and keeping busy in order to redirect themselves from feeling feelings of grief:

Not feeling like myself this morning. I woke up thinking about my friend that passed. I think about them every day. I try to stay busy to avoid thinking about them. But I can’t get my mind off it.

Further reinforcing the desire to distract from grief, Participant 6 posts about the pain associated with being reminded of the person they lost: “I don't want to be around people who remind me of her, its too painful right now. One day they'll bring me joy but right now it's too much.”

Another user describes the somatic experience of dissociation, or detaching from one’s environment in order to avoid their emotions: “sometimes i feel like a still picture and everything around me is going”

Given the intensity and range of emotions that can accompany grief, some IESO users indicate attempting to avert the pain of these intense emotions by avoiding people and things that trigger their grief. These users posted expressions of pain, intense emotionality, and consequent desire to avoid or redirect emotions, especially when experiencing a volume of grief that exceed normative social boundaries.

2.1.2 Institutional and Systemic Losses (48 posts)

Posts categorized as institutional and systemic losses address societal loss, hardship, and tragedy. Within this content, we identified four sub-themes: mass shootings, police brutality, COVID-19, and climate change. Posts assigned to mass shootings discussed school shootings, mass gun violence, and other shootings directly mentioned by IESO users. Posts assigned to police brutality discussed violence, specifically racialized violence, at the hands of police. Posts assigned to COVID-19 discussed implications of COVID-19, and often dovetailed with other stressors including emotional and financial distress. Lastly, posts assigned to climate change discussed environmental change, as well as potential loss from climate change. Some IESO users describe these events as stressors and catalysts for grief in their lives. Additionally, users worked to make meaning of large-scale losses that may not have affected them personally, but that had larger institutional implications. Users’ posts about institutional and systemic losses weave personal feelings and perceptions into the broader tapestry of social challenge…and often discuss a shared lack of agency to address social issues. By contextualizing personal sentiments within systemic dynamics, users emphasize the interconnected nature of individual experiences and the urgent need for comprehensive solutions and change that address the root causes of institutional and systemic loss.

2.1.2.1 Mass Shootings

IESO users posted their reactions to mass shootings, condemning them and calling for changes to both current legislation and the power structures that allow these events to happen. Posters usually describe the age of the victims and their shock at the seeming inevitability of mass shootings. At two separate times, Participant 8 posted about school shootings, and their distress as a result of school shooting in the U.S. First, they reflected:

The school shooting have me worried what is happening to US how can this keep happening. The thought of someone dropping off their child and to pick them up and they not alive is overwhelming. How can one cope with this hard burden? Lost for words!

Participant 8, referencing a separate incident, posted: “Disgusted. The shooting in Albany. I watched the video. To know that someone is actually just shooting real human beings is HORRIBLE! Destroyed families.”²

Participant 9 noted:

I've been thinking a lot about systemic violence recently. It probably has to do with all the recent mass shootings. The recent elementary school shooting in Texas is especially horiffic. I feel powerless. All we can do is pray for them. I'm sick of this happening. And I don't even know the kids.”³

Users’ posts reflect worry about the reality of gun violence, and children dying in mass shootings. Users expressed intense emotions of disgust towards people who enact gun violence, and horror that systems and institutions charged to protect people are failing. Users also expressed feelings of outrage for those injured and killed, and lack of hope that anything will change. These responses showcase IESO users’ intense feelings of anguish, hopelessness, and distress due to mass shootings and mass violence, not just interpersonal losses.

2.1.2.2 Police Brutality

Police brutality also appeared in posts about institutional and systemic loss. Like posts referencing mass shootings, users express feelings of anger and frustration, responding to the violence that police exert on communities of color. In two separate posts, Participant 10 discussed feeling upset and outraged regarding racially motivated police shootings: “I was thinking about All the African men, woman, and children who was killed by police officers. It seems that they have a license to kill black and brown people.” In a separate post, Participant 10 reflected: “[I’m] upset. A police officer shot a teenager because he thought the water gin was a real gun.”

Similarly, Participant 11 discussed police brutality: “I feel angry. There is a real war between the people and police it’s really needs to stop”

IESO users’ posts about mass shootings are concerned with systemic issues of violence and the government’s perceived lack of response. Similarly, in posts about police brutality, users are particularly concerned with the systemic issue of racism. They acknowledge that race, and race-based loss, underlie police brutality and emerge from systemic racism. Although perhaps not directly affected by mass shootings and police brutality, IESO users are still grappling with intense emotions of fear, disgust, anger, and helplessness in the face of violence and ubiquitous grief—especially compounded with other events, including COVID-19.

2.1.2.3 COVID-19 Pandemic

IESO users’ posts about the COVID-19 pandemic focus primarily on concern about not receiving adequate resources to cope with losses, and anxiety related to isolation and further spread of the virus. We include three separate posts that represent reflections on lockdown experiences, as well as current distresses about contracting the disease and its long-term effects:

Participant 12 states, “I feel at times that I cant beat the feeling I have of being worried. I contracted covid in 2020 and ever since I did, my body hasn't been the same. I almost died and so did my fiancee in 2020 due to covid. It was very traumatic and I just hope we live a long life because the after-effects is what worries me.”

Participant 9, who discussed systemic violence, also referenced feelings of frustration at the limited cultural discussion and memorialization of COVID-19:

It's strange how we have all moved on from COVID. Like it was this massive traumatic event that had killed millions of people. But we are conditioned to move on from it. Did we just forget about all the victims? How are we supposed to make sense of the scale of this disaster?

