Moisture content (MC) is one of the most important quality parameters of green coffee beans. Therefore, its fast and reliable measurement is necessary. This study evaluated the feasibility of near infrared (NIR) spectroscopy and chemometrics for rapid and non-destructive prediction of MC in intact green coffee beans of both Coffea arabica (Arabica) and Coffea canephora (Robusta) species. Diffuse reflectance (log 1/R) spectra of intact beans were acquired using a bench top Fourier transform NIR instrument. MC was determined gravimetrically according to The International Organization for Standardization (ISO) 6673. Samples were split into subsets for calibration (n = 64) and independent validation (n = 44). A three-component partial least squares regression (PLSR) model using raw NIR spectra yielded a root mean square error of prediction (RMSEP) of 0.80% MC; a four component PLSR model using scatter corrected spectra yielded a RMSEP of 0.57% MC. A simplified PLS model using seven selected wavelengths (1155, 1212, 1340, 1409, 1724, 1908, and 2249 nm) yielded a similar accuracy (RMSEP: 0.77% MC) which opens the possibility of creating cheaper NIR instruments. In conclusion, NIR diffuse reflectance spectroscopy appears to be suitable for rapid and reliable MC prediction in intact green coffee; no separate model for Arabica and Robusta species is needed.

Species adulteration is a common problem in the coffee trade. Several attempts have been made to differentiate among species. However, finding an applicable methodology that would consider the various aspects of adulteration remains a challenge. This study investigated an ultraviolet–visible (UV-Vis) spectroscopy-based determination of caffeine and chlorogenic acid contents, as well as the applicability of non-targeted near-infrared (NIR) spectroscopy, to discriminate between green coffee beans of the Coffea arabica (Arabica) and Coffea canephora (Robusta) species from Java Island, Indonesia. The discrimination was conducted by measuring the caffeine and chlorogenic acid content in the beans using UV-Vis spectroscopy. The data related to both compounds was processed using linear discriminant analysis (LDA). Information about the diffuse reflectance (log 1/R) spectra of intact beans was determined by NIR spectroscopy and analyzed using multivariate analysis. UV-Vis spectroscopy attained an accuracy of 97% in comparison to NIR spectroscopy’s accuracy by selected wavelengths of LDA (95%). The study suggests that both methods are applicable to discriminate reliably among species.

The butterfly pea flower (Clitoria ternatea L.) is a flower that can be grown as a medicinal plant because it contains anthocyanins, which have antioxidant activity. Anthocyanins are compounds that play a role in giving red, purple, and blue colors to petals and fruits. The purpose of this study was to determine the effect of three methods of spray, vacuum, and freeze drying on the physicochemical properties of powdered butterfly pea extract and determine the appropriate type of drying to produce powdered butterfly pea extract as a natural dye. The method used in this study was a randomized block design (RBD) with three treatment levels and four repetitions. The treatments used included drying using a vacuum, spraying, and freezing. The analysis carried out was a physicochemical analysis, which included water content, antioxidant activity, total anthocyanin, dissolution time, solubility, color intensity, and hygroscopicity level. The results showed that the vacuum-, spray-, and freeze-drying methods had a significant effect on water content, antioxidant activity, dissolving time, solubility, color, and hygroscopicity but did not have a significant effect on total anthocyanin. Freeze drying is the best treatment on the parameters of water content, antioxidant activity, and color intensity. Meanwhile, for the solubility level and dissolution time parameters, the best results were in the spray-drying treatment. Vacuum drying is the best treatment on the parameters of total anthocyanin and hygroscopicity.

Sorghum has become increasingly popular for diverse applications, including food, feed, and bioenergy. The present study aimed to perform a bibliometric and content analysis on sorghum research in the industrial-agricultural sector to evaluate global research patterns, emerging topics, and future trends. The bibliometric analysis of the Scopus bibliographic metadata was done using Bibliometrix and Vosviewer. The results of this investigation show the descriptive analysis underscored fluctuation in research output and citation impact over the years. The identified four patterns through co-occurrence analysis encompassed central themes of sorghum research, including genetic expression and plant stress cluster, cereal crop and yield cluster, biomass and microbial bioprocessing, and metabolism and antioxidant cluster. The thematic map portrays a comprehensive view of cereal crops and yields, biomass and microbial bioprocessing, metabolism and antioxidant research, environmental remediation and sustainable agriculture practices, and the declining theme of starch. The findings provide valuable insights into the global pattern of sorghum research and emerging trends, thus directing future multidisciplinary research efforts for enhanced knowledge and sustainable practices in the industrial-agricultural sector.

This study evaluates the efficacy of machine learning models for predicting soil nitrogen (N), phosphorus (P), and potassium (K) concentrations from near-infrared (NIR) spectral data (750–2499 nm). A comparative analysis was conducted on models from three distinct categories: linear-based, non-linear kernel-based, and neural network-based, using data from 145 soil samples collected across four Indonesian provinces. Regularised linear modelling performed best: Ridge Regression achieved R ²/MAE of 0.999/0.00005% for N-Total, 0.868/0.01408% for P-Total, and 0.763/0.01239% for K-Total. The non-linear SVR delivered moderate fit ( R ² = 0.821 for N-Total) but produced the lowest MAE for K (0.00829%) at lower explained variance ( R ² = 0.419). Neural network-based models underperformed the linear baselines. This study demonstrates that, for this dataset, a simpler regularised linear model outperformed more complex architectures, underscoring the critical role of rigorous model selection in developing accurate spectroscopic tools for precision agriculture. Future research could explore hybrid or ensemble methods to combine the strengths of different model types, potentially improving prediction accuracy and robustness.