Hasil untuk "Municipal refuse. Solid wastes"

Rodrigo Silva, José Evans, José Isidro et al.

City council minutes are typically lengthy and formal documents with a bureaucratic writing style. Although publicly available, their structure often makes it difficult for citizens or journalists to efficiently find information. In this demo, we present CitiLink, a platform designed to transform unstructured municipal meeting minutes into structured and searchable data, demonstrating how NLP and IR can enhance the accessibility and transparency of local government. The system employs LLMs to extract metadata, discussed subjects, and voting outcomes, which are then indexed in a database to support full-text search with BM25 ranking and faceted filtering through a user-friendly interface. The developed system was built over a collection of 120 minutes made available by six Portuguese municipalities. To assess its usability, CitiLink was tested through guided sessions with municipal personnel, providing insights into how real users interact with the system. In addition, we evaluated Gemini's performance in extracting relevant information from the minutes, highlighting its effectiveness in data extraction.

Kyuri Im, Shuzhou Yuan, Michael Färber

While large language models (LLMs) have increasingly been applied to hate speech detoxification, the prompts often trigger safety alerts, causing LLMs to refuse the task. In this study, we systematically investigate false refusal behavior in hate speech detoxification and analyze the contextual and linguistic biases that trigger such refusals. We evaluate nine LLMs on both English and multilingual datasets, our results show that LLMs disproportionately refuse inputs with higher semantic toxicity and those targeting specific groups, particularly nationality, religion, and political ideology. Although multilingual datasets exhibit lower overall false refusal rates than English datasets, models still display systematic, language-dependent biases toward certain targets. Based on these findings, we propose a simple cross-translation strategy, translating English hate speech into Chinese for detoxification and back, which substantially reduces false refusals while preserving the original content, providing an effective and lightweight mitigation approach.

A. I. Lebedev, I. A. Sluchinskaya, V. N. Demin et al.

EXAFS spectroscopy is used to study the local environment of lead and selenium atoms in PbTe$_{1-x}$Se$_x$ solid solution. In addition to a bimodal distribution of bond lengths in the first shell, unusually large Debye--Waller factors for the Pb--Pb interatomic distances in the second shell and a substantial deviation of these distances from Vegard's law are observed. Valence force field (VFF) calculations show that these observations are due to the complex structure of the distribution function for Pb--Pb distances. It is found that the number of Se--Se pairs in the second shell surpasses the statistical value, which indicates that chemical factors play an important role in the structure of the solid solution. The contribution of these chemical factors to the enthalpy of mixing of the solid is estimated to be approximately 0.5 kcal/mole, which is comparable to the strain contribution.

Jincheng Liu, Oluwabunmi Iwakin, Carlos E. Romero et al.

The heterogeneity in the composition of municipal solid wastes (MSW) poses significant challenges in the production of biofuel and bioproducts. This research aims to enhance the accuracy and efficiency of waste analysis and characterization by introducing a fast characterization approach for MSW-derived refuse-derived fuels (RDF) by combining Laser-Induced Breakdown Spectroscopy (LIBS) with advanced machine learning (ML) techniques. The approach combines data pre-processing of LIBS spectra of RDF, and the development of ML models trained on domain and theory-based spectral features for predicting process parameters. These models are adept at predicting key process parameters like High Heating Value (HHV), carbon content, and volatile matter. This approach can achieve an average RRMSE of 2.13% and R2 of 0.98 or higher for all considered parameters on testing data. This work demonstrates significant potential for improving waste sorting, processing efficiency, and environmental compliance over traditional labor- and time-intensive laboratory waste analysis and characterization.

Tiago Teixeira Alves, Grasiele Soares Cavallini, Nelson Luis Gonçalves Dias Souza

