A context-based digital flowchart aided drawing method and device

Abstract

The application discloses a context-based digital flowchart auxiliary drawing method and device, wherein the method comprises the following steps: step S1, obtaining context information of a current drawing action of a digital flowchart; step S2, counting the collocation frequency between legends in all historical drawings to obtain a co-occurrence relationship matrix; step S3, calculating the co-occurrence frequency score of each legend and the context information based on the context information and the co-occurrence relationship matrix, and dynamically generating a legend recommendation list based on the co-occurrence frequency score. The application can fully perceive the drawing action of a user and dynamically recommend legends based on the drawing action of the user.

Description

Technical Field

[0001] This invention relates to the field of digital flowchart-assisted drawing technology, and in particular to a context-based digital flowchart-assisted drawing method and apparatus. Background Technology

[0002] Flowcharts, a common tool in nuclear power plant operation and production, are an important part of nuclear power plant digitization. Currently, in the process of creating digital flowcharts, users need to manually search for and drag equipment symbols from a vast symbol library, which is cumbersome and inefficient, especially for new users or those unfamiliar with the symbol library, where searching for symbols is time-consuming. At the same time, the drawing process lacks intelligent guidance, requiring users to frequently switch tools and adjust attributes, affecting the smoothness of the drawing process; complex equipment combinations, such as pump groups and valve groups, need to be drawn repeatedly, lacking a reuse mechanism. Summary of the Invention

[0003] In view of this, it is necessary to provide a context-based digital flowchart-assisted drawing method and apparatus to effectively solve the technical problem of low legend input efficiency when drawing digital flowcharts.

[0004] This invention provides a context-based digital flowchart-assisted drawing method, comprising the following steps: Step S1: Obtain the context information of the current drawing action of the digital flowchart; Step S2: Count the frequency of use of legends in all historical drawings to obtain the co-occurrence matrix; Step S3: Calculate the co-occurrence frequency score of each legend with the context information based on the context information and the co-occurrence relationship matrix, and dynamically generate a legend recommendation list based on the co-occurrence frequency score.

[0005] Preferably, step S1 specifically comprises: The user's most recent legend placement action is obtained as the current drawing action. Multiple legend placement actions prior to the current drawing action are obtained as the context information. The context information includes the legend type identifier, spatial location coordinates, and attribute parameter configuration corresponding to the current drawing action, as well as the legend type, position, attributes, and pipeline connection topology relationships between the multiple legend placement actions.

[0006] Preferably, step S1 further includes: The recording of legend placement actions adopts an event-driven mechanism. The system triggers event recording every time a legend is added or deleted, and appends the operation events to the context sequence in the order of the operation occurrence. The length of the context extraction window is configurable, and the legend placement actions with the most recent preset number of times are retained as valid context.

[0007] Preferably, step S2 specifically comprises: The number of times multiple legends are used in the same order in historical drawing is counted. Legends with a higher number of times they are used together are grouped into legend groups. The co-occurrence frequency of each legend group is calculated based on the number of times they are used together. The co-occurrence relationship matrix is ​​obtained by combining the co-occurrence frequencies of different legend groups. Each element of the co-occurrence relationship matrix represents the association strength of the two legends used together in all historical drawing records.

[0008] Preferably, step S2 further includes: During the construction of the co-occurrence relationship matrix, global usage frequency statistics and individual usage frequency statistics are maintained simultaneously. The global usage frequency statistics record the cumulative number of times each legend is used in the historical drawings of all users, and the individual usage frequency statistics record the cumulative number of times each legend is used in the historical drawings of the current user. The global usage frequency is used to calculate the global popularity score, and the individual usage frequency is used to calculate the personalized preference score.

[0009] Preferably, step S3 specifically comprises: The context information is paired with each legend to construct different legend groups. The co-occurrence frequency of each legend group is obtained based on the co-occurrence relationship matrix. The co-occurrence frequency score is calculated based on the co-occurrence frequency. A legend recommendation list is generated for the user based on the co-occurrence frequency score. If a user places a new legend after the current legend recommendation list is generated, the system retrieves new context information based on the new legend placement action and generates a new legend recommendation list, thus achieving dynamic updating of the legend recommendation list.

[0010] Preferably, step S3 further includes: The system calculates the individual usage frequency of each legend and its score based on that frequency. It also calculates the global usage frequency of each legend and its global popularity score. The system then performs a weighted average of the co-occurrence frequency score, the individual usage frequency score, and the global popularity score to obtain a total recommendation score. Finally, it dynamically generates a legend recommendation list based on the total recommendation score.