This user attempts to make meaning from large-scale loss, and feels as though society is pretending like nothing happened, rather than face, and integrate, the grief stemming from lives lost from COVID. Participant 9 expresses incredulousness that there has been such little memorialization for those who have passed, as well as a lack of institutionally-recognized meaning-making for those who have survived them.

Participant 13 reflects:

[I’m] exasperated and tired. Hearing about the new BA variant for COVID arising, especially in LA, really makes me so frustrated. I feel like everyone, including myself, has loosened up a lot about COVID and everything here seems more or less back to normal with COVID cases and daily life. However, just as it seems like the worst is over and we can continue to live life, a new variant shows up….I really can’t deal with another lockdown.

Participant 13’s post reflects other users’ posts surrounding the coronavirus– they are striving to comprehend the ongoing changes and risks associated with the COVID-19 virus. As they do so, IESO users express anxiety, exasperation, and frustration concerning COVID-19’s destabilizing effects, including mourning the lives lost and worrying about the future. Specific concerns about the future include how society moves forward after experiencing a collective trauma. Some users address coping with the trauma of nearly dying from the virus themselves, while others discuss feeling fatigued and exasperated from hearing about consistently rising numbers of reported cases. Participant 13 mentions that they cannot deal with another lockdown, highlighting the mental and emotional strain of isolation, and their fear that new variants will continue to emerge.

2.1.2.4 Climate Change

IESO users post about systemic issues, such as distress regarding climate change, where users lack a feeling of individual agency, and express the unknown impact it will have on their own futures.

Participant 10, in addition to discussing COVID and systemic violence, also discussed climate change and a sense of distress about potential loss from climate change:

Is climate change grief a real thing? Like this overwhelming, deep sense of forthcoming doom. I don't know because it's not like I have personally lost anything yet. But I feel for others.

Participant 14 reflects:

I kind of want kids but right now im not really sure. obviously im young and have time to decide but at this point, I dont think the world is a safe place in general. climate change, racism, poverty, it just seems like an awful idea to have kids

In their posts, users note that, although they have not experienced interpersonal loss from climate change, they struggle to envision their own futures given the high potential for loss and crisis. Climate anxiety is depicted via language that expresses doubt and uncertainty regarding the future –or potential lack thereof– users anticipate. The presence of anticipatory ecological grief affects how users imagine their futures, what choices they make, and what actions they take.

With these posts about institutional and systemic losses, users broaden the scope of what is typically considered to be grief to examine structural issues. Participant 15, discusses current events and exposure to such events on social media:

It’s interesting how long-term exposure and media can desensitize us to violence and issues in our world, so in thinking about why myself and others maybe not be as bothered by headlines sometimes, I feel guilty and contemplative about this. I wonder if it’s a way of dealing with grief? Perhaps because it is too much to constantly get berated with these events and have it deeply wound us and turn us melancholic all the time, so eventually people get tired of caring—in a way.

With this post, Participant 15 summarized a major theme that emerged in the data: desensitization to current events, such as climate change and ubiquitous grief due to COVID, results in emotional fatigue.

By framing personal experiences within the context of institutional problems, users contribute to a narrative that advocates systemic change. Like the meaning-making that occurs in posts regarding interpersonal grief, users confronted with large-scale losses appear to be reminded of their own mortality, prompting additional reflection.

2.1.3 Everyday Stressors and Mental Health (462 posts)

Posts assigned to the theme Everyday Stressors and Mental Health include a diverse range of challenges affecting mental wellbeing in daily life. Within this theme, three subthemes were identified: masking emotions, desire for connection, and miscellaneous stressors. Posts categorized as masking emotions highlight internal struggles associated with hiding emotions to conform to societal expectations. Posts categorized as desire for connection discuss the need for emotional support and connection within relationships. Posts categorized as miscellaneous stressors address the impact of diverse personal stressors such as work, academics, and relationships. The following examples illustrate how these elements—ranging from external pressures to internal feelings of isolation and longing for support—interact.

2.1.3.1 Masking Emotions

The subtheme ‘Masking Emotions’ refers to users concealing their true emotions, often with the notion that if they showed their true feelings of loneliness, sadness, and struggle, they would burden others. Posts from two separate users exemplify masking emotions. Participant 16 observes: “Just imagine having to put on a smile face mask when others come around but behind that mask I’m probably the most saddest person you’ll ever meet.”

Participant 17, who discussed distress due to climate change, racism, and the general state of the world in an earlier example, also stated:

My parents dont understand mental health. It's so frustrating when they tell me I have no right to be depressed or anxious because I have a better life than they did. I love them but it's so hard opening up to them

The first quote, describing the act of putting on a “smile face mask,” expresses sadness hidden behind a facade of happiness, highlighting the emotional strain that accompanies concealing one’s true feelings. The second quote adds complexity, demonstrating how family dynamics can contribute to this disconnection, especially when mental health is misunderstood. Both participants demonstrate how masking emotions can lead to a sense of disconnection and exacerbate mental health issues.

2.1.3.2 Desire for Connection

In posts that highlight desire for connection, IESO users describe the human need for emotional support and connection, often in the context of feeling disconnected from their communities or naming that they are receiving inadequate support from family members, friends, or loved ones. It is important to note that these posts were all written during the COVID-19 pandemic, during a time when isolation and loneliness markedly rose across the country.

