Objective To address the challenges of fragmented knowledge, inconsistent standards, and heavy reliance on expert judgment in larch plantation site quality evaluation, this study developed a knowledge graph–based service system aimed at enabling intelligent management, efficient querying, and precise application of site evaluation knowledge.

Method An ontology engineering method was adopted to construct an ontology model for site quality evaluation. The DeepSeek large language model was used to extract structured knowledge from multi-source heterogeneous data (literature, technical regulations, research reports) on larch distribution and site quality evaluation. A larch knowledge graph including geographical distribution, environmental factors, and evaluation indicators was established. A knowledge service system was developed based on the Neo4j graph database to realize intelligent querying, factor correlation analysis, and visual display of site knowledge.

Result An ontology model for larch site quality evaluation was constructed, consisting of 195 entity classes, 13 types of semantic relationships, and 35 constraint rules, forming a unified conceptual system and semantic framework. A site quality evaluation knowledge graph was developed, encompassing 1 424 entity nodes and 3 152 semantic relationships, integrating the distribution and environmental information of Larix principis-rupprechtii, Larix olgensis, Larix gmelinii, and Larix kaempferi. The validation results indicate that the precision of geographical distribution information was 94.2%, while for environmental factor information, the numerical range compliance rate and categorical consistency accuracy were 85.0% and 88.3%, respectively. Additionally, the interval compliance rate for Site Index reached 88.3%. System performance testing indicated that the accuracy rates of the intelligent query, factor correlation analysis, and intelligent question-answering modules were 89%, 85%, and 87%, respectively.

Conclusion The proposed knowledge service system for larch site quality evaluation based on knowledge graphs effectively integrates fragmented and heterogeneous site evaluation knowledge, providing a standardized knowledge management tool for larch afforestation selection and contributing to the improvement of the level of intelligence of forest site quality evaluation.