[0011] Preferred options also include: Step S4: Record the user's selection behavior for each recommendation; if the user accepts the recommendation, record the correlation strength between the corresponding recommendation legend and the context; if the user ignores the recommendation and searches manually, record the keywords the user manually searches for, and update the legend's alias library and semantic vector based on the keywords; if the user accepts the recommendation but adjusts the legend's position or attributes, record the adjustments to optimize the corresponding layout algorithm.

[0012] Preferred options also include: Intelligent recommendations are made based on the user's real-time actions while drawing. Specifically, the intelligent recommendations include recommending legends related to the legends where the mouse position is located, based on the user's mouse position information. A typical configuration library is constructed, which stores predefined legend combination templates. The legend combination templates include legend combinations and their connection rules. When a user places a legend, the complete combination is automatically recommended based on the co-occurrence relationship. After the user confirms the recommended content, all devices in the corresponding combination are automatically generated and automatically connected according to preset rules. After the user selects and places a recommended legend, the system checks the required attributes of the selected legend, provides prompts for missing or incorrect attributes, and automatically fills in the relevant or recommended attribute values ​​based on the selected legend.

[0013] The present invention also provides a context-based digital flowchart-assisted drawing device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the context-based digital flowchart-assisted drawing method.

[0014] Compared with existing technologies, the advantages of this invention are as follows: This invention calculates a contextual association score, i.e., a co-occurrence score, based on the current drawing context information, and dynamically generates an intelligent recommendation list. This invention fully perceives the user's historical drawing actions; after the user places the first part of the legend, it identifies and recommends the next most likely legend, reducing legend search and placement time and improving drawing efficiency. Attached Figure Description

[0015] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart of an embodiment of a context-based digital flowchart-assisted drawing method provided by the present invention. Detailed Implementation

[0016] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0017] Example 1

[0018] Please see Figure 1 The context-based digital flowchart-assisted drawing method in this embodiment specifically includes the following steps: Step S1: Obtain the context information of the current drawing action of the digital flowchart; Step S2: Count the frequency of use of legends in all historical drawings to obtain the co-occurrence matrix; Step S3: Calculate the co-occurrence frequency score of each legend with the context information based on the context information and the co-occurrence relationship matrix, and dynamically generate a legend recommendation list based on the co-occurrence frequency score.

[0019] The context-based digital flowchart-assisted drawing method provided in this embodiment, hereinafter referred to as "this method", achieves an intelligent drawing experience and significantly improves drawing efficiency and standardization through technologies such as context awareness, combination recommendation, semantic understanding and intelligent completion.

[0020] The following provides a detailed explanation of each step.

[0021] Step S1: Obtain the current drawing action and historical drawing action sequence of the digital flowchart. For example, if the user places multiple legends consecutively, the legend sequence is recorded. The context information specifically includes the type, two-dimensional coordinate position, size scale, rotation angle, layer affiliation, and pipeline connection topology relationship between legends added by the user in the current drawing session.

[0022] Specifically, the process of obtaining context information is as follows: After the user starts a drawing session, the system starts an event listener to monitor legend operation events in the drawing area in real time, including the addition and deletion of legends; for each operation event, the system records the timestamp of the event, the operation type, and the unique identifier of the legend involved; the above operation events are appended to the historical drawing action sequence of the current session in chronological order to form a complete context event chain; when the user performs a new legend placement, the system extracts the most recently placed legend from the event chain as the current drawing action, and uses a continuous operation record before the current drawing action as context information. The window length of the context information can be set according to the system configuration, for example, retaining 3 to 10 most recently placed legends as valid context.

[0023] Furthermore, the structured storage of context information adopts the following approach: a context record object is created for each legend, which contains attribute fields such as legend type identifier, placement coordinates, size parameters, rotation angle, layer to which it belongs, and topological connection relationship with other legends; context record objects are linked by timestamps and topological relationships to form a context knowledge graph that can be quickly retrieved; the system also maintains a session-level context state cache, which updates the latest state of all placed legends in the current drawing area in real time, allowing symbiotic relationship analysis and recommendation engines to read efficiently, avoiding repeated traversal of the complete action sequence for each recommendation.

