In China’s road freight market, there exists the issue of decentralized transportation capacity, requiring online freight platforms to allocate logistics tasks according to travel preferences of freight operators and optimize the matching of vehicles and goods. Thus, a data-model collaborative-driven classification method for truck logistics patterns is proposed. Firstly, based on the truck trajectory data, 5 characteristic parameters including turning radius, activity entropy, average daily travel locations, average daily travel time, and average daily travel distance are constructed. After dimensionality reduction by Principal Component Analysis (PCA) and clustering by K-Means Cluster, trucks are categorized into 3 types of logistics patterns: long-distance round-trip, short-distance fixed-point, and short-distance multi-point. Secondly, the network motif identification technology from graph theory is introduced, generating a directed logistics network from truck OD data, using the DotMotif Algorithm for motifs recoginition and selecting p-value to test the significance of motifs. Finally, by deeply analyzing the connection between network motifs and the typical topological structures of truck travel chains, the differences in significant motifs within the logistics networks of different logistics modes of trucks are explained, verifying the accuracy of the the truck logistics pattern classification results.