Participant 3, who previously discussed their father’s passing, also stated:

Everytime I crave love, I don’t really receive it. It’s usually given to me unexpectedly. Which is okay. But yesterday I wanted to love and affection and i might sound selfish because I understand everyone has their own issues. But I just needed that. So bad.

Participant 16, who previously discussed having to mask their emotions to interface with others, separately wrote:

Today I feel sad I’ve been going through it mentally financially… people don’t really have my back like I have theirs and it just makes me want to cry so bad

Participant 17, states:

I'm home alone and I feel very lonely, My Husband gets off late, my son is away, and I'm still looking for my Pet while being in this house all by myself. I'm missing him more when I'm home alone. God please give me the strength to get through this.

In these quotes, users emphasize the significance of mutual support when someone is experiencing a mental health event. Conversely, users also express challenges of unmet needs surrounding care and affection, and the emotional toll inadequate connection takes on individuals seeking support for mental health challenges. In addition, users also reference ideas of religion/spirituality. As one user writes, “God please give me the strength to get through this.” While naming their negative emotions, among them loneliness and sadness, and identifying what they lack in their lives, users also discuss their needs. “I wanted love and affection,” says one user. “People don't really have my back,” says another. Desire for connection posts highlight users’ loneliness and desire for supportive community when they felt isolated, disconnected, and lonely. The gap between perceiving the need for emotional support and the receipt of support is a key aspect of this theme, emphasizing the role that nurturing, reciprocal relationships play in mental wellbeing.

2.1.3.3 Miscellaneous Stressors

The subtheme everyday stressors encompasses the array of daily challenges that impact mental health. It includes various personal and situational stressors like academic pressures; workplace issues; and platonic, familial, or romantic relationships. Several IESO users also discussed financial hardships and burdens, particularly in light of COVID-19, that add to their feelings of stress and anxiety. The nature of these stressors is varied and subjective. Everyday stressors underscores the cumulative impact of stress on an individual's mental well-being and includes posts discussing stress and anxiety due to personal issues, financial stress, mental health, and relational/family dynamics.

As Participant 18 observes:

TODAY I TRIED SO HARD TO CONTROL MY ANXIETY. I HAVE SO MANY THOUGH GOING THROW MY HEAD. I HAVE SO MUCH GOING ON THAT I CAN BEARLY CATCH MY BREATH. SOMETIMES I THINK IT'S JUST TOO MUCH FOR ME AND I END UP FEELING LIKE CRAP. I'M JUST GOING TO TRY TO TAKE IT A STEP AT A TIME BECAUSE I KNOW THAT I'M A STRONG WOMAN. BUT SOMETIMES I LOSE MYSELF.

While discussing COVID-19, Participant 19 writes:

Months of disrupted schedules, canceled milestone celebrations and daily logistical challenges. Little wonder so many of us feeling lonely, frustrated and vulnerable, physically, emotionally and financially.

Participant 20 notes:

Super Stressed, work has been an absolute nightmare. I have been pushed to the brink and i'm trying not to crossover: We had a showcase for the culmination of a program. I have been running about hither and yon to prepare for the day. Prep that should've been done the day before was not down at all and the director is breathing down my neck

Participant 18 discusses a main feeling that many IESO users echoed: feeling overwhelmed by daily life, and feeling as though it is hard to function due to anxiety. Workplace stress, as described by Participant 20, points to the mental strain caused by professional demands. As Participant 19 summarized, emotional wellbeing has taken a different shape during COVID, a time when financial and emotional stressors dramatically increased. In addition to these experiences, users discussed anxiety due to relationships with families, friends, or partners. Together, these quotes reveal the wide spectrum of everyday stressors, each uniquely contributing to overall mental health challenges. Miscellaneous stressors was a large theme—with many IESO users discussing feelings of overwhelm and anxiety. In light of the pandemic, and as Participant 19 points out, with increased social isolation, IESO users reflected mental distress surrounding everyday issues—work, romantic relationships, and school. As Participant 18 writes, “I CAN BARELY CATCH MY BREATH”—this sentiment was present in multiple IESO posts—and shows a complete lack of ability to cope with everyday stressors. Vulnerable posts such as this one highlighted the fact that mental health, for many IESO users, was incredibly treacherous during this time in their lives.

2.1.4 Self-care and Joy (278 posts)

While discussing everyday stressors and interpersonal and systemic losses, IESO users often reported how they were coping and caring for themselves during times of distress. We categorized these posts as Self Care and Joy. Within the theme, four subthemes were identified: Self-care, religion and spirituality, hope, and joy. In posts assigned to self-care, users share methods for relaxation and discuss how they cope with daily stressors. Posts assigned to religion & spirituality describe the solace found in religion and spirituality during tough times, and often discuss God and/or faith and belief in God. Posts assigned to hope expressed feelings of excitement or hopefulness about the future. Posts assigned to joy include expressions of gratitude for overcoming challenges; feeling connected within relationships; happiness; and fond reflections on personal achievements. Across this theme, users describe taking an active role in supporting their own well-being and happiness, and feeling supported by family and friends.