Water pollution from herbicide contamination poses a significant environmental challenge, necessitating effective regenerative materials for their removal. 2,4-dichlorophenoxyacetic acid and glyphosate are among the most widely used herbicides for weed control. This study aimed to synthesize polymeric materials for the removal of these compounds from aqueous media. The study evaluated adsorption capacity, isotherms, kinetics, regeneration capacity, and the influence of pH on adsorption, alongside disinfection tests. Biodegradable polymers including chitosan, sodium alginate, and guar gum were cross-linked and characterized using infrared and Raman spectroscopy. Two samples (experiment C and M) exhibited adsorption capacities of 49.75 ± 1.474 mg g⁻¹ and 26.53 ± 1.326 mg g⁻¹ for glyphosate and 2,4-dichlorophenoxyacetic acid, respectively. Optimal adsorption was observed at pH 3.00 and 6.00 for glyphosate and 3.00 for 2,4-dichlorophenoxyacetic acid. The Langmuir and Dubinin–Radushkevich isotherms best described the adsorption behavior of glyphosate and 2,4-dichlorophenoxyacetic acid, respectively. Kinetic studies indicated that the adsorption process followed a pseudo-second-order model. Infrared and Raman absorption spectra confirmed cross-linking in the polymer samples. Regeneration tests showed that 2,4-dichlorophenoxyacetic acid adsorption remained consistent over three reuse cycles, while glyphosate adsorption increased. Disinfection tests using <i>Escherichia coli</i> and total coliforms demonstrated a significant reduction in colony-forming units, supporting the suitability of the material for this application.

Tetiana Perga

The article analyses the early stages of wastepaper collection in the Ukrainian SSR during the 1920s and early 1930s, with a focus on the key actors and their conflicting interests. The significance of makulatura is considered in the context of its economic, political and ideological importance in the early Soviet Union. Special attention is given to mass mobilisation campaigns and wastepaper collection in housing cooperatives. The desperate struggle of archival institutions to preserve their documentary heritage is highlighted. The article also reveals the role of administrative resources as a tool of directive planning, used to lobby the interests of specific companies. It demonstrates how the organisational flaws in the state wastepaper collection system contributed to the development of the black market, where wastepaper flows were redirected through unofficial channels. The article argues that speculators were the only ones to make substantial economic profits, while the state primarily derived political and ideological benefits.

Srecko Stopic, Richard Schneider, Duško Kostić et al.

This study explores both pyrometallurgical and hydrometallurgical methods for decarbonizing and recovering valuable metals from bauxite residue, with hydrogen plasma reduction and direct acid leaching as the primary approaches. The goal is to offer innovative techniques for extracting metals from bauxite residue, a by-product of the Bayer process, which cannot be disposed of in an environmentally sustainable manner. Additionally, reducing the volume of bauxite residue through combined treatments is a key objective. In contrast to traditional carbon-based reductive melting, which generated significant CO₂ emissions, hydrogen is now being investigated as a cleaner alternative. Through hydrogen plasma reduction, approximately 99.9% of iron is recovered as crude metallic iron, which can be easily separated from the slag containing other valuable metals. Thermochemical analysis was used to predict slag formation and chemical analysis of slag during hydrogen reduction. To further recover metals like aluminum and titanium, the slag is subjected to sulfuric acid leaching under high-pressure of oxygen in an autoclave avoiding silica gel formation. The results demonstrated a leaching efficiency of 93.21% for aluminum and 84.56% for titanium, using 5 mol/L sulfuric acid at 150 °C, with almost complete iron recovery. Assisted ultrasound leaching of slag with sulphuric acid under atmospheric pressure leads to 54% leaching efficiency of titanium.

Ibrahim Batcham, Djiwonou Koffi Adjalo, Koko Zébéto Houedakor et al.

The overconsumption of plastic packaging has alarming repercussions on the environment, notably through waste accumulation in public spaces and clogged drains. This study identifies factors driving plastic proliferation, analyzes their impacts, and proposes strategies for sustainable waste management. A cross-sectional design combined document review, field observations, and interviews with 156 households and 24 informants. Descriptive statistics characterized consumption patterns and service access. Impacts were assessed through litter hotspots, blocked drains, flood-prone points, and reported health risks. Households used five to six plastic bags daily, while collection coverage remained below 50%, sustaining persistent leakage. Findings reveal excessive reliance on plastics, shaped by technical, social, and institutional gaps, including weak segregation and limited pre-collection. Agoè-Nyivé 4, a fast-growing peri-urban commune within Greater Lomé, faces limited services but high consumption, making it a relevant case for rapidly growing municipalities. Yet the population often adopts counterproductive practices, hampering responsible waste management. A policy mix is outlined: expanding pre-collection and door-to-door services, integrating informal collectors, and targeted community sensitization. Without urgent interventions, plastic leakage will intensify environmental degradation, flooding, and health risks. The study recommends integrated policy measures to curb single-use dependence and foster a local circular economy.

Kun Hu, Guoning Zhou, Jia Chen et al.

