Search

CN-121981687-A - ChatOps-based IT operation and maintenance real-time cooperation and work order automatic generation method

CN121981687ACN 121981687 ACN121981687 ACN 121981687ACN-121981687-A

Abstract

The invention belongs to the technical field of automatic operation and maintenance, and particularly relates to a ChatOps-based IT operation and maintenance real-time collaboration and worksheet automatic generation method, which generates structured data representing intents and entities by monitoring and analyzing operation and maintenance dialogues; the method comprises the steps of carrying out a first operation on a work order data to be filled, carrying out a second operation on the work order data to be filled, carrying out a first operation on the work order data to be filled, carrying out a second operation on the work order data to be filled, carrying out a third operation on the work order data to be filled, carrying out a fourth operation on the work order data to be filled, and carrying out a third operation on the work order data to be filled, and finally synchronously driving the work order system to establish a structural chemical order and an automatic engine to execute operation, thereby realizing a closed loop from unstructured dialogue to a traceable and executable operation flow, and improving the cooperation efficiency and the disposal accuracy.

Inventors

  • LIANG YUGANG

Assignees

  • 上海速擎软件有限公司

Dates

Publication Date
20260505
Application Date
20260409

Claims (10)

  1. 1. The IT operation and maintenance real-time cooperation and work order automatic generation method based on ChatOps is characterized by comprising the following steps of: based on the continuous monitoring of the instant dialogue content in the chat platform, collecting the instant dialogue content containing unstructured text; Understanding and characterizing the instant dialogue content through a configured semantic analysis model, and calculating to generate a first structured data set for characterizing dialogue core handling intention and key entity information; Performing task allocation between a preset work order information field and an automatic operation and maintenance operation to be triggered according to the first structured data set to obtain a second structured data set used for being filled into a standard chemical engineering single template and an automatic operation instruction set used for driving an automatic operation and maintenance engine to execute; Under the constraint of a preset work order integrity rule and an information accuracy verification rule, carrying out constraint verification and complementation on the second structured data set to obtain a second structured data set after verification and complementation, and generating a third control instruction based on the first structured data set and the automatic operation instruction set, wherein the third control instruction is used for representing the arrangement of execution logic and sequence of the automatic operation instruction set; and outputting the second structured data set after the verification and complementation to a creation interface of a work order management system as a work order generation instruction, and outputting the third control instruction to a control loop of an automatic operation and maintenance engine as an automatic arrangement instruction.
  2. 2. The method for real-time collaboration and automatic generation of a work order based on ChatOps IT operation and maintenance of claim 1, wherein the matching degree between key entity information in the second structured data set after completion of verification and corresponding key entity information in the first structured data set is not lower than a first preset threshold value, and the matching degree between an automation operation instruction set arranged by the third control instruction and dialogue core handling intention characterized in the first structured data set is not lower than a second preset threshold value.
  3. 3. The IT operation and maintenance real-time collaboration and work order automatic generation method based on ChatOps of claim 2, wherein computing generates a first structured data set characterizing dialogue core treatment intent and key entity information, comprising: Normalizing the instant dialogue content to obtain a normalized dialogue text sequence; Mapping each word element into a dense vector through a word vector model in an embedding layer based on the normalized dialogue text sequence to obtain a word vector sequence; Based on the word vector sequence, performing forward and backward sequence modeling through a two-way long-short-term memory network in the coding layer to obtain semantic vector representation of the dialogue fused with the context information; Based on the semantic vector representation of the dialog, nonlinear transformation is performed through a full connection layer in a first decoding branch, probability distribution is calculated through a Softmax function, and an intention classification label representing the dialog core handling intention and a corresponding confidence level are obtained.
  4. 4. The IT operation and maintenance real-time collaboration and work order automatic generation method based on ChatOps of claim 3, wherein computing generates a first structured data set characterizing dialogue core treatment intent and key entity information, further comprising: Based on semantic vector representation of the dialogue, performing sequence labeling through a conditional random field model in a second decoding branch, classifying each word element in the normalized dialogue text sequence, and obtaining a preliminary key entity information set, wherein each element in the preliminary key entity information set comprises an entity type label and a start-stop position index of the entity in the normalized dialogue text sequence; The method comprises the steps of carrying out context coding on a normalized dialogue text sequence, respectively obtaining semantic vector representation of a current round of dialogue and semantic vector representation of each history round of dialogue, splicing the semantic vector representation of the current round of dialogue with the semantic vector representation of the history round of dialogue based on a preset context window, calculating the dependency weight of the current semantic vector on the history semantic vector through an attention mechanism, carrying out weighted summation on the history semantic vector according to the dependency weight, and fusing the weighted summation result with the current semantic vector to obtain the enhanced semantic representation fused with the history dialogue context.
  5. 5. The method for automatically generating an IT operation and maintenance real-time collaboration and a work order based on ChatOps of claim 4, wherein computing generates a first structured data set characterizing dialogue core treatment intent and key entity information, further comprising: Based on the enhanced semantic representation, calculating vector cosine similarity between different entity references in the preliminary key entity information set through a coreference resolution layer, merging a plurality of entity references with cosine similarity exceeding a preset threshold value into the same entity identifier, and acquiring a unique key entity information set which is used for eliminating the reference ambiguity and has a uniform identifier; and based on the intention classification tag and the unique key entity information set, assembling and serializing the data format according to a preset key value through an information fusion layer to obtain a first structured data set representing the dialog core handling intention and the key entity information.
