169 results about "Logical consistency" patented technology

The invention provides a large model thinking chain knowledge distillation method and system based on anti-factual reasoning, and belongs to the technical field of artificial intelligence and natural language processing. Comprising the steps that minimum semantic disturbance is applied to a key position of an original reasoning task data set, and an anti-fact problem is constructed; based on the original problem and the anti-fact problem, calling a large language model to generate a multi-view reasoning chain; a high-loyalty distillation data set is screened out through causal intermediary analysis, and thinking chain consistency optimization training is carried out on the large language model; and deploying the large language model after optimization training to a downstream task. According to the method, by analyzing the anti-fact space of the input problem and quantifying the causal contribution of the inference chain to the final decision, the causal reliability and logic consistency of the large model thinking chain can be improved, and then effective data support is provided for downstream inference application.

The invention relates to the technical field of equipment management, in particular to an equipment full-life-cycle digital twinborn decision-making system enabling a large model. Comprising a digital twin modeling unit; a large model enabling analysis unit, wherein the large model enabling analysis unit adopts a hydroelectric equipment multi-modal causal constraint analysis model; a whole-process closed-loop management and control unit; and an intelligent decision output unit. According to the method, the multi-modal causal constraint analysis model adaptive to the working condition of the hydroelectric equipment is constructed, and a causal chain verification backtracking mechanism is introduced, so that the problem that the reasoning result lacks logic verification is effectively solved, the logic consistency of a fault reasoning conclusion is guaranteed, and the reliability of decision output is improved; through a scene adaptation mode of'pre-training + fine tuning 'of a large model, multi-modal feature fusion processing and deep linkage of a workflow engine and a digital twinborn body, full-life-cycle management requirements of equipment are fully covered, and the refinement and intelligence level of hydroelectric equipment management is further improved.

The invention discloses an operation data processing method and system based on a compliance verification library and a risk control plan, and belongs to the technical field of data compliance and risk control, and the method comprises the steps: obtaining an execution plan of a target operation task and risk control strategy information of the execution plan, analyzing operation data items, distributing corresponding compliance verification time windows and levels, and forming an operation data verification scheduling table; according to the verification level and the business domain, a rule set is extracted from a compliance verification library, and rules comprise predefined rules, model output rules and historical violation rules; field-level matching, logic consistency verification and cross-source linkage verification are executed in the verification time window; when the verification result meets the triggering condition, generating an abnormal event, and executing response operations such as delay, recheck or interception; and finally, writing the execution state and the verification result into an abnormal log queue for risk control strategy parameter updating or re-evaluation. According to the method, the dynamic compliance verification and strategy closed-loop optimization of the operation data are realized, and the intelligent compliance control capability of the system is improved.

The invention discloses a large language model multi-source inference knowledge graph question answering method and system, and the method comprises the steps: extracting entities in a question, linking the entities with knowledge graph entities, and determining a subject entity set; searching paths in the knowledge graph by adopting a graph constraint reasoning method, and obtaining candidate answers and reasoning paths; detecting incomplete path coverage or answer dimension deficiency by using a large language model, triggering question decomposition to generate logically complementary sub-questions, and obtaining answers; establishing an inference source coordination layer, a coordination graph constraint inference source, a planning-retrieval inference source and a sub-problem inference source for multi-source evidence; and standardizing the output of each inference source into a structured evidence triple, sorting and selecting the first K evidences according to relevance, and generating a final answer by utilizing a large language model to carry out inductive inference. According to the method, reasoning can be dynamically adjusted, multi-source evidences can be effectively aggregated, logic consistency and semantic integrity are guaranteed, and the integrity and accuracy of knowledge graph multi-hop questions and answers are improved.

The invention relates to the technical field of document automatic processing, and discloses a bidding and tendering document generation and automatic review method and system based on a large language model.The method comprises the steps that firstly, a source heterogeneous constraint document is obtained and analyzed, and a structured semantic unit is obtained; secondly, based on the structured semantic unit, constructing a dynamic abstract semantic graph; then, based on the topological structure of the dynamic abstract semantic graph, controlling the large language model to generate contents node by node by adopting a multivariate constraint guide mode, and performing instant verification; when a user modifies the content, executing a bidirectional consistency propagation algorithm based on a graph, reversely mapping text change into node attribute update, and automatically detecting global logic conflicts; and finally, combining the final text of each node in the graph, and outputting a complete bidding and tendering file. According to the method, through structured modeling and bidirectional consistency maintenance, the accuracy, controllability and logic consistency of the generated content are improved, and automatic consistency maintenance of the whole life cycle of the document is realized.