To verify the possible roles of calcium peroxide (CaO₂) in addressing the key challenges of aerobic composting of food waste, including long composting duration, poor compost product quality, and gas emissions during composting, this study conducted a 38-day composting experiment using artificially blended food waste. Five containers were employed for investigating the effects of five doses of CaO₂ (0%, 5%, 10%, 15%, and 20%, <i>w</i>/<i>w</i>) on physicochemical parameters, organic matter (OM) degradation, and humification during composting. Additionally, more evidence from a microbial perspective was provided by analyzing the effects of CaO₂ additions on microbial community succession. The results indicated that CaO₂ additions increased the relative abundance of mineralization bacteria, accelerated the temperature increase of compost in the early composting stage, and elevated the peak temperature. It also facilitated the decomposition of OM and enhanced the synthesis of humic acid during the early composting stage. However, the addition of CaO₂, especially at relatively high doses, impacted the humification process. Compared with the control, only the 5% CaO₂ treatment had a significantly greater humification coefficient, reaching 1.73 ± 0.11. Moreover, adding CaO₂ reduced the total ammonia emissions from composting by 17.1% to 59.7%. Overall, CaO₂ is an effective additive for ameliorating key issues in food waste composting.

Iris Borowy

This paper examines the early development of international waste governance from 1966 to 1976. As industrialised nations generated more waste due to urbanisation and changing consumption, international organisations faced pressure to respond. The World Health Organization led efforts by creating the International Reference Centre for Wastes Disposal (IRCWD) in collaboration with the Swiss Federal Institute for Water Supply, Sewage Purification and Water Pollution Control (EAWAG) and the International Association for Public Cleansing (INTAPUC). The IRCWD aimed to centralise global waste management knowledge and coordination. However, institutional fragmentation, funding issues and differing views on whether waste was a technical or systemic issue hindered its success. While some saw waste as a symptom of flawed modernisation, most treated it as a technical problem requiring improved disposal methods. Competition from other organisations, WHO ambivalence, and reluctance to challenge economic systems weakened momentum. Ultimately, waste became a shared concern but was addressed in fragmented ways. The study shows how institutional and political factors, rather than environmental priorities, shaped early international waste policy.

Jiachen Zhao, Jing Huang, Zhengxuan Wu et al.

LLMs are trained to refuse harmful instructions, but do they truly understand harmfulness beyond just refusing? Prior work has shown that LLMs' refusal behaviors can be mediated by a one-dimensional subspace, i.e., a refusal direction. In this work, we identify a new dimension to analyze safety mechanisms in LLMs, i.e., harmfulness, which is encoded internally as a separate concept from refusal. There exists a harmfulness direction that is distinct from the refusal direction. As causal evidence, steering along the harmfulness direction can lead LLMs to interpret harmless instructions as harmful, but steering along the refusal direction tends to elicit refusal responses directly without reversing the model's judgment on harmfulness. Furthermore, using our identified harmfulness concept, we find that certain jailbreak methods work by reducing the refusal signals without reversing the model's internal belief of harmfulness. We also find that adversarially finetuning models to accept harmful instructions has minimal impact on the model's internal belief of harmfulness. These insights lead to a practical safety application: The model's latent harmfulness representation can serve as an intrinsic safeguard (Latent Guard) for detecting unsafe inputs and reducing over-refusals that is robust to finetuning attacks. For instance, our Latent Guard achieves performance comparable to or better than Llama Guard 3 8B, a dedicated finetuned safeguard model, across different jailbreak methods. Our findings suggest that LLMs' internal understanding of harmfulness is more robust than their refusal decision to diverse input instructions, offering a new perspective to study AI safety.

Keiho Matsumoto

We construct a covariant realization functor, denoted \textsc{Solidm}, from the category of motives with modulus to the derived category of solid modules in the sense of Clausen--Scholze. For any smooth modulus pair (X, D), the dual of Solidm(X, D) recovers the Hodge realization of Kelly--Miyazaki for (X, D). Using Ren's pro-solid comparison theorem, we give an explicit description of Solidm(X, D) and compute Solidm of the cone of M(U, D restricted to U) $\to$ M(X, D), in the setting where X is a smooth proper variety over a field, D $\subset$ X is a simple normal crossings divisor, and U $\subset$ X is an open immersion. We identify the result via the formal completion of X along the complement X $\setminus$ U.

