CN-122001767-A - Network module verification flow generation method, system and storage medium

CN122001767ACN 122001767 ACN122001767 ACN 122001767ACN-122001767-A

Abstract

The application relates to a network module verification flow generation method, a system and a storage medium, and relates to the technical field of embedded communication, wherein the method comprises the following steps of S1, acquiring an AT instruction document to be analyzed, and preprocessing to obtain a plurality of plain text segments; S2, inputting the plain text segment into a preset semantic analysis engine for semantic extraction to obtain a semantic triplet, S3, constructing an AT instruction knowledge graph according to the semantic triplet, and S4, generating a network module verification flow corresponding to the AT instruction document according to the AT instruction knowledge graph. Compared with the prior art, the application can realize zero manual reading, support multi-language and multi-format documents, understand the time sequence among instructions, rely on conditional branches, automatically generate executable graphic verification flow and support incremental updating.

Inventors

WANG JIEKAI
LI NING
ZHOU GUANGHAI

Assignees

广州南方测绘科技股份有限公司

Dates

Publication Date: 20260508
Application Date: 20251225

Claims (10)

1. A method for generating a network module verification process, comprising: S1, acquiring an AT instruction document to be analyzed, and preprocessing to obtain a plurality of plain text segments; S2, inputting the plain text segment into a preset semantic analysis engine for semantic extraction to obtain a semantic triplet; s3, constructing an AT instruction knowledge graph according to the semantic triples; And S4, generating a network module verification flow corresponding to the AT command document according to the AT command knowledge graph.
2. The method according to claim 1, wherein in step S2, the semantic triplet includes an instruction including an instruction name and parameter information, a state indicating that a response result returned by the network module is expected after the instruction is input, a condition indicating a conditional branch based on the state; the semantic analysis engine comprises a rule channel and a deep learning channel; When the plain text segment is input to the semantic parsing engine, judging whether the format of the plain text segment is a standard format or not: if the format of the plain text segment is a standard format, carrying out semantic extraction on the plain text segment through a regular expression and a context-free grammar preset by the rule channel to obtain the semantic triplet; and if the format of the plain text segment is not the standard format, performing sequence labeling and semantic extraction on the plain text segment through a BERT-BiLSTM-CRF model preset by the deep learning channel to obtain the semantic triplet.
3. The method of claim 2, wherein in step S3, according to the instruction, the state and the condition in the semantic triplet, an instruction node, a state node and a conditional branch edge of the AT instruction knowledge graph are respectively constructed; The sides of the AT instruction knowledge graph also comprise time sequence dependency sides which are expressed as dependency relations of the sequence between two nodes; And storing the AT instruction knowledge graph through Neo4j and supporting the Cypher query.
4. The method for generating a network module verification flow according to claim 3, wherein in step S3, when the version of the AT command document is changed, updating the AT command knowledge graph through a version Diff algorithm specifically includes: Generating a new version of AT instruction knowledge graph based on the new version of AT instruction document through the step S1 and the step S2, and mapping the new version and the old version of AT instruction knowledge graph into embedded vectors respectively; And updating the AT instruction knowledge graph of the old version based on the delta patch and synchronously updating the verification process of the AT instruction knowledge graph, or manually correcting the verification process based on the delta patch and feeding back and updating the AT instruction knowledge graph.
5. The method of claim 4, wherein step S3 further comprises, when the AT command knowledge graph is broken, correcting the AT command knowledge graph by using a preset branch logic correction algorithm, and the method specifically comprises: If a path from a current node to a target node does not exist in the AT instruction knowledge graph, and the current node is the instruction node for activating PDN, a virtual node is inserted between the current node and the target node and used for waiting for a response result of a registration state returned by the network module; adding a loop edge from the virtual node to the instruction node for checking registration status, wherein a transition condition of the loop edge depends on a response value of a registration status query instruction of the network module; if the response value does not belong to the instruction preset value set in the registered state or the unregistered state, returning to the virtual node through the circulating edge to wait for retry; and if the response value belongs to the instruction preset value set of the registered state or the unregistered state, correcting the edges of the current node and the target node through a minimum cost flow algorithm.
6. The method for generating a network module verification flow according to claim 5, wherein in step S4, the AT instruction knowledge graph is converted into a graphics block JSON, so as to obtain the network module verification flow, which specifically includes: Determining the execution sequence of each node in the AT instruction knowledge graph through DAG topological ordering based on the time sequence dependent edges in the AT instruction knowledge graph; and generating a condition judgment block based on the conditional branch edge, generating a circulation block based on the virtual node, performing blocking encapsulation on the condition judgment block and the circulation block according to the execution sequence of each node, and then converting the blocking encapsulation into the graphical block JSON, thereby obtaining the network module verification flow.
7. The method of claim 6, wherein different vendors adaptively generate the respective network module verification processes through transfer learning training.
8. The method according to any one of claims 1 to 7, wherein in step S1, the format of the AT instruction document input is any one or more of a scanned or text version PDF, word, HTML and a CSV; When the AT instruction document is a scanning PDF, extracting a text part by utilizing an OCR technology, and carrying out text correction by using a LLM error correction model based on BART; When the format of the AT instruction document is the text version PDF, word or HTML, directly extracting a text part of the AT instruction document; When the AT command document is CSV, extracting a command table in the AT command document through a TableBank model; Classifying the time sequence flow chart in the AT instruction document by CNN, and converting the classified time sequence flow chart into a Mermaid time sequence chart in Mermaid grammar form; And obtaining a plurality of plain text segments based on the text portion, the instruction form and/or the Mermaid time chart.
9. A network module authentication flow generation system, using a network module authentication flow generation method according to any one of claims 1 to 8, comprising: the document input module is used for acquiring an AT instruction document to be analyzed, and preprocessing the AT instruction document to obtain a plurality of plain text segments; The semantic extraction module is used for inputting the plain text segment into a preset semantic analysis engine to carry out semantic extraction to obtain a semantic triplet; the map construction module is used for constructing an AT instruction knowledge map according to the semantic triples; And the flow generation module is used for generating a network module verification flow corresponding to the AT command document according to the AT command knowledge graph.
10. A computer readable storage medium having stored thereon at least one instruction, at least one program, code set, or instruction set, the at least one instruction, at least one program, code set, or instruction set being loaded and executed by a processor to implement the method of any of claims 1-8.