The invention discloses an automatic causal structure generation method based on semantic representation and logical reasoning of a large language model. The method comprises the following steps: acquiring an input text; performing semantic coding and clustering on the obtained input text by utilizing a large language model, and establishing a candidate causal variable set; causal relationship detection is carried out on the established candidate causal variable set based on anti-fact intervention and do-calculation; performing causal direction judgment, and generating a directed acyclic causal graph meeting logic consistency; and on the basis of the generated directed acyclic causal graph, natural language interpretation is generated by using a large language model, and logic consistency closed-loop verification is carried out. According to the method, automatic generation from the natural language to the causal structure is realized, the causal variable set is automatically extracted and constructed from the unstructured natural language text, the defects that variables need to be manually defined and modeling depends on field experts in the existing causal modeling process are avoided, and the labor cost and professional threshold of causal structure construction are remarkably reduced.

The embodiment of the invention provides a knowledge graph fusion method and system based on a large model, and the method comprises the steps: carrying out the data standardization, entity feature enhancement and relation semantic annotation of heterogeneous knowledge graphs from different sources; through semantic similarity calculation, context reasoning and alignment confidence evaluation, a multi-stage and multi-mode entity alignment mechanism is constructed in combination with a large language model, and cross-source entity matching is performed on entities in heterogeneous knowledge maps of different sources; based on conflict detection, dynamic weight distribution and a conflict resolution strategy, generating a data fusion result, and unifying multi-source data; and generating a high-quality unified knowledge graph through missing relationship prediction, logic consistency verification and an incremental updating mechanism. According to the knowledge graph fusion method and device, full-process automation, precision and dynamics of knowledge graph fusion are realized, the problems of high manual dependence, weak semantic processing capability, poor cross-domain adaptability and the like in the prior art are effectively solved, and the efficiency and quality of knowledge graph fusion are remarkably improved.

The invention relates to the technical field of unmanned aerial vehicle control and management, in particular to an unmanned aerial vehicle parameter consistency configuration method based on template management, which adopts a structured parameter configuration template in a JSON format, comprises four blocks of metadata, parameter definition, constraint rules and digital signatures, establishes an MAVLink communication link with an unmanned aerial vehicle flight controller through a template management terminal, and provides an unmanned aerial vehicle parameter consistency configuration method. And executing an automatic configuration process based on a finite-state machine, and ensuring parameter logic consistency by using a difference comparison and constraint verification mechanism. According to the method, temporary storage, batch writing and atomic submission of parameter changes are achieved through the atomization transaction submission protocol, the integrity of the configuration process is guaranteed, the digital signature auditing log containing the timestamp, the device identifier, the template hash and the change details is generated after configuration is completed, whole-process tracing is supported, and the method is high in practicability and easy to popularize. Headless mode operation, parameter white list control and hardware abstraction layer adaptation are supported, and high automation, standardization and safety of unmanned aerial vehicle parameter configuration can be achieved.

The invention provides an interaction quality assurance method for improving the reliability of big language model questions and answers, and relates to the technical field of big language model questions and answers, and the method comprises the following steps: responding to an initial natural language inquiry, generating and outputting a guiding question, endowing each structured field in a preliminary medical description with an initial medical semantic weight, in response to the intervention instruction, generating an enhanced medical description with adjustable weight; identifying a subject anaphora relationship in the current interaction sequence in real time, and performing subject label calibration and logic isolation on context data generated by interaction based on an identification result so as to construct a structured session context with subject consistency; performing integrity scoring and logic consistency verification on the enhanced medical description in the structured session context; and when the verification is passed, packaging the enhanced medical description, the weight information corresponding to the enhanced medical description and the related constraint strategy called from the medical constraint strategy library into a structured cue word, and inputting the structured cue word into the large language model to obtain a medical question and answer result.

The invention discloses a commercial password evaluation report automatic generation and quality control system based on an artificial intelligence technology, and the system comprises a multi-modal data processing and knowledge graph construction module which constructs a knowledge graph of a password evaluation field based on the multi-modal data related to the commercial password evaluation field; the field large model fine tuning module is used for constructing an instruction fine tuning data set in the password evaluation field, performing field adaptation training on the pre-trained large language model by adopting a hybrid fine tuning strategy, and generating a password evaluation field expert large model; the report automatic generation module is used for automatically generating a commercial password evaluation report based on the knowledge graph and the expert large model; the quality control module is used for carrying out automatic quality verification and evaluation on the report; and the system integration and optimization module is used for integrating the modules into an interactive application system and continuously optimizing the knowledge graph and the expert large model. According to the method, the commercial password evaluation report can be automatically generated, and the correctness and logic consistency of the report can be ensured.

