Abstract
<jats:p>The article presents the conceptual architecture of an intelligent dynamic pricing model for cargo transportation automation systems operating on the basis of digital platforms as organizational systems requiring coordination of interaction between multiple market participants. The article considers the problem of price formation in conditions of high variability of transaction parameters and the need to process data from unstructured sources – text correspondence and audio recordings of negotiations between customers and carriers. A system is proposed that includes four main blocks: a data extraction module based on natural language processing methods for structuring information about routes, cargo characteristics and time parameters; a predictive module using gradient boosting and recurrent neural networks to account for nonlinear interactions between features and long-term demand trends; a mechanism for automatically adapting model parameters when new data is received on completed transactions, which implements online training to adjust weights without complete retraining; the block for forming the final price recommendation based on the platform strategy. Mathematical relations are described for calculating the forecast price, adjusting the weights of the model and determining the final recommendation. The classification of price adjustment strategies depending on market conditions is presented: maximizing load, maximizing margin, balancing and competitive pressure. The stages of practical implementation of the model are formulated, including data preparation with a volume of at least ten thousand records, training of basic algo-rithms, integration with the platform and pilot testing on a limited segment of routes. The technical requirements for the infrastructure and metrics for evaluating the quality of the system are defined. The flexibility of the proposed architecture is noted, which makes it possible to adapt the composition of features and algorithms to the specifics of specific transport platforms. The results of the study are of interest to managers of digital transport platforms, specialists in the field of organizational systems management and researchers involved in the automation of decision-making in logistics.</jats:p>