2.1.4.1 Self Care

Users' expressions of self care emphasized the importance of experiences rooted in leisure while attending to present needs. Although language varied across posts, calmness was a desired outcome. Participant 21 writes:

Today I will watch a movie and eat popcorn and take a sip of wine. I gotta treat myself. I’m always thinking about others

This user takes time for themselves, noting that they are constantly thinking of others rather than doing things that would benefit them. Participant 16, who has discussed masking emotions and feeling mental and financial distress, writes about feeling grounded on a particular day and how they engage in self-care:

Today I feel ok. One good tip when I’m feeling down I prepare the tub and make a bubble bath with jazz slow music and just relax and it helps me a lot. Today I will be calm

This user is taking an active role in cultivating calm by engaging in self-care practices that help them relax. Other users discuss seeing friends and exercising as self-care strategies. In addition to external self-care strategies and coping mechanisms, IESO users also named the IESO platform as a helpful tool for emotional release and sharing. As one participant notes, “Feels good to have some place to talk, I feel like I could never talk to anyone of my friends or family.”

2.1.4.2 Religion & Spirituality

Religious beliefs and practices, such as worship, allow users to honor their relationship with a higher power in times of despair. Phrases like “I want to thank God” illustrate outward expressions of faith. Participant 21, who previously discussed their self-care strategies, also wrote:

Today I woke up with peace in my heart. I just want to thank god that eve through the storm he has never lifted my side.

Similarly, Participant 22 observed:

I have felt this peaceful in a very long time I just want to thank god for everything that he has giving me even though I had a hard past.

Users discuss feeling peace due to their connection with God and religion, and gratitude for God's presence in their lives. They also name the prominence of faith and God’s presence during hardship. Participant 23 bridges their feeling of helplessness with how they seek support from faith in God: “I feel like I’m loosing myself. I know god will always guide me even in my darkest time” This user speaks to currently feeling lost or unsure, but having a deep trust in God, and leaning on God for guidance.

2.1.4.3 Hope

As in posts focused on religion and spirituality, users express hope as a pathway for self-determination and positivity. One user notes that anything is possible with enough effort. Hope invites one to reimagine their future and promotes resiliency. Other posts include:

Participant 24: “Im feeling hopeful because I feel like I am finally getting back to myself after thats been occurring these past couple of years. Alot of hope was lost due to personal, financial, and emotional issues due to covid and other events. However I see an opening and many opportunities presenting themselves to me that I can take advantage of now.”

Participant 21, who discussed self-care strategies and spirituality, also offered words of hope and encouragement for other IESO users: “Good night every I hope y’all have a great evening. Just remember stay positive. Remember that the storm won’t last for ever stay strong. We can do this👐”

Similar to mentions of hardship as “the storm” in posts categorized religion and spirituality, users discuss the importance of a positive mindset in order to achieve success and endure difficult times. Hope is best outlined in Participant 21’s phrase, “We can do this,” and the post’s accompanying open-hand emoji: showcasing solidarity in the we, and a sense of community achievement.

2.1.4.4 Joy

Users embody joy in response to change, love, and everyday pleasures. For example, one user expresses relief and happiness because their brother is no longer incarcerated. Posts often discussed feeling gratitude for family members, friends, and loved ones. As Participant 22 shares: “I love my boyfriend so much sometimes I can scream.” In a similar expression of joy stemming from relief, Participant 23 observes: “Happy and Excited: I found out that my Mother doesn't have cancer!!!”

IESO users, although asked to describe feelings of distress, often discussed feeling happy, relaxed, and joyful, celebrating large and small events. One user’s post outlines their joy and gratitude for simply being alive: Participant 24 writes, “Happy as hell I’m alive and still enjoying the sun & moon everyday” Other users discuss milestones that have led to joy, such as graduating school, passing exams or tests, and feeling excited to enjoy meals with friends. Despite many posts surrounding loss, emotional distress, and pain, IESO posts expressed joy, hope, and self-care abound, highlighting users’ resilience and gratitude.

2.2 Computational Results (Topic Modeling)

To augment the qualitative analysis surfacing themes across IESO platform posts, we also consider themes surfaced through quantitative analysis. In particular, we examine topics discovered by a model fit on the IESO texts, supporting themes coded by human annotators while elevating those that may have been overlooked. We first evaluate how well the topic model fits the IESO data, and the quality of the discovered topics, and then examine those topics in relation to the aforementioned qualitative themes.

Table 1

**) Topic metrics by number of topics.** Topic Coherence, Diversity, and Quality scores for a sample of the tested values of K (number of topics). The model achieving the highest Quality score is bolded, denoting the selected value of K = 12 for all topic modeling results.
# Topics (K)	Coherence	Diversity	Quality
11	0.03	0.66	0.02
12	0.12	0.62	0.08
13	0.09	0.58	0.05
14	0.05	0.52	0.03
15	0.11	0.57	0.06

2.2.1 Topic Correlation

2.2.2 Automatically Extracted Topics

Table 2

**) Selected ProdLDA topics and most probable terms.** The top 10 most probable terms according to the fitted ProdLDA model for 5 selected topics. Topics are mapped to qualitative Potential Themes based on interpretation of the top words and are not output by the model.
Potential Theme	Topic #	Top Words
Everyday Stressors	1	Day	good	today	feeling	little	na	super	mind	try	better
	2	Day	really	right	work	hard	make	sad	things	stressed	bad
	7	Today	day	good	lot	work	make	little	home	days	better
Grief/Loss	9	Time	love	life	know	family	hard	got	miss	good	best
Positive/ Self-care	12	happy	got	person	hard	think	im	super	stress	makes	right

Table 2 shows the top terms for 5 selected topics drawn from the fitted topic model, that is, the 10 terms with the largest weight under the topic model for each. As the topics can be difficult to interpret from the top terms alone, Fig. 2 also presents the correlations between topics and emotions detected by a pre-trained model. Similarly to topic correlations, correlations are calculated over the distribution over topics and distribution over emotion labels for each document. Using this information, the presented topics in Table 2 could be interpreted as relating to 3 of the qualitative themes found in IESO posts.

