Element attribute dynamic binding method for three-dimensional design software

Abstract

The invention discloses a three-dimensional design software-oriented component attribute dynamic binding method. The method comprises the following steps of: S1, receiving component classification information input by a user; s2, based on a pre-trained language model, converting the element classification name into vector representation; s3, calculating a similarity score between the element classification vector and a plurality of attribute table vectors; s4, recommending the most matched attribute table according to the similarity score; s5, displaying the recommended attribute table for the user to confirm or modify, and obtaining a final attribute table confirmed by the user; and S6, based on the attribute table confirmed by the user, updating the attribute relation graph maintained by the system to record the association strength of the classification and the attribute table. Compared with the prior art, the association process of the element classification and the attribute table is automated through the intelligent algorithm, and the problems of tedious operation, low efficiency, high error rate and poor user experience in the prior art are effectively solved.

Description

TECHNICAL FIELD

[0001] The application relates to a component attribute dynamic binding method for a three-dimensional design software, and belongs to the technical field of computer-aided design (CAD) and building information modeling (BIM) data management. BACKGROUND

[0002] In the field of computer-aided design (CAD) and building information modeling (BIM) data management, component coding management is a key technology involving the association operation between component classification and attribute table to ensure data consistency and operation efficiency.

[0003] In the traditional engineering design and modeling process, component attribute hanging usually relies on user manual selection and association of attributes from pre-defined attribute tables. The specific method is as follows: after the user inputs the component classification information, the system displays all available attribute tables, the user needs to browse the list, remember or search for related attributes, and manually perform matching operations. The implementation of this method is based on static database query and interface display, and the attribute table is stored through a relational database, the drop-down list or check box is provided for user selection through a graphical user interface (GUI), and the basic CRUD (create, read, update, delete) operation handles user input.

[0004] However, the above-mentioned component coding management has the following problems: first, the operation is tedious and inefficient, the user must manually select the attributes step by step, which increases the operation steps and time cost, especially when the number of attribute tables is large, the search and matching process takes a long time; second, the error rate is high, since it depends on user memory and subjective judgment, it is easy to cause attribute omission or incorrect association, which affects data quality; third, the interface is complex, the existing system often displays all attribute options, which increases the user's cognitive load and reduces the user experience. SUMMARY

[0005] Based on the above, the application provides a component attribute dynamic binding method for a three-dimensional design software to solve the problems of tedious and inefficient component attribute hanging operation, high error rate in the prior art.

[0006] The technical scheme of the application is: a component attribute dynamic binding method for a three-dimensional design software, comprising:

[0007] Step S1, receiving user input component classification information;

[0008] Step S2, converting the component classification name into a vector representation based on a pre-trained language model;

[0009] Step S3, calculating the similarity score between the component classification vector and the plurality of attribute table vectors;

[0010] Step S4, recommend the most matching attribute table according to the similarity score;

[0011] Step S5, show the recommended attribute table for user confirmation or modification, and obtain the final attribute table confirmed by the user;

[0012] Step S6, update the attribute relationship graph maintained by the system to record the association strength between the classification and the attribute table based on the attribute table confirmed by the user.

[0013] In one of the examples, it also includes:

[0014] Step S7, load the user historical behavior data based on the user identification, and retrieve the user operation log from the database;

[0015] Step S8, analyze the operation mode using a prediction model based on the user historical operation sequence, and predict the attribute of the next possible operation of the user;

[0016] Step S9, adjust the graphical interface layout, and display the predicted attribute preferentially and hide irrelevant attributes through priority sorting.

[0017] In one of the examples, in step S2, the pre-trained language model uses a distributed semantic representation method to map the element classification name to a high-dimensional vector space based on the word vector representation technology, wherein the vector representation is generated by training a large-scale text corpus to capture the semantic features of the name.

[0018] In one of the examples, in step S3, the calculation of the similarity score is based on the vector angle correlation measure, specifically using the cosine similarity algorithm to calculate the cosine value of the angle between the vectors, and outputting the score list for sorting the attribute tables by score.

[0019] In one of the examples, in step S4, recommending the most matching attribute table includes selecting the attribute table with the highest similarity score, and outputting in the form of key-value pairs, where the key is the attribute table identifier and the value is the similarity score.

[0020] In one of the examples, in step S8, the method for predicting the attribute of the next possible operation of the user is to build a Markov chain model, calculate the transition probability between attributes, and generate an attribute list sorted by probability. Through the probability model, the prediction accuracy is improved, and the user interface is more in line with actual needs.