Description

Network module verification flow generation method, system and storage medium Technical Field The present application relates to the field of embedded communications technologies, and in particular, to a method, a system, and a storage medium for generating a network module verification flow. Background With the rapid development of the internet of things technology, a communication module (hereinafter referred to as a "module") has become a core component of various terminal devices accessing a network. Module manufacturers (such as remote, guangdong, mig, and generous) typically issue detailed AT instruction documents along with their hardware products for secondary development by engineers to implement core functions such as dial-up networking, short message transceiving, and positioning. The current industry situation is that each module factory (remote, wide and general, american, square and the like) issues a Chinese/English AT instruction document with a non-uniform format, document contents are mixed, instruction grammar, return examples, use attention, a time sequence chart and a parameter table coexist, an engineer needs to manually extract a key instruction sequence (such as setting an APN (access point name), activating a PDN (public data network), inquiring a resident network and the like before dialing), omission or sequence errors are easy, the same function instructions of different modules have large difference, the document is updated frequently, the workload of manually maintaining a flow template is large, and the version is easy to disorder. The automatic AT document analysis method based on the keyword matching has the following defects that 1) no public tool can automatically read an AT document and generate an executable verification flow, 2) the traditional OCR and keyword matching cannot process instruction dependency relationship and conditional branches, 3) the traditional automatic AT document analysis scheme lacks semantic level understanding, and the analysis precision is low. Therefore, there is a need for an automatic AT document parsing, verification flow chart generation, and complete scheme of graphical tool execution importing, which realizes intelligent crossing of "document in and flow out". Disclosure of Invention Based on this, it is necessary to provide a network module verification flow generation method and system capable of realizing zero manual reading, supporting multi-language and multi-format documents, understanding the timing sequence between instructions, relying on conditional branching, automatically generating executable graphic verification flow and supporting incremental updating, aiming AT the technical problem of poor AT document analysis efficiency and precision in the prior art. In order to solve the technical problems, the technical scheme of the invention is as follows: In a first aspect, a method for generating a network module verification flow includes: S1, acquiring an AT instruction document to be analyzed, and preprocessing to obtain a plurality of plain text segments; S2, inputting the plain text segment into a preset semantic analysis engine for semantic extraction to obtain a semantic triplet; s3, constructing an AT instruction knowledge graph according to the semantic triples; And S4, generating a network module verification flow corresponding to the AT command document according to the AT command knowledge graph. In a second aspect, a network module verification process generating system, using a method for generating a network module verification process for automatically parsing a document as described above, includes: the document input module is used for acquiring an AT instruction document to be analyzed, and preprocessing the AT instruction document to obtain a plurality of plain text segments; The semantic extraction module is used for inputting the plain text segment into a preset semantic analysis engine to carry out semantic extraction to obtain a semantic triplet; the map construction module is used for constructing an AT instruction knowledge map according to the semantic triples; And the flow generation module is used for generating a network module verification flow corresponding to the AT command document according to the AT command knowledge graph. In a third aspect, a computer readable storage medium has at least one instruction, at least one program, a code set, or an instruction set stored thereon, where the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement a network module verification process generation method as described above. Compared with the prior art, the technical scheme of the invention has the beneficial effects that: The invention provides a vendor self-adaptive AT instruction knowledge graph self-growth mechanism for the first time, which performs semantic level extraction on vendor documents by constructing an instruction-st