Topics 1, 2 and 7 describe posts with frequent use of terms like day and today as well as stressed, feeling, bad, and work, suggesting that many may be using IESO to discuss daily emotions and stress. The dominant emotions of topics 1 and 7 (relief and pride), as well as the dominant emotion of topic 2 (sadness) further suggest that these topics may revolve around resilience to and emotional response to daily stressors respectively.

Topic 9 gives strong weight to terms like love and family, but also time, life, hard, and miss, which might suggest a theme surrounding grief and the loss of loved ones. While grief is among the considered emotion labels, we note that it is the least represented emotion in the training data, leading resulting models to perform poorly on detecting grief (Demszky et al., 2020). The topic did correlate, however, with emotion labels love and surprise, reinforcing the overall theme when considering the top terms.

Finally, topic 12 largely contains terms similar to those of topics 1, 2, and 7, such as hard and stress. The top term is happy, suggesting that the topic may describe documents focusing on self-care and positive experiences. This is further supported by the fact that topic 12 and joy are the most correlated topic and emotion pair among all topics and emotions.

Social media platforms have redefined and transformed how we communicate emotions—particularly expressions of grief and distress in daily life. Interpreting expressions of grief and everyday stressors is challenging, and further complicated when considering potential bias associated with deciphering African American English and short-form textual expressions popular across social media platforms. To address these challenges, we assembled an interdisciplinary research team composed of social work scientists, computer scientists, and linguists, to develop the IESO platform and to collect and analyze posts related to grief and everyday stressors. We used qualitative and computational methods to interpret users’ posts, validating emerging themes using topic modeling. This approach enables a clearer understanding of how Black people in New York City express experiences of grief and everyday stress on a digital platform within the broader context of societal stressors like COVID-19, systemic injustice, and police violence.

Ongoing stressors, such as school or a recent romantic break-up, bring an additional burden to the bereavement process. These stressors may reduce the ability to seek effective social support and resources, increasing the likelihood of mental health complications and physical manifestations of grief (e.g. inability to sleep). The grief literature defines adaptive grief as including “gratifying, uplifting, and/or pleasant responses as well as distressful, disheartening, and/or painful ones,” where “integrating internal and external changes catalyzed by a grief-inducing loss is essential” (Darian, 2014, p. 200). Several IESO users embodied adaptive grief, such as naming gratitude for knowing the person they lost, and working through how to memorialize and make meaning of a loss. Other less-studied and non-traditional forms of grief were prevalent amongst IESO posts, among them miscarriages; the loss of pets, opportunities, and relationships; and anticipatory ecological grief, or “the grief felt in relation to…anticipated ecological losses” and the resulting consequences (Cunsolo & Ellis, 2018, p. 275).

Many users highlighted feelings of loneliness and desire for a supportive community, exacerbated by isolation as result of the COVID-19 pandemic (Lee et al., 2020). The prevalence of disconnectedness, masking emotions, and sense of isolation throughout the posts – in conjunction all posts were written during COVID-19 – suggests that social isolation from COVID contributed to increased feelings of psychological distress (Knox et al., 2022). Conversely, some IESO users felt supported, expressing gratitude for their families and friends, and emphasized the importance of practicing self-care routines. Qualitative and quantitative findings indicated the prevalence of both distress and joy among IESO users’ posts.

IESO does not share the volatility of other social media platforms: Users know fewer of their followers personally and thus the risk for unwarranted comments from people in their social circles is lower (Ma, et al., 2016). Social media platforms like Facebook, Twitter, and Instagram, are not inherently anonymous, and there is often a significant overlap of online followers and offline social circles (Correa, et al., 2021). Users may not want to share content conveying intimate feelings of grief or distress that close family members and friends likely see (Mondal, et al., 2020). IESO provided anonymity to unmask true feelings.

IESO data and themes present mental health clinicians an opportunity to consider integrating patients’ online description of grief and other triggering emotional events into more traditional treatment methods. Our findings highlight how Black users’ experiences of grief and ongoing stress could inform adaptation of current treatments for grief and stress. Findings offer an inroad of understanding into how Black community expressions of grief appear online, providing benefit to both social work and counseling practitioners, as well as those using computational methods to understand Black communities.

3.1 Limitations

Our study has limitations in both the qualitative approaches and topic modeling. From a qualitative perspective, IESO was created to conduct research and maintain poster anonymity. As such, replicability is limited to sites and/or platforms that can replicate those same conditions. Further, our definitions of grief and distress, while informed by the corpus of grief literature, may not fully incorporate a diverse understanding of experiences of grief and stress; as such findings cannot be generalized to groups outside the study populations. As stated earlier, post anonymity may have contributed to users feeling free to more openly discuss emotions or topics they would keep confidential on non-anonymized social media platforms, such as Facebook or Twitter (Nitschinsk, et al., 2023).