Adel Khorramrouz, Sharon Levy

Safety guardrails in large language models(LLMs) are developed to prevent malicious users from generating toxic content at a large scale. However, these measures can inadvertently introduce or reflect new biases, as LLMs may refuse to generate harmful content targeting some demographic groups and not others. We explore this selective refusal bias in LLM guardrails through the lens of refusal rates of targeted individual and intersectional demographic groups, types of LLM responses, and length of generated refusals. Our results show evidence of selective refusal bias across gender, sexual orientation, nationality, and religion attributes. This leads us to investigate additional safety implications via an indirect attack, where we target previously refused groups. Our findings emphasize the need for more equitable and robust performance in safety guardrails across demographic groups.

Md Abdullah Al Mamun, Ihsen Alouani, Nael Abu-Ghazaleh

Large Language Models (LLMs) are aligned to meet ethical standards and safety requirements by training them to refuse answering harmful or unsafe prompts. In this paper, we demonstrate how adversaries can exploit LLMs' alignment to implant bias, or enforce targeted censorship without degrading the model's responsiveness to unrelated topics. Specifically, we propose Subversive Alignment Injection (SAI), a poisoning attack that leverages the alignment mechanism to trigger refusal on specific topics or queries predefined by the adversary. Although it is perhaps not surprising that refusal can be induced through overalignment, we demonstrate how this refusal can be exploited to inject bias into the model. Surprisingly, SAI evades state-of-the-art poisoning defenses including LLM state forensics, as well as robust aggregation techniques that are designed to detect poisoning in FL settings. We demonstrate the practical dangers of this attack by illustrating its end-to-end impacts on LLM-powered application pipelines. For chat based applications such as ChatDoctor, with 1% data poisoning, the system refuses to answer healthcare questions to targeted racial category leading to high bias ($ΔDP$ of 23%). We also show that bias can be induced in other NLP tasks: for a resume selection pipeline aligned to refuse to summarize CVs from a selected university, high bias in selection ($ΔDP$ of 27%) results. Even higher bias ($ΔDP$~38%) results on 9 other chat based downstream applications.

Nicole L. Shaw, Arvin Farid, Zahra Taheri Sarteshnizi

The process of purifying agricultural products, at various food processing plants, generates waste materials that consist of precipitated calcium carbonate, organic debris, and trace amounts of soil and agricultural contaminants. A specific food-processing waste, hereafter referred to as a food industry byproduct, FIBP, is typically stockpiled on land adjacent to the corresponding food processing facilities due to its large volume and chemical composition. The FIBP also contains commercially available unspent lime products, which makes its reuse viable in various applications. An example is construction applications where an organic content of up to 5% by weight is allowed, such as treating expansive clays. Traditionally, lime stabilization has been used for improving the properties of expansive clays, where ground improvement methods are necessary for a large area. However, the process of producing lime is resource- and energy-intensive as it includes crushing and heating limestone in kilns to extract lime. Therefore, one specific doubly sustainable application is the treatment of expansive clays using the FIBP instead of lime. The main application tested here is the treatment of expansive clayey soils underneath a stretch of State Highway 95 near Marsing, ID. Other potential applications are in road and embankment construction. To evaluate the potential of expansive clay stabilization utilizing the FIBP, a series of geotechnical and environmental laboratory testing were conducted to measure the engineering properties (e.g., swell potential, permeability, and strength properties) of expansive clay amended with FIBP. Preliminary testing on blends with an expansive clay suggests benefits such as decreased swelling potential, increased density, and leachate immobilization.

Dong-Hee Kang, James G. Hunter, Anastasia Chirnside

Recycled Concrete Aggregate (RCA) of the chemical and biological effects must be understood to avoid potential adverse impacts to the bay’s aquatic ecosystem. RCA application as a base material for oyster reefs did not adversely affect oyster spat growth and survival, or the surrounding environment. Evaluated RCA leaching for petroleum byproducts showed that RCA as a base material for oyster reefs did not leach any hydrocarbon chemicals, and no water extractable SVOC were detected. The research found potential RCA application to the Chesapeake Bay watershed as a bottom conditioning material for oyster aquaculture. Overall, the findings support the use of RCA for oyster aquaculture.

Ioannis Makrygiannis, Vassilis Athanasiadis, Theodoros Chatzimitakos et al.