[0021] In one of the examples, in step S9, adjusting the graphical interface layout includes dynamically rendering interface components through the front-end framework, and gradually revealing advanced attributes according to user operation behavior.

[0022] The beneficial effects of the present application: by adopting a pre-trained language model and similarity calculation to realize intelligent attribute recommendation, the present application automates the element attribute binding process, significantly reduces the user manual operation steps, and improves the coding efficiency; and, through the user confirmation mechanism, the accuracy of the binding is ensured, and the error rate and data inconsistency are reduced; at the same time, through the dynamic updating of the attribute relationship graph, the system self-learning optimization is realized, and the response speed and adaptability are improved. Compared with the prior art, the present application automatically classifies elements and associates the attribute table through intelligent algorithms, effectively solving the problems of complicated operation, low efficiency, high error rate and poor user experience in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 A flowchart of a three-dimensional design software-oriented element attribute dynamic binding method. DETAILED DESCRIPTION

[0024] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings. In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the scope of the present application, so the present application is not limited to the specific implementation disclosed below.

[0025] Please refer to Figure 1 The embodiment of the present application is a three-dimensional design software-oriented element attribute dynamic binding method, which comprises the following steps:

[0026] Step S1, receiving element classification information input by the user

[0027] The system receives the element classification information input by the user through a graphical user interface (GUI). Specifically, the user inputs the element classification name, such as "electrolytic cell" or "industrial boiler", through an input box or a drop-down menu. The system uses an event listener to capture user input events and passes the input data to the backend processing module through an application programming interface (API). The input data is in the form of a text string, and the system performs basic verification, such as checking for non-empty and correct format, to ensure data integrity.

[0028] Step S2, converting the element classification name into a vector representation based on a pre-trained language model

[0029] Specifically, the pre-trained language model is based on the word vector representation technology, and specifically adopts the distributed semantic representation method such as GloVe, to map the element classification name to a high-dimensional vector space, wherein the vector representation is generated by training a large-scale text corpus to capture the semantic features of the name.

[0030] In one example, the model is trained to generate by large-scale text corpus (Wikipedia and professional engineering documents) to capture the semantic features of the name. For example, for the input "electrolytic tank", the model outputs a 300-dimensional floating-point number vector representing its position in the vector space. The model training process is as follows: data collection downloads text data from public corpus, pre-processing (such as word segmentation, stop word removal); feature extraction using TF-IDF; training using neural network architecture (Skip-gram), parameter settings: vector dimension 300, window size 5, learning rate 0.025. After training, the model performance is evaluated using the similar word task.

[0031] Step S3, calculating the similarity score between the component classification vector and the plurality of attribute table vectors

[0032] Specifically, the attribute table vectors are pre-stored in the database, and each attribute table corresponds to a vector representation. The calculation of the similarity score is based on the vector angle correlation measure, and the cosine similarity algorithm is used to calculate the cosine value of the angle between the vectors, and the score list is output for sorting the attribute tables according to the score. For example, the score list is sorted in descending order, which facilitates subsequent recommendation.

[0033] The cosine similarity formula is:

[0034]

[0035] where, is the component classification vector, is the attribute table vector, denotes the dot product, denotes the modulus of the vector. The output is a similarity score list, and the score range is between [-1, 1], and the higher the value, the more similar.

[0036] When calculating, the system queries all attribute table vectors from the database, iteratively calculates the cosine similarity of each with the input vector, and stores the score list.

[0037] Step S4, recommending the most matched attribute table according to the similarity score

[0038] Specifically, recommending the most matched attribute table includes selecting the attribute table with the highest similarity score, and outputting in the form of key-value pair, where the key is the attribute table identifier and the value is the similarity score. The system selects the maximum value from the score list, and the corresponding attribute table identifier is obtained from the database mapping. The output data format is JSON, which is convenient for subsequent system processing, such as transmission to the user interface or storage to the log.

[0039] Step S5, showing the recommended attribute table for user confirmation or modification, and obtaining the final attribute table confirmed by the user

[0040] Specifically, the detailed information of the recommended attribute table is displayed through the user interface (web page), including attribute name and value, and confirmation and modification options (such as buttons or forms) are provided. User operations are captured through event handling, and the system obtains the final selection.