Multi-stage hybrid network integrates relationship regularization links and explainable elements to improve alignment with human values, explainability, robustness, and efficiency. The network comprises neural components, event prediction elements, and probability models across multiple stages, with relationship constraints enforcing structured knowledge representation. Explainable elements provide interpretable rationales for decisions, enhancing transparency. Training incorporates supervised learning, human-guided refinement, semi-automated knowledge engineering, and adversarial robustness techniques. A Socratic reasoning module detects contradictions and refines outputs for logical consistency. Indexed model elements enable dynamic memory optimization for improved efficiency. Candidate outputs may be scored, verified, or selected using neural and symbolic criteria. The invention supports retry loops and configurable subsystem pipelines to improve output quality. Applications include text generation, speech recognition, translation, and decision support. By combining structured constraints, human oversight, and modular architectures, the system improves the trustworthiness, safety, and adaptability of AI systems across diverse modalities and tasks.

The invention discloses a large model illusion detection and inhibition method and system based on logic type guidance, and belongs to the technical field of large model illusion detection, and the method comprises the following steps: logic diagram matching, multi-hop path screening, consistency verification, semantic reconstruction correction and text verification output. According to the method, a text reasoning logic diagram is constructed, a multi-hop reasoning relation, causal constraints and a logic sequence which are implied in a natural language are subjected to explicit structured representation, and a knowledge graph is introduced to perform logic consistency verification. On the basis, based on logic type constrained multi-hop path matching and global consistency evaluation, an implicit reasoning relation between long-distance entities is captured from the structural level, hallucination missing detection caused by logic chain breakage or semantic drift in the deep reasoning process is effectively avoided, the accuracy and coverage capacity of large model hallucination detection are remarkably improved, and the method has the advantages of being high in practicability and easy to popularize. The problem that in the prior art, complex inference type illusion in large model generation content is difficult to effectively recognize is effectively solved.

The invention discloses a transformer area high-frequency acquisition data quality improvement method, system and device and a medium, and the method comprises the steps: constructing a multi-level rule field, converting a rule into a continuous potential energy function with directivity and range control, and forming a regular force field in a semantic space; defining a field state function, and generating a plurality of field completion candidate values through rule-driven nonlinear evolution reasoning; a candidate value path diagram is constructed, nodes of the path diagram represent candidate values, and edge weights are determined according to semantic and rule association features among the candidate values; calculating the global path participation degree of each node based on the path diagram, and selecting an optimal complement value according to the global path participation degree; and executing multi-dimensional consistency verification and high-confidence judgment, and marking and supplementing the credibility level according to a verification result, thereby realizing closed-loop improvement of data quality. According to the method, the accuracy, the logic consistency and the high confidence of missing field supplementation of the high-frequency acquisition data of the transformer area in a complex interference environment are remarkably improved.

The invention discloses a context understanding and memory management system and method in large-model multi-round dialogues. The system comprises a five-layer distributed micro-service technology architecture; wherein the business logic layer is integrated with a hierarchical memory management module, a dynamic context coding module, a semantic association engine module and a personalized adaptation module which cooperate with one another; the hierarchical memory management module is responsible for retrieval, hierarchical storage and dynamic updating of dialogue history; the dynamic context coding module is responsible for generating structured context representation; the semantic association engine module is responsible for constructing a cross-round semantic association network; the personalized adaptation module is responsible for generating personalized candidate responses. The technical problems of context loss, memory attenuation, incoherent semantic understanding, insufficient personalized adaptation, low resource utilization rate and the like of an existing large-model multi-round dialogue system are solved, and logic consistency guarantee, memory durability maintenance and personalized experience improvement in a long dialogue scene are realized.

The invention discloses a one-stop adaptive architecture and method for constructing a trusted electromagnetic radar field multi-mode large model, and the method comprises the steps: obtaining unsupervised text data of an electromagnetic radar field, and carrying out the incremental pre-training of a preset base language model, so as to construct a fact base model; training is carried out by using supervised question and answer data in the field of electromagnetic radars through an efficient parameter fine tuning method, and alignment of instruction following and dialogue interaction capability is realized; performing multi-modal fine tuning by using the multi-modal data containing the radar image or the electromagnetic signal; generating a preliminary answer by using the multi-modal fine-tuned model; and starting an auto-reflection module to check the logic consistency and the fact accuracy of the preliminary answer so as to generate a final checked or corrected answer. According to the method, the credibility construction process of the model is decomposed into clear and progressive stages, and the problem that knowledge depth, dialogue ability and multi-modal understanding are difficult to consider in a single fine tuning method is solved.

