Poli Nemkova

Poli Nemkova, Solomon Ubani, Suleyman Olcay Polat et al.

The present-day Russia-Ukraine military conflict has exposed the pivotal role of social media in enabling the transparent and unbridled sharing of information directly from the frontlines. In conflict zones where freedom of expression is constrained and information warfare is pervasive, social media has emerged as an indispensable lifeline. Anonymous social media platforms, as publicly available sources for disseminating war-related information, have the potential to serve as effective instruments for monitoring and documenting Human Rights Violations (HRV). Our research focuses on the analysis of data from Telegram, the leading social media platform for reading independent news in post-Soviet regions. We gathered a dataset of posts sampled from 95 public Telegram channels that cover politics and war news, which we have utilized to identify potential occurrences of HRV. Employing a mBERT-based text classifier, we have conducted an analysis to detect any mentions of HRV in the Telegram data. Our final approach yielded an $F_2$ score of 0.71 for HRV detection, representing an improvement of 0.38 over the multilingual BERT base model. We release two datasets that contains Telegram posts: (1) large corpus with over 2.3 millions posts and (2) annotated at the sentence-level dataset to indicate HRVs. The Telegram posts are in the context of the Russia-Ukraine war. We posit that our findings hold significant implications for NGOs, governments, and researchers by providing a means to detect and document possible human rights violations.

Poli Nemkova, Haeshitha Indukuri, Jaedon Charles

Agentic AI systems are increasingly proposed for social-good domains, often invoking the United Nations Sustainable Development Goals (SDGs) as a vocabulary of global benefit. Yet claims of social good do not establish accountability to the communities a system claims to serve. We present a structured survey of 112 papers on agentic AI for social good published between 2015 and 2026. We find a moral-geographic asymmetry: papers are least likely to specify geographic context in precisely the domains where local political, legal, and cultural context matters most. Across the corpus, 82 of 112 papers (73%) specify no geographic context. Papers aligned with health or physical/ecological SDGs specify geography 37-40% of the time, while papers aligned with institutional and social-policy SDGs do so only 13%. SDG 16, peace, justice, and strong institutions, is both the most-covered goal in the corpus and the one with the lowest geographic-specification rate. We interpret this as moral abstraction: agentic AI for social good often treats institutional good as universal in ways it does not treat health or ecological good. A second finding compounds this: only 28 of 112 papers (25%) report any real-world deployment or small-scale test. We identify five accountability gaps and propose a minimal reporting standard for more context-specific, participatory, and accountable agentic AI for social good.

George Mihaila, Suleyman Olcay Polat, Poli Nemkova et al.

Local explanation methods such as LIME (Ribeiro et al., 2016) remain fundamental to trustworthy AI, yet their application to NLP is limited by a reliance on random token masking. These heuristic perturbations frequently generate semantically invalid, out-of-distribution inputs that weaken the fidelity of local surrogate models. While recent generative approaches such as LLiMe (Angiulli et al., 2025b) attempt to mitigate this by employing Large Language Models for neighborhood generation, they rely on unconstrained paraphrasing that introduces confounding variables, making it difficult to isolate specific feature contributions. We introduce LIME-LLM, a framework that replaces random noise with hypothesis-driven, controlled perturbations. By enforcing a strict "Single Mask-Single Sample" protocol and employing distinct neutral infill and boundary infill strategies, LIME-LLM constructs fluent, on-manifold neighborhoods that rigorously isolate feature effects. We evaluate our method against established baselines (LIME, SHAP, Integrated Gradients) and the generative LLiMe baseline across three diverse benchmarks: CoLA, SST-2, and HateXplain using human-annotated rationales as ground truth. Empirical results demonstrate that LIME-LLM establishes a new benchmark for black-box NLP explainability, achieving significant improvements in local explanation fidelity compared to both traditional perturbation-based methods and recent generative alternatives.

Antonia Karamolegkou, Angana Borah, Eunjung Cho et al.

Recent advancements in large language models (LLMs) have unlocked unprecedented possibilities across a range of applications. However, as a community, we believe that the field of Natural Language Processing (NLP) has a growing need to approach deployment with greater intentionality and responsibility. In alignment with the broader vision of AI for Social Good (Tomašev et al., 2020), this paper examines the role of NLP in addressing pressing societal challenges. Through a cross-disciplinary analysis of social goals and emerging risks, we highlight promising research directions and outline challenges that must be addressed to ensure responsible and equitable progress in NLP4SG research.

Poli Nemkova, Amrit Adhikari, Matthew Pearson et al.

Humanitarian organizations face a critical choice: invest in costly commercial APIs or rely on free open-weight models for multilingual human rights monitoring. While commercial systems offer reliability, open-weight alternatives lack empirical validation -- especially for low-resource languages common in conflict zones. This paper presents the first systematic comparison of commercial and open-weight large language models (LLMs) for human-rights-violation detection across seven languages, quantifying the cost-reliability trade-off facing resource-constrained organizations. Across 78,000 multilingual inferences, we evaluate six models -- four instruction-aligned (Claude-Sonnet-4, DeepSeek-V3, Gemini-Flash-2.0, GPT-4.1-mini) and two open-weight (LLaMA-3-8B, Mistral-7B) -- using both standard classification metrics and new measures of cross-lingual reliability: Calibration Deviation (CD), Decision Bias (B), Language Robustness Score (LRS), and Language Stability Score (LSS). Results show that alignment, not scale, determines stability: aligned models maintain near-invariant accuracy and balanced calibration across typologically distant and low-resource languages (e.g., Lingala, Burmese), while open-weight models exhibit significant prompt-language sensitivity and calibration drift. These findings demonstrate that multilingual alignment enables language-agnostic reasoning and provide practical guidance for humanitarian organizations balancing budget constraints with reliability in multilingual deployment.