Additionally, topic models are often difficult to interpret in isolation, because the surfaced topics are represented as lists of key-terms. To mitigate this limitation, we considered topics alongside manually identified themes, and analyzed documents representative of each topic to provide context. As is traditional, we omit stopwords and encode the text by the most frequent terms. These preprocessing steps, however, are intended for mainstream English and may not be suited to handling features of African American English or specific social media conventions. Future work may focus on adapting topic modeling and data preprocessing approaches to the varieties of language represented in IESO specifically and social media more broadly. Experimentation with different topic modeling techniques is limited, and other approaches may surface higher quality topics. For an initial analysis to complement identified themes, we focus on more traditional methods fitted from scratch on the IESO posts.

A particular strength of the IESO platform is its moderation by social work students, who monitored for content that may trigger other users, including harmful or violent language, to ensure a space where users felt solidarity with limited emotional risk. By doing so, IESO moderators were able to control what users encountered, and to provide resources and information for users who expressed potentially harmful sentiments, such as self-harm. The moderation (or lack thereof) of other social media platforms is particularly why further research would need to be conducted on a platform designed specifically for users to post about distress and emotions on an online forum that is 1) anonymous and 2) moderated to keep potentially triggering or harmful posts away from other users of the site.

The research team conducted qualitative and computational analysis of themes that emerged from participants postings about grief and daily stressors on the IESO platform. Participant posts were stored separately in a password-protected remote server, along with username and demographic information. All study data and confidential participant information is stored on a remote server owned by the computer science department at a large university in the northeast, with access-management permissions that afforded only members of the computer science research team access to the files. Users were encouraged to create pseudonymous logins. After collecting those logins, the research team removed any data with personally identifiable information. The research team recruited users for this study from two main sources: our Harlem community partners, via a partnership with a large Baptist church, as well as undergraduate and graduate students attending a large university in the northeast. We posted flyers advertising the study on sites including social media, email listservs, and other community spaces. In all, 125 users created 949 posts on the IESO platform during 2022.

IESO posts required approval from in-house moderators (social work graduate students) prior to publication, to identify and remove potentially triggering content, such as suicidality, self-harm, etc. Users’ content that was flagged and disapproved by moderators did not appear publicly on the IESO feed, ensuring other users could not read it. Users that posted disapproved content received a message with available resources and an explanation describing why the post was not published. The research team developed a mixed-methods approach, using both human annotation and thematic analysis as well as computational analysis. This approach was selected because it creates a richer depth of understanding amongst the data, and highlights certain themes that would not be as evident if only one methodology was chosen.

Table 3

**) Demographics of participants.** The size of each demographic group among IESO participants.
Ethnicity/Race	Count
Hispanic or Latino	49
White	8
Black or African American	42
Native Hawaiian or Pacific Islander	0
Native American or Alaska Native	1
Asian	8
More than one race/ethnicity	4
Other	0
Do not wish to disclose	13
Total	125

Along with the username and password form, the first interface (Fig. 3), asked users to enter their racial and ethnic identity if they wished to do so. This data was collected in an effort to deepen our understanding of racial differences in expressions of grief and stress. Users were free to provide any self-reported identity they wished, including selecting multiple groups or even withholding this information. As shown in Table 3, the largest group of participants self-identified as Hispanic or Latino, and a third of all participants self-identified as Black.

When posting on IESO, users accessed the form shown in Fig. 4. Participants provided a free-text description of their emotions, selected any emotion labels that applied, and described any event that may have elicited the emotional reaction in their post.

A second follow-up page (Fig. 5) asked users to rate the intensity of the emotions they chose on a 10-point Likert scale and whether their experiences and emotions had any relation to the ongoing COVID-19 pandemic, issues relevant to police brutality, or any other relevant topics.

Trained social work graduate students moderated the website, where they approved or disapproved posts to appear on the public website feed where other users could see them. Content that involved self-harm, suicidality, triggering language, or personally identifying information was disapproved, and a message was sent to the user with appropriate resources. Each post had to be approved by two moderators before it was displayed on the site to ensure that identifying information or triggering content was not displayed to other users. Posts that were not approved were annotated by graduate students, but were not publicly visible on the IESO site.

Finally, the anonymous posts (without connections to their users or usernames) and data regarding emotions and events were exported separately and provided to social work graduate students, who qualitatively analyzed and annotated each post to discover recurring themes and ideas.

We engaged in a two-step analysis process. First, we conducted qualitative analysis of annotated IESO posts. Second, to confirm or deny the qualitative themes that emerged, we compared them against a topic modeling analysis, which aimed to automatically identify substantive labels or codes in text data sets.

5.1 Qualitative Analysis

Social work graduate students conducted an inductive analysis of IESO posts via line-by-line coding. Graduate students annotated posts by systematically identifying emotions mentioned, events that were mentioned in posts, whether the event was related to the identified emotion, and most importantly, whether grief was identified in the post. The research team began the process of identifying emotions by focusing on two main topics: grief and distress. To make informed identifications, the research team consulted with a grief expert and psychiatrist to identify the parameters for interpersonal grief versus systemic loss. The grief expert advised the research team to differentiate between grief and systemic losses that may not fit the psychological criteria for grief. Prior to analysis, graduate students were trained to identify interpersonal grief if any of the following was present: yearning, longing, and preoccupying thoughts that focus on the person they lost. Systemic loss is distinct in that it stems from large-scale and highly publicized loss, including losses from police brutality or school shootings, that may not involve the user personally but may result in distress or intense emotionality. After graduate students individually coded posts, all posts were discussed within the research team and reviewed to reach consensus on labels of grief and other emotions. Annotators thematically organized posts based on the most salient emotions and concepts discussed by IESO users.