[0041] Step S6, based on the user-confirmed attribute table, update the attribute relationship graph maintained by the system to record the association strength between the category and the attribute table

[0042] The attribute relationship graph is a graph data structure, where nodes represent element categories and attribute tables, and edges represent association strength, with strength values as numerical values. The update includes incremental updating of association strength values, adjusting strength values through weighted averaging or exponential decay methods to reflect the weight and historical impact of user confirmation operations. For example, use the formula:

[0043]

[0044] where, is the decay factor (such as 0.9), and confirmation_weight is the weight of user confirmation (such as 1.0). The system queries the existing relationship graph, calculates the new strength, and updates the database.

[0045] Step S7, based on user identification, load user historical behavior data, retrieve user operation logs from the database

[0046] Query user operation logs from the database, data format is timestamp and operation details. Operation details field includes operation type (such as "attribute selection") and attribute ID.

[0047] Step S8, based on user historical operation sequence, use prediction model to analyze operation mode, predict the attribute that the user may operate next

[0048] Specifically, the method for predicting the attribute that the user may operate next is: build a Markov chain model, calculate the transition probability between attributes, and generate a probability-sorted attribute list.

[0049] First, extract attribute operation sequences from user operation logs, each sequence consists of a series of attribute operation states (for example, state is represented as attribute ID).

[0050] Then, for each current state, count the number of times in history that transitions from the current state to each next state , and the total number of times the current state appears . Transition probability is calculated by the formula To avoid zero probability problems, use Add-one smoothing processing, adjust the formula to where N is the total number of attribute operation states (i.e. the number of all possible next states).

[0051] Finally, for a given current state, the probabilities of all possible next states are computed, a list of probabilities is generated, and sorted in descending order of probability value.

[0052] Step S9, adjust the graphical interface layout to prioritize display of predicted attributes and hide irrelevant attributes by prioritizing.

[0053] In particular, interface components are dynamically rendered by a front-end framework such as React, with CSS and JavaScript used to control component display and hiding. In one example, predicted attributes are highlighted and irrelevant attributes are collapsed.

[0054] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be noted that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the scope of protection of the patent of the present application should be subject to the appended claims.

Claims

1. A method for dynamically binding component attributes to 3D design software, characterized in that, include: Step S1: Receive component classification information input by the user; Step S2: Based on the pre-trained language model, convert the component classification names into vector representations; Step S3: Calculate the similarity score between the component classification vector and multiple attribute table vectors; Step S4: Based on the similarity score, recommend the most matching attribute table; Step S5: Display the recommended attribute table for the user to confirm or modify, and obtain the final attribute table confirmed by the user. Step S6: Based on the attribute table confirmed by the user, update the attribute relationship graph maintained by the system to record the association strength between the category and the attribute table.

2. The method for dynamically binding component attributes according to claim 1, characterized in that, Also includes: Step S7: Load user historical behavior data based on user identifier and retrieve user operation logs from the database; Step S8: Based on the user's historical operation sequence, use a predictive model to analyze the operation pattern and predict the attributes of the user's next possible operation. Step S9: Adjust the layout of the graphical interface, prioritize the display of predicted attributes and hide irrelevant attributes by sorting them by priority.

3. The method for dynamically binding component attributes according to claim 1, characterized in that, In step S2, the pre-trained language model is based on word vector representation technology and uses a distributed semantic representation method to map the component classification names to a high-dimensional vector space, where the vector representation is generated through training on a large-scale text corpus to capture the semantic features of the names.

4. The method for dynamically binding component attributes according to claim 1, characterized in that, In step S3, the similarity score is calculated based on the vector angle correlation metric. Specifically, the cosine similarity algorithm is used to calculate the cosine value of the angle between vectors, and the output score list is used to sort the attribute table according to the score.

5. The method for dynamically binding component attributes according to claim 1, characterized in that, In step S4, the recommended most matching attribute table includes selecting the attribute table with the highest similarity score and outputting it in key-value pair form, where the key is the attribute table identifier and the value is the similarity score.

6. The method for dynamically binding component attributes according to claim 2, characterized in that, In step S8, the method for predicting the attributes that the user may perform the next action is as follows: construct a Markov chain model, calculate the transition probabilities between attributes, and generate a list of attributes sorted by probability.

7. The method for dynamically binding component attributes according to claim 2, characterized in that, In step S9, adjusting the graphical interface layout includes dynamically rendering interface components through the front-end framework and gradually revealing advanced attributes based on user actions.

Images