The invention provides a knowledge graph construction method and device, a question answering method and device, equipment, a medium and a program product, and relates to the technical field of intelligent question answering, the knowledge graph construction method comprises the following steps: cutting an original text into a plurality of text blocks with coherent contents to determine nodes in a knowledge graph; determining edges in the knowledge graph and initial weights of the edges according to a cross reference relationship between contents of the text blocks and / or similarity between themes; based on the relation type to which the edge belongs, adjusting the initial weight to determine the target weight of the edge; and constructing the knowledge graph based on the node, the edge and the target weight. According to the embodiment of the invention, through modular analysis of papers and dynamic perception of the association relationship, multi-level knowledge node and implicit topic association are fused, the logic relationship between modules is accurately quantified, and the constructed knowledge path can effectively capture the logic consistency and cross-document reference transitivity in the document, so that the reliability of the document is improved. And the multi-hop reasoning capability in a complex problem is obviously enhanced.

The invention relates to the technical field of artificial intelligence, in particular to an official document semantic verification and intelligent error correction method and system based on a large language model, and the method comprises the steps: executing document analysis and preprocessing to form structured text data; key entities and association relationships are extracted from a large language model subjected to official document special corpus fine adjustment and stored as an official document knowledge graph; based on the official document knowledge graph, verifying entity expression uniformity, fact data matching degree and context logic consistency; marking and checking inconsistency, and generating a modification suggestion with a modification basis by using a large language model subjected to official document special prediction fine adjustment. The document semantic verification and intelligent error correction method realizes document semantic verification and intelligent error correction, and is suitable for processing scenes of documents (such as reports, notices, requests, replies and the like) of administrative organizations and enterprises and public institutions.

The invention relates to the technical field of big data, and particularly discloses a natural language and semantic mapping-based business data model automatic construction method, which comprises the following steps of: performing structured analysis on a natural language demand input by business personnel, and extracting a dimension entity, a measurement index and a user intention; performing semantic matching on the business terms and the metadata fields by using a Sentence-BERT sentence vector model to generate a field mapping list; constructing a data consanguinity map based on a historical SQL, and reasoning an optimal multi-table association path through a shortest path algorithm by taking a center table as a root node; automatically generating a target wide table structure definition and a complete SQL (Structured Query Language) code by combining a template engine; and performing grammar and logic consistency verification on the output statement. According to the method, full-automatic, high-precision and interpretable end-to-end modeling from natural language requirements to executable wide tables can be realized, the data modeling efficiency and accuracy are remarkably improved, and the participation threshold of business personnel is reduced.

The invention relates to the technical field of business process modeling, in particular to a meta-model-driven interactive business process intelligent modeling method and system. The method comprises the following steps: constructing a meta-model framework containing core elements such as activities, events and gateways, and establishing a meta-model constraint rule base to define a structure legality rule of a BPMN2.0 specification; based on a domain ontology Schema and the domain exclusive large language model after instruction fine tuning, extracting a business entity and a process relationship thereof from the unstructured text; carrying out information integrity verification and logic consistency detection on the extracted BPMN elements by utilizing a constraint satisfaction algorithm and Petri net reachability analysis; displaying a difference label of the process sketch through an interactive verification interface, and realizing semantic correction and model iteration of a user on the process elements based on a natural language feedback mechanism; an XML code is generated by adopting a template-driven code generation strategy, and a flow chart is generated through a BPMN parser. According to the method, the business process modeling efficiency can be remarkably improved, and the manual intervention proportion in the modeling process is reduced.

The invention discloses a two-stage reinforcement learning SAT solving method for structure perception and a related device, and belongs to the field of chip design. Chip signal polarity and logic consistency information is introduced in a variable initialization stage, an initial feature vector for structure perception is constructed, and the limitation that traditional graph modeling can only represent a local logic relation is broken through; a two-stage reinforcement learning framework composed of a main path and a sampling path is designed, the sampling path is used for exploring a chip logic solution space, the main path is used for stably utilizing an optimal verification strategy, and the problems of unstable training convergence, single strategy and the like are effectively relieved. Based on a reward-guided feature updating mechanism, dynamic evolution of chip signal features and enhanced feedback are explicitly bound, the interpretability of a solver is improved, and the difference between neural reasoning and chip logic symbol reasoning is reduced. A signal assignment decision is executed based on GNN multi-round message passing results, and a verification track is continuously optimized, so that efficient chip design verification capability can be obtained without depending on an artificial heuristic mode.