Stone fruits, such as the apricot (<i>Prunus armeniaca</i> L.), are frequently consumed. As such, a substantial volume of apricot waste is generated at each stage of the food supply chain, including harvesting, processing, packaging, warehousing, transportation, retailing, and eventual consumption. This generates tons of waste annually on a global scale. The significant amounts of phenolics present in these wastes are primarily responsible for their antioxidant capacity and the subsequent health advantages they provide. As such, apricot pulp by-products could be a valuable reservoir of bioactive compounds, such as tocopherols, polyphenolic compounds, proteins, dietary fibers, etc. Moreover, apricot kernels are also recognized for their abundance of bioactive compounds, including polyphenols and tocopherols, which find utility in diverse sectors including cosmetology and the food industry. Both conventional and green methods are employed, and generally, green methods lead to higher extraction efficiency. The antimicrobial properties of apricot kernel essential oil have been widely recognized, leading to its extensive historical usage in the treatment of diverse ailments. In addition, apricot kernel oil possesses the capacity to serve as a viable resource for renewable fuels and chemicals. This review examines the potential of apricot waste as a source of bioactive compounds, as well as its utilization in diverse applications, with an emphasis on its contribution to health improvement.

Xinpeng Wang, Chengzhi Hu, Paul Röttger et al.

Training a language model to be both helpful and harmless requires careful calibration of refusal behaviours: Models should refuse to follow malicious instructions or give harmful advice (e.g."how do I kill someone?"), but they should not refuse safe requests, even if they superficially resemble unsafe ones (e.g. "how do I kill a Python process?"). Avoiding such false refusal, as prior work has shown, is challenging even for highly-capable language models. In this paper, we propose a simple and surgical method for mitigating false refusal in language models via single vector ablation. For a given model, we extract a false refusal vector and show that ablating this vector reduces false refusal rate while preserving the model's safety and general capabilities. We also show that our approach can be used for fine-grained calibration of model safety. Our approach is training-free and model-agnostic, making it useful for mitigating the problem of false refusal in current and future language models.

Yasunori Okumura

This paper investigates the strategic implications of the uniform rank-minimizing mechanism (URM), an assignment rule that selects uniformly from the set of deterministic assignments minimizing the sum of agents' reported ranks. We focus on settings in which agents may refuse their assignment and instead receive an outside option. Without the refusal option, we show that truth-telling is not strictly dominated under any fair rank-minimizing mechanism; that is, one satisfying equal treatment of equals. However, introducing the refusal option significantly changes strategic incentives: specific manipulations, called outside option demotion strategies, dominate truth-telling under the URM. Moreover, such manipulations can lead to inefficient outcomes, as desirable objects may be refused by misreporting agents and consequently remain unassigned. To address this issue, we propose a modification of the URM that restores undominated truth-telling, although it introduces incentives to underreport acceptable objects. Our results highlight a fundamental trade-off in the design of fair rank-minimizing mechanisms when agents can refuse their assignments.

Priyanshu Kumar, Elaine Lau, Saranya Vijayakumar et al.

For safety reasons, large language models (LLMs) are trained to refuse harmful user instructions, such as assisting dangerous activities. We study an open question in this work: does the desired safety refusal, typically enforced in chat contexts, generalize to non-chat and agentic use cases? Unlike chatbots, LLM agents equipped with general-purpose tools, such as web browsers and mobile devices, can directly influence the real world, making it even more crucial to refuse harmful instructions. In this work, we primarily focus on red-teaming browser agents, LLMs that manipulate information via web browsers. To this end, we introduce Browser Agent Red teaming Toolkit (BrowserART), a comprehensive test suite designed specifically for red-teaming browser agents. BrowserART is consist of 100 diverse browser-related harmful behaviors (including original behaviors and ones sourced from HarmBench [Mazeika et al., 2024] and AirBench 2024 [Zeng et al., 2024b]) across both synthetic and real websites. Our empirical study on state-of-the-art browser agents reveals that, while the backbone LLM refuses harmful instructions as a chatbot, the corresponding agent does not. Moreover, attack methods designed to jailbreak refusal-trained LLMs in the chat settings transfer effectively to browser agents. With human rewrites, GPT-4o and o1-preview-based browser agents attempted 98 and 63 harmful behaviors (out of 100), respectively. We publicly release BrowserART and call on LLM developers, policymakers, and agent developers to collaborate on improving agent safety