pith. sign in

arxiv: 2509.12288 · v2 · pith:YGQOLDI4new · submitted 2025-09-15 · 💻 cs.SI · cs.AI· cs.CY· cs.IR

Digital Voices of Survival: From Social Media Disclosures to Support Provisions for Domestic Violence Victims

Kanlun Wang , Zhe Fu , Wangjiaxuan Xin , Lina Zhou , Shashi Kiran Chandrappa This is my paper

Pith reviewed 2026-05-21 22:14 UTC · model grok-4.3

classification 💻 cs.SI cs.AIcs.CYcs.IR
keywords domestic violencesocial mediaself-disclosureonline supportcomputational frameworktopic summarizationsupport mapping
0
0 comments X

The pith

A four-component framework detects domestic violence disclosures on social media and maps them to community support provisions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes a computational framework to analyze social media posts where domestic violence victims disclose experiences and receive responses from online communities. It breaks the analysis into four steps: spotting disclosures, grouping similar posts, summarizing the main topics, and pulling out the specific support being offered or requested. A sympathetic reader would care because domestic violence remains a widespread problem and clearer patterns from real posts could guide more effective ways to connect victims with help. The work tests the framework on data from relevant online communities to show how self-disclosure connects to actual support.

Core claim

This study proposes a novel computational framework for modeling DV support-seeking behavior alongside community support mechanisms. The framework consists of four key components: self-disclosure detection, post clustering, topic summarization, and support extraction and mapping. When implemented and evaluated with data collected from relevant social media communities, the approach advances existing knowledge on DV self-disclosure and online support provisions and enables victim-centered digital interventions.

What carries the argument

The four-component computational framework that performs self-disclosure detection, post clustering, topic summarization, and support extraction and mapping to connect victim posts with community responses.

Load-bearing premise

Data collected from relevant social media communities is representative and sufficient to implement and evaluate the four-component framework.

What would settle it

Testing the framework on a fresh, independent set of social media posts and finding that it fails to accurately detect disclosures or correctly map support offers would falsify the central claim.

Figures

Figures reproduced from arXiv: 2509.12288 by Kanlun Wang, Lina Zhou, Shashi Kiran Chandrappa, Wangjiaxuan Xin, Zhe Fu.

Figure 1
Figure 1. Figure 1: A computational framework for modeling DV support [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 4
Figure 4. Figure 4: Particularly, support 1 (prioritiz [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 4
Figure 4. Figure 4: DV self-disclosure topics and support provisions Other forms of support are more specific to certain types of abuse or emotional challenges. For example, Support 6 (reclaiming independence) helped in posts about custody battles, long-term effects, and piecing together abuse histories. Support 7 (trusting your instincts) is associated with clusters related to disbelief, gaslighting, and internal doubt, high… view at source ↗
read the original abstract

Domestic Violence (DV) is a pervasive public health problem characterized by patterns of coercive and abusive behavior within intimate relationships. With the rise of social media as a key outlet for DV victims to disclose their experiences, online self-disclosure has emerged as a critical yet underexplored avenue for support-seeking. In addition, existing research lacks a comprehensive and nuanced understanding of DV self-disclosure, support provisions, and their connections. To address these gaps, this study proposes a novel computational framework for modeling DV support-seeking behavior alongside community support mechanisms. The framework consists of four key components: self-disclosure detection, post clustering, topic summarization, and support extraction and mapping. We implement and evaluate the framework with data collected from relevant social media communities. Our findings not only advance existing knowledge on DV self-disclosure and online support provisions but also enable victim-centered digital interventions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a novel computational framework consisting of four components—self-disclosure detection, post clustering, topic summarization, and support extraction and mapping—to model domestic violence (DV) support-seeking behavior and community support mechanisms on social media. It states that the framework was implemented and evaluated using data collected from relevant social media communities, with findings claimed to advance knowledge on DV self-disclosure and online support provisions while enabling victim-centered digital interventions.

