arXiv subjects

K. J. Zhang

Publications and source records attributed to K. J. Zhang.

5 recordsLinked to original sources

arXiv2026

FraudSMSWalker: Benchmarking Agentic Large Language Models for SMS-to-Webpage Fraud Detection

SMS fraud is increasingly cross-channel: a message directs the user to a webpage, and the final risk depends on how the SMS claim aligns with the page content and requested user action. However, existing evaluations either focus on message-only smishing classification or expose URL and domain cues that allow models to rely on reputation shortcuts. To address this gap, we introduce \textbf{FraudSMSWalker}, a controlled benchmark for URL-masked SMS-to-webpage fraud judgment. FraudSMSWalker contains 699 bilingual chains, including 332 fraudulent and 367 benign cases, across ten service scenarios. The model-visible input consists of the SMS context and sanitized webpage evidence, while raw URLs, hosts, domains, IPs, redirects, and reputation metadata are withheld. The benchmark further includes hard benign cases whose pages contain login, payment, verification, or account-management elements that are plausible under the service context but also appear in scam flows. We evaluate nine web agents under masked browser-agent protocols and conduct URL-visibility ablations. The results show that current agents can detect suspicious cues, but struggle to preserve benign recall and often produce positive predictions that are weakly supported by the observed evidence. These findings position FraudSMSWalker as a benchmark for measuring whether web agents can make fraud judgments that remain both accurate and evidence-grounded when direct reputation shortcuts are suppressed. The associated code and dataset are accessible at the \href{https://anonymous.4open.science/w/FraudMessageWalker-Bench}{anonymous link}.

cs.CL↗

arXiv2025

Revisiting the energy distribution and formation rate of CHIME fast radio bursts

Based on the first CHIME/FRB catalogue, three volume-limited samples of fast radio bursts (FRBs) are built, with samples 1, 2, and 3 corresponding to a fluence cut of 5, 3, and 1, respectively. The Lynden-Bell's c$^-$ method was applied to study their energy function and event rate evolution with redshift ($z$). Using the non-parametric Kendall's $τ$ statistics, it is found that the FRB energy ($E$) strongly evolves with redshift as $E(z)\varpropto(1+z)^{1.24}$ for sample 1, $E(z)\varpropto(1+z)^{0.98}$ for sample 2, and $E(z)\varpropto(1+z)^{1.99}$ for sample 3. After removing the redshift dependence, the local energy distributions of the three samples can be well described by a broken power-law form with a broken energy of $\sim10^{40} \rm erg$. Meanwhile, the redshift distributions of samples 1 and 2 are identical but different from that of sample 3. Interestingly, we find that the event rates of samples 1 and 2 are independent of redshift, and sample 3 decreases as a single power-law form with an index of -2.41. The local event rates of the three samples of CHIME FRBs are found to be consistently close to $\sim 10^4\rm{\,Gpc^{-3}yr^{-1}}$, which is comparable with some previous estimates. In addition, we notice that the event rate of sample 3 FRBs with lower energies not only exceeds the star formation rate at the lower redshifts but also always declines with the increase in redshift. We suggest that the excess of FRB rates compared with the star formation rate at low redshift mainly results from the low-energy FRBs that could originate in the older stellar populations.

astro-ph.HE↗

arXiv2023

New evidence of multiple channels for the origin of gamma-ray bursts with extended emission

Gamma-ray bursts (GRBs) are the most intense explosions in the universe. GRBs with extended emission (GRB EE) constitute a small subclass of GRBs. GRB EE are divided into EE-I GRBs and EE-II GRBs, according to the Amati empirical relationship rather than duration. We test here if these two types of GRB have different origins based on their luminosity function (and formation rate). Therefore, we use Lynden-Bell's c^- method to investigate the LF and FR of GRBs with EE without any assumption. We calculate the formation rate of two types of GRBs. For EE-I GRBs, the fitting function can be written as ρ(z) \propto {(1 + z)^{ - 0.34 \pm 0.04} for z < 2.39 and ρ(z) \propto {(1 + z)^{ - 2.34 \pm 0.24}} for z>2.39. The formation rate of EE-II can describe as ρ(z) \propto {(1 + z)^{ - 1.05 \pm 1.10}} for z<0.43 and ρ(z) \propto {(1 + z)^{ - 8.44 \pm 1.10}} for z>0.43. The local formation rate are ρ(0) = 0.03 Gpc^{-3}yr^{-1} for some EE-I GRBs and ρ(0) = 0.32 Gpc^{-3}yr^{-1} for EE-II GRBs. Based on these results, we provide a new evidence that the origins of EE-I GRBs are different from EE-II GRBs from the perspective of event rate. The EE-I GRB could be produced from the death of the massive star, but EE-II GRB bursts may come from other processes that are unrelated to the SFR. Our findings indicate that the GRB with EE could have multiple production channels.

astro-ph.HE↗

arXiv2023

Properties of Gamma-Ray Bursts Associated with Supernovae and Kilonovae

We systematically compare the temporal and spectral properties of 53 Supernova (SN)-associated and 15 Kilonova (KN)-associated Gamma-Ray Bursts (GRBs). We find that the spectral parameters of both types GRBs are identically and lognormally distributed, consistent with those normal GRBs. The bolometric luminosities of SN/GRBs and KN/GRBs have a triple form with the corresponding break luminosities of SN/GRBs are roughly two orders of magnitude larger than those of KN/GRBs. We build the power-law relations between the spectral lag and the luminosity of prompt $γ$-rays with indices of $-1.43\pm0.33$ for SN/GRBs and $-2.17\pm0.57$ for KN/GRBs in the laboratory frame, which are respectively coincident with the rest-frame values. We verify that both SN/GRBs and KN/GRBs comply with their own Amati relations that match those of long and short GRBs, respectively. Analyzing X-ray afterglows with good plateau segments, we build the power-law relations between the X-ray luminosity and the plateau time with an index of $-1.12\pm0.17$ for KN/GRBs and $-1.08\pm0.22$ for SN/GRBs, which can be well explained by the relativistic shock driven by an energy injection. The plots of luminosity-lag, Amati relation and luminosity-time show heavy overlap between the two types of GRBs, implying that they might share the same radiation mechanism despite originating from different progenitors or central engines.

astro-ph.HE↗

arXiv2000

Light Scattering from Nonequilibrium Concentration Fluctuations in a Polymer solution

We have performed light-scattering measurements in dilute and semidilute polymer solutions of polystyrene in toluene when subjected to stationary temperature gradients. Five solutions with concentrations below and one solution with a concentration above the overlap concentration were investigated. The experiments confirm the presence of long-range nonequilibrium concentration fluctuations which are proportional to $(\nabla T)^2/k^4$, where $\nabla T$ is the applied temperature gradient and $k$ is the wave number of the fluctuations. In addition, we demonstrate that the strength of the nonequilibrium concentration fluctuations, observed in the dilute and semidilute solution regime, agrees with theoretical values calculated from fluctuating hydrodynamics. Further theoretical and experimental work will be needed to understand nonequilibrium fluctuations in polymer solutions at higher concentrations.

cond-mat.soft↗