5.2 Topic Modeling

To validate qualitative themes, the research team also employed topic modeling. Probabilistic topic models are a set of computational tools that analyze the co-occurrence of words in a collection of documents to uncover topics within the corpus. After fitting such a model, each topic takes the form of a list of related terms each weighted by how strongly the term represents its topic. These topics can later be interpreted manually by the researcher to better understand the data.

Formally, topic models are latent variable mixed-membership models of a corpus of documents (Blei, 2003), where the distribution over words in each topic, the distribution over topics in each document, and the topic each individual word is assigned are all assumed to be latent variables. Mixed-membership models assume that a set of data points each belong to overlapping groups or clusters in different proportions, and these groups are considered latent or unobserved characteristics of the data. In fitting a mixed-membership model, the aim is to identify the most probable groups as well as the extent to which each data point belongs to each group. In this case, each data point is an individual word, and each group represents a topic or weighted set of terms. An example of these topics and the top-weighted terms are included in Table 2. Importantly, topic models make an exchangeability assumption, meaning that the order of terms in a document are not considered or that documents are represented by a Bag-of-Words (BOW). With this formulation, the latent variables and consequently, the topics, can be derived through posterior inference through the observed variables, the terms in each document.

In this work, we specifically employ Product of Experts Latent Dirichlet Allocation (ProdLDA; Srivastava & Sutton, 2017), a variation of Latent Dirichlet Allocation (LDA; Blei, 2003) which relaxes the assumption that each individual term is assigned to a discrete topic. We conduct posterior inference using Stochastic Variational Inference (SVI; Hoffman et al., 2013), a scalable form of approximate posterior inference that permits efficient topic modeling with large corpora of data. In applying ProdLDA to the corpus of IESO posts, we extract underlying topics to compare with the qualitative themes.

We evaluate the topics of our model with three metrics commonly used in topic modeling literature: Coherence, Diversity, and Quality. Coherence is a measure between 0 and 1 (1 being most coherent, 0 being least coherent) of how well a topic is supported by the terms in the corpus. Formally, coherence measures the Normalized Pointwise Mutual Information between each pair of words in the top-\(\:l\) terms of each topic. To compare models, we use the average coherence among all topics. Diversity is a measure between 0 and 1 (1 meaning all topics are unique, 0 meaning all topics are identical) of how different each resulting topic is among the set of topics. Formally, diversity measures the proportion of unique terms among the top-\(\:l\) terms of each topic. Finally, Quality is a single measure between 0 and 1 (1 being highest quality, 0 being lowest quality) that allows for direct comparison between two sets of topics. Quality is calculated as the product of coherence and diversity. In selecting the best topic model (namely, selecting the best number of topics, K), we fit ProdLDA for K = 1 to K = 20 and select the K value achieving the highest Topic Quality.

5.2.1 Automated Emotion Detection

To complement the automatically discovered topics, we also employ an emotion detection model in additional analysis. We use a BERT model (Devlin et al., 2018) fine-tuned on the GoEmotions dataset (Demszky et al., 2020), which, given a text, predicts a score between 0 and 1 for each of 27 emotions including basic emotions such as happiness and sadness, as well as more complex emotions such as gratitude and disapproval (see Demszky et al., 2020 for the full set of emotions and emotion model details). Using the most probable emotion output by the model for each document, we examine the correlations between topics discovered by the best ProdLDA variant and dominant emotions represented.

In conclusion, to understand how IESO users spoke about grief and distress, how their grief and feelings of distress manifested, and what events may have prompted those feelings, the combination of human annotation and topic modeling led to a comprehensive mixed-methodological approach to data analysis. Both computational and human annotation methods identified similar themes and sub-themes, indicating that computational models can bolster and affirm qualitative analysis.

Acknowledgements

The authors acknowledge the following funding sources: National Science Foundation and Columbia World Projects for their support of this work. This material is based upon work supported by the National Science Foundation under Grant Number 2106666 RI: Medium: Automatically Understanding and Identifying Digital Expression of Black Grief.

Author Contributions

These authors contributed equally: Desmond Patton, Shana Kleiner, Shug Miller, and Nick Deas

K.M., J.G., J.S.: Data curation, introduction, writing--review & editing

D.P., S.K., & S.M.: conceptualization, data curation, introduction, qualitative results, methodology, discussion, writing--review & editing

N.D., E.T., & F.E.: Data curation, quantitative results, methodology, discussion, analysis, writing-- review & editing

Competing Interests

D.P.: Advisory and consulting, unrelated to this work (Spotify, Axon, TikTok, MascotYou Humma.AI, PRIM&R).

S.K., F.E., N.D., S.M., J.G., E.T., J.S., K.M.: None

Data Availability

To protect the privacy and anonymity of participants in this work, the dataset of IESO posts cannot be shared or be made publicly available beyond the examples included in this manuscript.

Code Availability

The underlying code for fitting the topic model, the weights of the selected topic model, and the full set of topics can be accessed at https://github.com/jman23bball08/aaltopicmodeling/tree/main.