  6. 6. The method for real-time collaboration and automatic generation of a work order based on ChatOps's IT operation and maintenance of claim 5, wherein the task allocation between the preset work order information field and the automatic operation and maintenance operation to be triggered comprises: Based on the intention classification labels and the corresponding confidence coefficient in the first structured data set, matching a preset work order template selection rule by combining a decision tree classifier with confidence coefficient threshold verification, and filtering paths of which the confidence coefficient does not meet a threshold value to obtain a target work order template identifier; Based on the target work order template identifier, inquiring a template-field association database by combining a field priority ordering database inquiry function to obtain a requisite field name set, an optional field name set and a field priority ordering result; Based on the unique key entity information set and the entity identifier in the first structured data set, mapping is completed and the entity identifier is associated according to the iteratively updated entity type-work list field corresponding relation lookup table through the entity type-field mapping function supporting dynamic updating, and a preliminary field filling mapping relation is obtained.
  7. 7. The method for real-time collaboration and automatic generation of a work order based on ChatOps's IT operation and maintenance of claim 6, wherein the task allocation between the preset work order information field and the automatic operation and maintenance operation to be triggered further comprises: based on the preliminary field filling mapping relation, the unique key entity information set and the field priority ordering result, extracting entity text content according to priority through a field value filling function with unique verification, completing field filling and de-duplication, and generating a second structured data set; based on the intention classification label, the unique key entity information set and the entity attribute information in the first structured data set, triggering a rule engine to load a hierarchical conditional action rule set through the automatic operation of multi-level condition matching, and matching to generate an automatic operation instruction set with an execution priority mark; Based on the execution priority, the logic dependency relationship and the execution premise of the automatic operation instruction set, constructing and optimizing a task execution flow chart through a directed acyclic graph construction algorithm with dependency conflict detection, and acquiring initial automatic arrangement logic; Based on the initial automation arrangement logic, the current load of the system, the resource occupancy rate and the historical execution time consumption data, the instruction execution sequence is adjusted through a multi-objective optimized resource scheduling optimization function, system resources are allocated, and a third control instruction representing the automatic operation instruction set execution logic, the execution sequence and the resource allocation scheme is generated.
  8. 8. The method for real-time collaboration and automatic generation of a work order based on ChatOps's IT operation and maintenance of claim 6, wherein the performing constraint verification and completion on the second structured data set under the constraint of a preset work order integrity rule and an information accuracy verification rule includes: Based on the target work order template identifier, inquiring a preset work order integrity rule database by executing a structured inquiry language statement to obtain a word-to-be-filled definition list and a field format rule list which are associated with the target work order template identifier; Based on the necessary-filling character segment definition list and a field name set contained in the second structured data set, acquiring a missing necessary-filling character segment name by calculating a set difference set to form a missing field list; And based on the field format rule list, performing pattern matching verification on the filling value corresponding to each field in the second structured data set through a regular expression engine, and identifying the field with the format not conforming to the field format rule to form a format exception field list.
  9. 9. The method for real-time collaboration and automatic generation of a work order based on ChatOps's IT operation and maintenance of claim 8, wherein the performing constraint verification and completion on the second structured data set under the constraint of a preset work order integrity rule and an information accuracy verification rule further comprises: acquiring authority configuration attribute information corresponding to each entity identifier by calling a hypertext transfer protocol application programming interface provided by a configuration management database based on the entity identifier contained in the unique key entity information set; Acquiring accuracy quantization scores of each field filling value by calculating a Levenstein distance and converting the Levenstein distance into similarity scores based on the field filling value associated with an entity in the second structured data set and corresponding authority configuration attribute information acquired from the configuration management database; based on a preset accuracy verification threshold, identifying a field with the accuracy quantization score lower than the accuracy verification threshold by comparing the accuracy quantization score with the accuracy verification threshold, and forming an accuracy abnormal field list.
  10. 10. The method for real-time collaboration and automatic generation of a work order based on ChatOps's IT operation and maintenance of claim 9, wherein the performing constraint verification and completion on the second structured data set under the constraint of a preset work order integrity rule and an information accuracy verification rule further comprises: based on the missing field list, accessing an operation and maintenance knowledge base or a historical work order database to perform information retrieval by executing a map query language query statement or a structured query language query statement, and generating corresponding filling values and updating the second structured data set if the retrieval is successful; Creating a task to be rechecked containing abnormal details and contexts through a workflow engine based on the format abnormal field list and the accuracy abnormal field list, submitting the task to be rechecked to a manual rechecking queue, and suspending an automatic work order generating process based on a current second structured data set; and after finishing information complementation and potential manual verification correction, generating a second structured data set after verification complementation which meets all preset rules by executing final integrity verification and data merging.

