Evaluating recommender systems remains challenging due to the gap between offline metrics and real user behavior, as well as the scarcity of interaction data.Recent work explores large language model (LLM) agents as synthetic users, yet they typically rely on few-shot prompting, which yields a shallow understanding of the environment and limits their ability to faithfully reproduce user actions.We introduce ALIGNUSER, a framework that learns world-model-driven agents from human interactions.Given rollout sequences of actions and states, we formalize world modeling as a next state prediction task that helps the agent internalize the environment.To align actions with human personas, we generate counterfactual trajectories around demonstrations and prompt the LLM to compare its decisions with human choices, identify suboptimal actions, and extract lessons.The learned policy is then used to drive agent interactions with the recommender system.We evaluate ALIGNUSER across multiple datasets and demonstrate closer alignment with genuine humans than prior work, both at the micro and macro levels.

Urban nitrogen dioxide nitric oxide (NO2) emission is a major source of total NO2 emission, yet brings large uncertainty of emission estimates because of its complicated internal environment. Current top-down methods, for example, exponentially modified Gaussian (EMG) have strict theoretical assumptions of atmospheric dispersion condition and simplify the whole urban region as an isolated point source, which differs from actual emission situation, resulting in poor temporal representativeness and large uncertainty. This article constructs a remote sensing estimation method for urban NO2 emissions based on the cross-sectional flux method which is insensitive to meteorological assumptions, and improved it with considerations of NO2 lifetime. Compared with ground-based observation, the mean absolute percentage error (MAPE) of this work decreases by 42.58% compared with EMG’s MAPE. On the total scale of annual stocktake, the MAPE of this work is reduced by 39.79% compared with the EMG method. This method weakens the impact of meteorology condition and provides higher temporal representativeness. The retrieved NO2 emission based on this method of New York City, Las Vegas, Chicago, Wuhan, Xi’an, and Paris shows a$61.55~\pm ~28.25$kt/yr differences compared with Emissions Database for Global Atmospheric Research (EDGAR) inventory results, with overestimation up to 135.39 kt/yr (Xi’an). For all study regions, a clearly temporal pattern of NO2 emission is found, with the monthly emission during ozone season increases 5.76 kt/month compared with nonozone season. The cross-sectional flux method, once improved, demonstrates greater accuracy in estimating urban NO2 emissions and is expecting to be applied to different gaseous emission to provide more reliable results.

Fusing multiple satellite aerosol optical depth (AOD) products to produce high-quality aerosol records is necessary for climate-related research. The Bayesian maximum entropy (BME) approach has unique advantages in this regard. However, the large accuracy differences and redundancy characteristics between aerosol products are not considered by BME resulting in limited fusion quality and large computational consumption. Therefore, in this study, we try to explicitly introduce a priori knowledge about the multisource AOD product accuracy, called the accuracy ranking matrix, into the BME fusion process to test whether the above issues can be alleviated. Eight publicly released aerosol products over China are used in the fusion experiments, and the fusion results are validated with ground-based measurements. Results suggest that, compared with fusion directly, the fusion accuracy and efficiency are significantly improved by introducing the accuracy ranking matrix; the proportion of satisfying the Global Climate Observing System (GCOS) accuracy requirements increases by 12.2%, and the consumption time decreases by 74.6%. Specifically, the accuracy improvement in vegetated and clean air conditions outperforms that in arid and polluted conditions, with the most significant improvement in clean conditions, where the GCOS fraction increased by 22.9%. Moreover, the spatial distribution of fusion results in typical regions indicates that the introduction of the accuracy ranking matrix makes the AOD spatial variation smoother. These results demonstrate that introducing the accuracy ranking matrix can generate higher quality AOD distributions with improved efficiency, which is expected to guide the fusion of AOD products or other remote sensing products.

Motivation: The metabolites of exogenous and endogenous compounds play a pivotal role in the domain of metabolism research. However, they are still unclear for most chemicals in our environment. The in silico methods for predicting the site of metabolism (SOM) are considered to be efficient and low-cost in SOM discovery. However, many in silico methods are focused on metabolism processes catalyzed by several specified Cytochromes P450s, and only apply to substrates with special skeleton. A SOM prediction model always deserves more attention, which demands no special requirements to structures of substrates and applies to more metabolic enzymes. Results: By incorporating the use of hybrid feature selection techniques (CHI, IG, GR, Relief) and multiple classification procedures (KStar, BN, IBK, J48, RF, SVM, AdaBoostM1, Bagging), SOM prediction models for six oxidation reactions mediated by oxidoreductases were established by the integration of enzyme data and chemical bond information. The advantage of the method is the introduction of unlabeled SOM. We defined the SOM which not reported in the literature as unlabeled SOM, where negative SOM was filtered. Consequently, for each type of reaction, a series of SOM prediction models were built based on information about metabolism of 1237 heterogeneous chemicals. Then optimal models were attained through comparisons among these models. Finally, independent test set was used to validate optimal models. It demonstrated that all models gave accuracies above 0.90. For receiver operating characteristic analysis, the area under curve values of all these models over 0.906. The results suggested that these models showed good predicting power. Availability and implementation: All the models will be available when contact with [email protected]. Contacts: [email protected] or [email protected]. Supplementary information: Supplementary data are available at Bioinformatics online.