This application relates to the field of fintech and discloses a dynamic hybrid parsing and deterministic verification system and method for fuzzy bond instructions. The system constructs structured data through an instruction preprocessing module; a routing decision module calculates complexity scores based on structure, semantics, and business characteristics, and generates path selection instructions using a resource-aware dual-threshold strategy; a hybrid parsing module responds to instructions, employing a rule engine to handle low-complexity tasks and a large-scale model with enhanced retrieval to handle highly ambiguous tasks; a fallback verification module performs confidence-weighted integration and logical consistency checks, and repairs anomalies through issuer set operations and edit distance matching; a mapping generation module constructs standard query statements using an abstract syntax tree; and a closed-loop optimization module dynamically updates parameters based on user interaction feedback. This invention effectively balances computational efficiency and parsing accuracy, eliminates model illusion risks, and achieves adaptive evolution of the system.

This invention relates to a method for checking the consistency of row-level data files and index files based on the OpenGauss database. This method can be used to check the consistency of the physical files of the main table, the physical files of the index table, and the logical consistency of the physical files of the main table and the index table. This consistency checking method overcomes the shortcomings of traditional data file consistency verification methods, such as delayed startup and the possibility of business failures or even database instance anomalies due to check failures. Through this method, DBAs can periodically and proactively perform full consistency checks on the physical files of specified tables during daily operations and maintenance. It can also check the consistency relationship between the physical files of the b-tree indexes related to the main table and the physical files of the main table, thereby enabling timely detection of data file consistency issues in the database and taking corresponding remedial and mitigation measures to minimize potential impact and losses.

The invention provides a training sample generation and model training method and device, a storage medium and a product. The training sample generation method comprises the steps of obtaining original dialogue materials of a target application scene, wherein the original dialogue materials comprise a model input question and a model output reply; performing named entity recognition on the model output reply, and extracting N target entities, N being greater than 0; on the basis of the N target entities and the incidence relation between the different target entities, the original dialogue material is disassembled into a plurality of question and answer pairs, so that each question and answer pair corresponds to one target entity or a group of target entities with the incidence relation; the multiple question and answer pairs are input into a language model, so that the language model conducts logic consistency detection on all the question and answer pairs and outputs corresponding detection results, and the detection results comprise inconsistent conclusions and combination of magic view types or consistent conclusions; and based on each question and answer pair and the corresponding detection result, generating a first training sample used for training the magic vision detection model.

This invention discloses a multi-dimensional dynamic peripheral security verification method and system for the boot phase of wireless communication devices, belonging to the field of embedded system security technology. The method is executed after the second stage of the U-Boot boot process of the wireless communication device and before jumping to the operating system kernel. It includes: triggering a security verification process and reading a trusted peripheral whitelist, where each entry contains the peripheral's fingerprint information; collecting fingerprint information for the peripheral to be verified using a differentiated method based on its interface type, comparing it with the whitelist entries, and performing a logical consistency check for USB interface peripherals; and determining the peripheral's risk level and executing corresponding security response actions based on a comprehensive assessment of firmware hash consistency, timing feature deviation, and logical consistency check results. By integrating multi-dimensional dynamic fingerprint verification, this invention enables deep security verification of peripherals during the embedded boot phase, effectively preventing identifier forgery and firmware tampering, and improving device boot security.

The invention relates to the technical field of real-time synchronization of knowledge maps, and discloses a multi-source data knowledge base supporting real-time synchronization and an intelligent agent system. The system comprises an event access module, a content deconstruction module, an entity matching module, a data anchoring module, a synchronous decision module and a network execution module. The system deconstructs semantics, data and identifiers by receiving a multi-source event stream in real time, and matches or creates entities and relationships in a dynamic entity relationship network. A data version is anchored to an entity, a synchronous decision instruction is intelligently generated according to a data source and an operation type, and instant reconstruction and updating of a network topology are driven. According to the system, millisecond-level real-time evolution of the knowledge base is achieved, logic consistency of multi-source data fusion is guaranteed through strategic synchronous coordination, and the real-time performance and reliability of knowledge support are improved.