Significance. If the implementation and evaluation prove robust with representative data, this applied framework could provide a structured computational approach to analyzing sensitive online disclosures, contributing to computational social science by linking self-disclosure patterns with support mechanisms and informing digital interventions for a major public health issue.

major comments (2)
  1. [§4] §4 (Data Collection): The manuscript states that the framework was implemented and evaluated with data from relevant social media communities but provides no details on collection protocol, platform or subreddit selection criteria, total post volume, temporal scope, or any validation against selection bias or demographic skew. This directly undermines the central assumption that the data is representative and sufficient for a comprehensive understanding.
  2. [§5] §5 (Evaluation): The abstract and framework description claim implementation and evaluation of the four components, yet no specific algorithms, performance metrics (e.g., precision or recall for detection tasks), validation procedures, or error analysis are reported. This leaves the empirical support for the framework's utility without visible grounding.
minor comments (1)
  1. [Abstract and §1] The abstract and introduction could more explicitly distinguish the proposed framework from prior work on social media analysis of sensitive topics to strengthen the novelty claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We agree that the current manuscript lacks sufficient transparency on data collection and evaluation procedures, which are essential for assessing the framework's robustness. We will revise the manuscript to address these points directly and provide the requested details.

read point-by-point responses

  1. Referee: [§4] §4 (Data Collection): The manuscript states that the framework was implemented and evaluated with data from relevant social media communities but provides no details on collection protocol, platform or subreddit selection criteria, total post volume, temporal scope, or any validation against selection bias or demographic skew. This directly undermines the central assumption that the data is representative and sufficient for a comprehensive understanding.

    Authors: We agree that these details are missing and weaken the claims of representativeness. In the revised manuscript, we will add a new subsection under Data Collection that specifies: the platform (Reddit), the exact subreddits chosen and the rationale for their selection based on community focus on DV support, the temporal scope of data collection, the total volume of posts retrieved, the keyword-based filtering and scraping protocol used, and steps taken to assess and mitigate selection bias (e.g., cross-validation against known DV-related keywords and manual review of a sample). We will also explicitly discuss limitations regarding demographic skew and generalizability. revision: yes

  2. Referee: [§5] §5 (Evaluation): The abstract and framework description claim implementation and evaluation of the four components, yet no specific algorithms, performance metrics (e.g., precision or recall for detection tasks), validation procedures, or error analysis are reported. This leaves the empirical support for the framework's utility without visible grounding.

    Authors: We acknowledge this gap in the reporting of empirical results. The revised manuscript will expand the Evaluation section to describe the specific algorithms and models used for each of the four components (e.g., the supervised classifier or LLM prompt for self-disclosure detection, the clustering algorithm, the summarization method, and the support extraction technique). We will report quantitative performance metrics including precision, recall, and F1 scores obtained via cross-validation or held-out annotated test sets, describe the annotation and validation procedures, and include an error analysis highlighting common failure modes. These additions will provide concrete grounding for the framework's utility. revision: yes

Circularity Check

0 steps flagged

No circularity: applied NLP framework on external data

full rationale

The paper describes a four-component computational framework (self-disclosure detection, post clustering, topic summarization, support extraction and mapping) that is implemented and evaluated directly on data collected from relevant social media communities. No equations, fitted parameters, predictions derived from subsets of the same data, or load-bearing self-citations appear in the derivation chain. The central claims rest on standard NLP techniques applied to external observations rather than any self-referential reduction or renaming of inputs as outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that social media data from relevant communities can meaningfully model real support-seeking behavior; no free parameters or invented entities are mentioned in the abstract.

axioms (1)
  • domain assumption Social media posts from relevant communities can be used to model DV support-seeking behavior and community support mechanisms.
    This premise underpins the entire framework and its evaluation on collected data.

pith-pipeline@v0.9.0 · 5698 in / 1267 out tokens · 48380 ms · 2026-05-21T22:14:42.902448+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages · 1 internal anchor

  1. [1]

    & Großberger, L

    https://doi.org/10.21105/joss.00861 Murvartian, L., Matías-García, J. A., Saavedra-Macías, F. J., & Crowe, A. (2024). A Systematic Review of Public Stigmatization Toward Women Victims of Intimate Partner Violence in Low - and Middle -Income Countries. Trauma, Violence, & Abuse , 25(2), 1349–

    work page doi:10.21105/joss.00861 2024

  2. [2]

    GPT-4o System Card

    https://doi.org/10.1177/15248380231178756 OpenAI, Hurst, A., Lerer, A., Goucher, A. P., Perelman, A., Ramesh, A., Clark, A., Ostrow, A. J., Welihinda, A., Hayes, A., Radford, A., Mądry, A., Baker -Whitcomb, A., Beutel, A., Borzunov, A., Carney, A., Chow, A., Kirillov, A., Nichol, A., … Malkov, Y. (2024). GPT-4o System Card (No. arXiv:2410.21276). arXiv. h...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1177/15248380231178756 2024