Description

ChatOps-based IT operation and maintenance real-time cooperation and work order automatic generation method Technical Field The invention belongs to the technical field of automatic operation and maintenance, and particularly relates to an IT operation and maintenance real-time cooperation and worksheet automatic generation method based on ChatOps. Background In the IT operation and maintenance cooperation process based on ChatOps, a robot (Bot) is generally adopted as an interaction center, and a back-end automation script or an API interface is called to execute corresponding operation and maintenance operation according to an instruction input by an operation and maintenance person in a chat platform or discussion of monitoring alarm, and an execution result is fed back to the chat channel in real time, so that transparent and traceable collaborative operation is realized. For transactions needing to be tracked, such as fault processing, service request and the like, the prior art usually needs to manually create and fill in a work order in an external work order system (ITSM) according to chat records to drive subsequent flow processing and knowledge precipitation, and for instant dialogs which are generated under complex fault scenes and contain multiple persons participating in, multiple rounds of interleaving and a large amount of unstructured texts, key information (such as accurate fault service identification, resource positioning, influence description and the like) needed for generating a standard chemical order is often scattered and expressed differently. In the prior art, automatic or semi-automatic generation of a work order usually depends on simple keyword matching or manual triggering of a fixed robot command template in a conversation to extract limited structured fields, and the method is applicable in the scene of sparse information and standard format, but has obvious limitation on the information extraction capability when processing the complex natural language conversation, a chat robot does not explicitly distinguish effective fact statement, irrelevant discussion, historical background description and fuzzy reference in the conversation, and lacks deep semantic understanding and information fusion capability on multi-round conversation context, so that when the real collaborative conversation with high information density and diversified expression is faced, the extraction method based on simple rules is extremely easy to generate information omission, recognition error or association misalignment. In summary, in the real-time collaboration of IT operation and maintenance based on ChatOps, in the face of a real-time dialogue with highly unstructured information and strong context dependence, the existing information extraction method based on keyword matching or fixed templates can not effectively solve the problem of fusion of index resolution (such as "IT", "that service" specifically refers to) and scattered information of cross-messages in a multi-round dialogue scene, so that identification errors, information splitting or omission occur when key entities (such as fault objects and influence ranges) for generating work orders are extracted from dialogue histories, and thus, the automation engineering order information is inaccurate and incomplete. Therefore, how to accurately realize reference resolution and cross-message key information fusion in a multi-round dialogue so as to generate high-fidelity structural chemical engineering single data is a specific technical problem to be solved. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a method for automatically generating IT operation and maintenance real-time collaboration and worksheet based on ChatOps, which comprises the steps of continuously monitoring and collecting instant dialogue content in a chat platform, further carrying out deep understanding and characterization on unstructured dialogue through a configured semantic analysis model to generate a first structured data set for representing core intention and key entity information, carrying out intelligent task allocation between worksheet information fields and automatic operation and maintenance operation according to the set to respectively obtain a second structured data set for filling worksheet templates and an automatic operation instruction set, checking and completing the second structured data set under the constraint of preset rules, combining the first structured data set and the automatic operation instruction set to generate a third control instruction for arranging execution logic, outputting the worksheet data after checking and completion to a worksheet system to automatically create a worksheet, and outputting the third control instruction to an automatic engine to drive the operation and maintenance operation to execute, thereby realizing closed loop from real-time dialogue to structured recording and aut