The invention discloses a science and technology document intelligent error correction and logic consistency verification method based on a large model, and belongs to the technical field of artificial intelligence and document intelligent processing. Comprising the steps of science and technology document access and structured analysis; carrying out semantic unit division and feature extraction on the science and technology document; performing multi-level intelligent error correction based on a large model; checking the logic consistency of the science and technology documents; and fusing and outputting error correction and verification results. According to the method, automatic identification and prompt of language errors, professional errors and cross-paragraph logic inconsistency problems in the document are realized through comprehensive analysis of the science and technology document structure, professional semantics and logic relationship, so that the normalization and technical reliability of the science and technology document are improved.

The invention relates to a key technical scheme intelligent screening and positioning method and system oriented to integrated circuit research and development, and the method comprises the steps: respectively extracting technical entities representing specific process steps and device structures and legal entities representing legal protection boundaries through building a semiconductor field heterogeneous data stream and utilizing a large language model; constructing a bimodal semantic topology network comprising technical association, legal association and cross-modal alignment edges; heterogeneous graph attention representation learning is executed, and deep semantic dependence between nodes is captured; generating a global semantic representation vector representing the logic consistency of the technical scheme based on the topological connection density and the cross-modal semantic alignment strength of the nodes; and finally, accurately positioning a key technical scheme based on the logic consistency confidence coefficient. According to the method, the process-protection logic consistency can be automatically verified, the problems that traditional keyword retrieval is loud in noise and a semantic model lacks logic support are solved, and objective and high-reliability data support is provided for technology pre-research, design avoidance and research and development path planning.

The application discloses a multi-agent collaborative reasoning method and system based on memory sharing and conflict resolution, comprising: constructing a hybrid structured memory pool containing a vector database for semantic recall and a graph database for storing logical structure; deploying functionally heterogeneous multi-agent to normalize original multi-source data into observation nodes and write into the graph database; using the graph database trigger mechanism to listen to the writing event, and automatically triggering the reasoning to generate the hypothesis node and establish the deduction edge when matching the preset rule; detecting whether there is a logical conflict for the hypothesis node of the same subject, and triggering dynamic arbitration if there is; the arbitration mechanism calculates the weighted score based on the confidence, data source credibility and context consistency calculated based on KL divergence, and finally outputs the reasoning conclusion and evidence chain path. The application breaks the information island through the hybrid memory pool, realizes the automatic reasoning by using the trigger, and significantly improves the accuracy, logical consistency and explainability of reasoning in complex scenarios.

The invention provides a text classification mapping method based on hierarchical semantic matching and fact data fusion, and relates to the technical field of text classification. The problems that in an existing text classification mapping technology, cross-category logic errors exist, the matching accuracy rate and the coverage rate are difficult to balance, and noise-containing external fact data cannot be effectively purified and fused are solved. According to the method, text classification mapping is realized through a two-stage process: firstly, data is preprocessed and vectorized, then full-text and high-level semantic similarity is subjected to weighted fusion, and prediction mapping is generated by using double thresholds; and then purifying external fact data, finally fusing the two types of mappings and labeling sources, and outputting a high-quality mapping table. According to the method, the mapping accuracy and logic consistency are effectively improved, noise data are filtered to enhance the reliability of a result, the accuracy and the coverage rate are balanced through a double-threshold strategy, high automation of the process is realized to improve the efficiency, and the technical framework is high in universality and can adapt to multi-field text classification mapping requirements.

The invention provides a customs code preprocessing method and system based on a structured tax number tree, and the method comprises the steps: analyzing a customs tax rule document through employing a natural language processing technology, and extracting the hierarchical structure of tax numbers, the commodity description corresponding to each tax number, and a commodity classification condition; establishing a tax number structure tree containing a father-child relationship, a brother relationship and an exclusion relationship; extracting feature words of multiple dimensions from the commodity description by adopting a TF-IDF algorithm to form a many-to-many mapping table; identifying an exclusion condition in the commodity classification conditions through a regular expression, converting the exclusion condition into an IF-THEN rule, binding the IF-THEN rule with a corresponding tax number in the tax number structure tree, and performing rule logic consistency verification; obtaining feature words of each tax number based on a many-to-many mapping table, calculating a semantic distance of each tax number, constructing a semantic vector space, and clustering similar tax numbers; and correcting the clustering result by using the exclusion item rule base, and outputting the tax number semantic association network and the clustering result for customs coding classification. According to the invention, automatic processing and intelligent maintenance of tax rule information can be realized.