CN-122021608-A - User input analysis method and device based on device design intelligent agent

CN122021608ACN 122021608 ACN122021608 ACN 122021608ACN-122021608-A

Abstract

The embodiment of the application provides a user input analysis method and device based on a device design intelligent agent, and relates to the technical field of device design. The method comprises the steps of responding to a natural language instruction input by a user, identifying device design operation intention represented by the natural language instruction, determining at least one parameter item to be analyzed which is associated with the operation intention based on the identified operation intention to form a parameter extraction framework, analyzing the semantics of the natural language instruction to extract or infer values corresponding to various parameters in the parameter extraction framework, and organizing the parameter item and the corresponding values into structured data which accords with the interface specification of target device design software. The implementation mode provided by the application not only can adapt to diversified expression modes, but also can dynamically adjust the analysis logic in combination with the context, thereby remarkably improving the accuracy and adaptability of the interaction between the intelligent agent and the user.

Inventors

LIANG YINGZONG
LIU FANG
CHEN YANNING
ZHAO DONGYAN
XIA GUOFANG
Wang siyun
QIAN ZHENHUA
ZHU YAXING
LI CHENFENG

Assignees

北京智芯微电子科技有限公司
国网江苏省电力有限公司营销服务中心
国网江苏省电力有限公司
国家电网有限公司

Dates

Publication Date: 20260512
Application Date: 20251226

Claims (13)

1. A user input analysis method applied to a device design intelligent agent is characterized by comprising the following steps: responding to a natural language instruction input by a user, and identifying the device design operation intention represented by the natural language instruction; determining at least one parameter item to be parsed associated with completing the operation intention based on the identified operation intention, forming a parameter extraction framework; analyzing the semantics of the natural language instruction to extract or infer the values corresponding to each parameter in the parameter extraction frame; the parameter items and their corresponding values are organized into structured data conforming to the target device design software interface specification.
2. The method of claim 1, wherein in response to a natural language instruction entered by a user, identifying a device design operational intent characterized by the natural language instruction comprises: identifying the natural language instruction by adopting a large language model to obtain an identification result; matching the recognition result with a preset operation intention set to obtain the confidence coefficient of the recognition result for each operation intention category in the preset operation intention set, and forming a confidence coefficient set; and obtaining the operation intention according to the confidence maximum value in the confidence coefficient set.
3. The method of claim 2, wherein the set of preset operational intents is set according to a common instruction type in a semiconductor design field.
4. The method according to claim 2, characterized in that the definition of the operational intent class and the parameter items in the parameter extraction framework is predefined based on knowledge of the semiconductor device design area.
5. The method of claim 1, wherein determining at least one parameter item to be parsed associated with completing the operational intent based on the identified operational intent forms a parameter extraction framework, comprising: determining a corresponding parameter extraction frame from a frame template set according to the operation intention, wherein the parameter extraction frame in the frame template set has a corresponding relation with the operation intention; and taking each parameter in the determined parameter extraction frame as a parameter set needing to be assigned.
6. The method of claim 1, wherein parsing the semantics of the natural language instruction to extract or infer values corresponding to parameters in the parameter extraction framework comprises: mapping the nonstandard expression of the natural language instruction about the device parameters into standardized parameter names and units; Analyzing the semantic information in the natural language instruction, and assigning the parameter values in the semantic information to the corresponding parameters.
7. The method of claim 1, wherein the parameters in the parameter extraction framework include necessary parameters and optional parameters; The value of the necessary parameter is determined according to the following sequence, namely, the necessary parameter is extracted from a natural language instruction, is obtained by reasoning from semantic information of the natural language instruction, is obtained from dialogue history of the natural language instruction, and is output to be independently prompted for the necessary parameter with missing value, and corresponding input is obtained; the values of the optional parameters are determined according to the order in which default values are extracted from natural language instructions.
8. The method of claim 1, wherein organizing the parameter items and their corresponding values into structured data that conforms to a target device design software interface specification comprises: Determining a structure of the structured data according to an interface specification with the semiconductor design software; the parameter items and their corresponding values are mapped to the structured data in key-value pairs.
9. The method of claim 1, further comprising data validating the resulting structured data conforming to the target device design software interface specification; The data verification includes verifying whether fields of the structured data are of an appropriate data type, verifying whether fields of the structured data meet numerical or character constraints, and verifying whether the structured data meet custom constraints.
10. A user input parsing apparatus for a device design agent, the apparatus comprising: an intent recognition module configured to, in response to a natural language instruction input by a user, recognize a device design operation intent characterized by the natural language instruction; A frame determination module configured to determine at least one parameter item to be parsed associated with completing the operation intent based on the identified operation intent, forming a parameter extraction frame; the semantic analysis and assignment module is configured to analyze the semantics of the natural language instruction so as to extract or infer the values corresponding to various parameters in the parameter extraction frame; A result structuring module configured to organize the parameter items and their corresponding values into structured data conforming to a target device design software interface specification.
11. An electronic device comprising at least one processor; a memory coupled to the at least one processor; Wherein the memory stores instructions executable by the at least one processor, the at least one processor implementing the device design agent-based user input parsing method of any one of claims 1 to 9 by executing the memory stored instructions.
12. A computer readable storage medium having stored thereon a computer program/instruction which, when executed by a processor, implements the device design agent based user input parsing method of any one of claims 1 to 9.
13. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the device design agent based user input parsing method of any one of claims 1 to 9.