Hall, C. Bereavement theory: Recent developments in our understanding of grief and bereavement. Bereavement Care, 33(1), 7–12 (2014).
De Vries, B., & Rutherford, J. Memorializing loved ones on the World Wide Web. OMEGA-J. Death and Dying, 49(1), 5–26 (2004).
Moore, J., Magee, S., Gamreklidze, E., & Kowalewski, J. Social media mourning: Using grounded theory to explore how people grieve on social networking sites. OMEGA-J. Death and Dying, 79(3), 231–259 (2019).
Deerwester, S., Dumais, S. T., Furnas, G. W., Landauer, T. K., & Harshman, R. Indexing by latent semantic analysis. J. American Soc. Inf. Sci., 41(6), 391–407 (1990).
Blei, D. M., Ng, A. Y., & Jordan, M. I. Latent dirichlet allocation. J. Mach. Learn. Res., 993–1022 (2003).
Cohen, Jacob. "A coefficient of agreement for nominal scales". Educ. Psychol. Meas., 20(1), 37–46 (1960).
Demszky, D. et al. GoEmotions: A Dataset of Fine-Grained Emotions. Conf. Proc. 58th Annu. Meeting of the ACL, 4040–4054 (2020).
Darian, C. B. A new mourning: Synthesizing an interactive model of adaptive grieving dynamics. Illness, Crisis & Loss, 22(3), 195–235 (2014).
Cunsolo, A., & Ellis, N. R. Ecological grief as a mental health response to climate change-related loss. Nat. Clim. Change, 8(4), 275–281 (2018).
Lee, C. M., Cadigan, J. M., & Rhew, I. C. Increases in loneliness among young adults during the COVID-19 pandemic and association with increases in mental health problems. J. Adolesc. Health, 67(5), 714–717 (2020).
Knox, L., Karantzas, G. C., Romano, D., Feeney, J. A., & Simpson, J. A. One year on: What we have learned about the psychological effects of COVID-19 social restrictions: A meta-analysis. Current opinion in psychology, 46, 101315 (2022).
Ma, X., Hancock, J., & Namaan, М. Anonymity, Intimacy and Self-Disclosure in Social Media. Conf. Proc. CHI, 3857–3869 (2016).
Correa, D., Silva, L. A., Mondal, M., Benevenuto, F., & Gummadi, K. P. The many shades of anonymity: characterizing anonymous social media content. Conf. Proc. ICWSM, 9(1), 71–80 (2021).
Mondal, M., Correa, D., & Benevenuto, F. Anonymity Effects: A Large-Scale Dataset from an Anonymous Social Media Platform. Conf. Proc. HT ’20, 69–74 (2020).
Nitschinsk, L., Tobin, S. J., Varley, D., & Vanman, E. J. Why do people sometimes wear an anonymous mask? Motivations for seeking anonymity online. Pers. Soc. Psychol. Bull., 0(0) (2023).
Srivastava, A., & Sutton, C. Autoencoding Variational Inference For Topic Models. ICLR. (2017).
Hoffman, M. D., Blei, D. M., Wang, C., & Paisley, J. Stochastic variational inference. J. Mach. Learn. Res., 14, 1303–1347 (2013).
Devlin, J., Chang, M.W., Lee, K., & Toutanova, K. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. Conf. Proc. NAACL. 1, 4171–4186 (2019).

All examples are direct and verbatim quotes from the IESO platform, presented as they were written by the user.
Although the user stated Albany, we believe they are referring to the grocery store shooting which was motivated by racialized violence in Buffalo, NY, and occurred around the same time this post was written. https://apnews.com/article/buffalo-supermarket-shooting-442c6d97a073f39f99d006dbba40f64b
School shooting in Uvalde, in May 2022. https://www.texastribune.org/2023/05/24/uvalde-school-shooting-what-to-know/

No competing interests reported.

Download PDF

Reviewers agreed at journal
05 Nov, 2024
Reviewers agreed at journal
04 Nov, 2024
Reviewers agreed at journal
23 Sep, 2024
Reviewers invited by journal
21 Sep, 2024
Editor assigned by journal
03 Sep, 2024
Submission checks completed at journal
27 Aug, 2024
First submitted to journal
22 Aug, 2024

You are reading this latest preprint version

Digital Narratives of Grief and Resilience: Insights from the Integrating Emotional Stories Online ( IESO) Platform

Status:

Version 1

Abstract

Figures

1. Introduction

2. Results

2.1 Qualitative Results

2.1.1 Interpersonal Grief

2.1.1.1 Adaptive Grief

2.1.1.2 Expressions of Longing

2.1.1.3 Avoidance of Emotions

2.1.2 Institutional and Systemic Losses (48 posts)

2.1.2.1 Mass Shootings

2.1.2.2 Police Brutality

2.1.2.3 COVID-19 Pandemic

2.1.2.4 Climate Change

2.1.3 Everyday Stressors and Mental Health (462 posts)

2.1.3.1 Masking Emotions

2.1.3.2 Desire for Connection

2.1.3.3 Miscellaneous Stressors

2.1.4 Self-care and Joy (278 posts)

2.1.4.1 Self Care

2.1.4.2 Religion & Spirituality

2.1.4.3 Hope

2.1.4.4 Joy

2.2 Computational Results (Topic Modeling)

2.2.1 Topic Correlation

2.2.2 Automatically Extracted Topics

3. Discussion

3.1 Limitations

4. Methods

5. Analysis

5.1 Qualitative Analysis

5.2 Topic Modeling

5.2.1 Automated Emotion Detection

Declarations

References

Footnotes

Additional Declarations

Status:

Version 1