Description

User input analysis method and device based on device design intelligent agent Technical Field The application relates to the technical field of device design, in particular to a user input analysis method based on a device design intelligent agent, a user input analysis device based on the device design intelligent agent, electronic equipment, storage media and a computer program product. Background In the actual development process of the device design intelligent agent, user interaction is usually performed in the form of natural language dialogue, and the interaction mode reduces the use threshold, but also brings the problems of input information ambiguity, irregular parameter expression and the like. Because the design of the semiconductor device involves multiple types of professional technical parameters such as process parameters, material characteristics, performance indexes and the like, users may describe the same concept in various expression modes, which requires that the intelligent agent has precise semantic analysis and parameter standardization capability. Specifically, user input analysis requires three key tasks, namely, accurately identifying design targets and constraint conditions in user demands, converting unstructured natural language expressions into structured design parameters, and solving the problems of relevance and conflict among the parameters. In conventional approaches, parsing of user input typically relies on rule-based approaches such as regular expressions (Regex) or keyword matching techniques. These methods extract key parameters from the user's natural language input through predefined patterns or fixed templates. The method has the remarkable limitations that firstly, the generalization capability of matching the regular expression and the key word is weak and is difficult to adapt to the diversified expression of the natural language, and secondly, the traditional method lacks context understanding capability and cannot process parameters of implicit semantics or dependence on dialogue history. More importantly, the design of the semiconductor device involves a large number of terms of art and parameters (such as the association of doping concentration and threshold voltage), and the conventional method requires manual maintenance of a huge rule base, so that the development cost is high, all possible user expressions are difficult to cover, finally, the interaction flexibility of the intelligent agent is poor, the fault tolerance is low, and the application effect of the intelligent agent in a complex design scene is severely limited. Further, the drawbacks of the conventional approach are particularly pronounced in advanced design scenarios involving complex constraint relationships between multiple rounds of interactions and parameters. For example, when a user proposes a compound instruction of "reducing the gate oxide thickness by 20% under the premise of ensuring that the breakdown voltage is not lower than X volts", a rule-based parser not only has difficulty in accurately separating an "optimization target" (thinning gate oxide) and a "constraint condition" (maintaining the breakdown voltage), but also cannot understand the inherent physical coupling relationship between the two parameters, so that the parsed design parameter sets may be in contradiction, and cannot be directly used for driving simulation or design tools. Disclosure of Invention The embodiment of the application aims to provide a user input analysis method and device based on a device design intelligent agent, which are used for constructing a prompt word template and a parameter extraction rule by combining the contextual understanding advantages of a large language model, so as to realize intelligent conversion from free expression of a user to standardized design parameters and at least solve part of problems in the background technology. In order to achieve the purpose, the application provides a user input analysis method based on a device design intelligent agent, which comprises the steps of responding to a natural language instruction input by a user, identifying a device design operation intention represented by the natural language instruction, determining at least one parameter item to be analyzed associated with the operation intention based on the identified operation intention to form a parameter extraction frame, analyzing the semantics of the natural language instruction to extract or infer values corresponding to various parameters in the parameter extraction frame, and organizing the parameter item and the corresponding values into structural data conforming to the interface specification of target device design software. Optionally, in response to a natural language instruction input by a user, identifying the device design operation intention represented by the natural language instruction, wherein the device design operation intention comprises the steps of identifyin