[0024] Specifically, considering the unique characteristics of digital flow diagrams for nuclear power plants, the context information also includes the following nuclear power-specific context dimensions: equipment system codes, i.e., determining the process system to which the current drawing belongs based on the functional location code and system code of the nuclear power equipment, such as RCV, RCP, RIS, and adjusting the weight of legends of the same system in subsequent recommendations accordingly; media type and flow direction information, i.e., filtering legends incompatible with the current media type based on the type of media handled by the placed legends, such as water, steam, oil, and compressed air, and the flow direction of the pipeline, such as upstream to downstream, and recommending only legend types that match the media; pressure and temperature levels, i.e., selecting equipment legends that meet the requirements of the current pipeline section's pressure and temperature levels, such as Class 150, Class 300, Class 1500, and avoiding recommending valves or instruments of incompatible levels.

[0025] Step S2: Co-occurrence Matrix. This step analyzes the frequency of legend pairings in historical drawing records. For example, "pump" often appears alongside "inlet valve," "outlet valve," and "pressure gauge." Specifically, this step analyzes the legend addition sequences in all drawing records in the historical drawing database, statistically analyzing the pairings of each legend. The resulting co-occurrence matrix reflects the frequency of pairings between legends. Each element of the matrix represents the strength of the association between the corresponding row legend and column legend in all historical drawing records; a larger value indicates a more frequent pairing of the two legends.

[0026] Specifically, the co-occurrence matrix is ​​constructed as follows: First, the system loads all legend addition sequence records of completed drawing sessions from the historical drawing database. Each sequence record corresponds to the addition order of legends in a completed flowchart. Then, a sliding window algorithm is used to traverse each sequence record. The window size is set to a preset value, such as 5. A pairing relationship is recorded between any two legends within the window. The order in which legends appear within the window is also considered, distinguishing between directional pairing and bidirectional pairing. After the statistics are completed, the original co-occurrence frequency matrix is ​​obtained. The rows and columns of the matrix are legend identifiers, and the matrix element values ​​represent the cumulative number of times the corresponding two legends co-occur in all historical drawing records.

[0027] Furthermore, the original co-occurrence frequency matrix is ​​normalized to eliminate the bias effect of high-frequency legends: for any element in the matrix The formula for calculating its normalized value is: ,in and Each represents a legend and legend The total number of independent occurrences in all historical records; the normalized matrix elements range from [0,1], with larger values ​​indicating a closer association between the two legends; in addition, Jaccard similarity coefficient or point mutual information (PMI) can be used as alternative normalization methods to adapt to the recommendation accuracy requirements of different scenarios.

[0028] Furthermore, during the construction of the co-occurrence relation matrix, the system simultaneously maintains two independent usage frequency counters: a global frequency counter records the cumulative number of times each legend has been used in all users' historical drawing sessions, which is used to calculate the popularity score of the legend across all users; and a personal frequency counter records the cumulative number of times each legend has been used in the current user's personal historical drawing sessions, which is used to calculate the personalized preference score of the legend for the current user. Both types of frequency counters are updated in batches after each drawing session, using an incremental update strategy to avoid the performance overhead caused by full recalculation.

[0029] Step S3: Recommendations are made based on the historical drawing action sequence. After the user places multiple legends in succession, the system identifies the pattern and recommends the next most likely legend. For example, after placing "pump", "inlet valve" is automatically recommended. Based on the user's historical drawing actions, such as placing "pump", several possible next legends are given based on legends that may appear in combination with pump, and co-occurrence scores are calculated.

[0030] Specifically, to calculate co-occurrence scores, a typical configuration library is set up for common equipment combinations, predefining frequently used equipment combinations. For example, a "pump group" includes a pump + inlet valve + outlet valve + pressure gauge + check valve, and a "regulating valve group" includes a regulating valve + upstream and downstream isolation valves + bypass valve. A co-occurrence relationship matrix is ​​constructed based on these common equipment combinations. When a user places the first device, the system calculates the co-occurrence score based on the co-occurrence relationship and automatically recommends the complete combination, displaying the combination layout in a semi-transparent preview. One-click combination generation: After user confirmation, the system generates all devices in the combination at once and automatically connects them according to preset rules, such as connecting valves to the pump inlet and outlet. Combination templates are saved, allowing users to customize combinations and save them as templates for later reuse.

[0031] For equipment combination scenarios unique to nuclear power plant flow diagrams, the typical configuration library also includes the following nuclear power-specific combination templates: Typical configuration for the Chemical and Volumetric Control (RCV) system: charging pump + inlet isolation valve + outlet isolation valve + outlet check valve + minimum flow pipeline and valves + pressure gauge + flow meter; Typical configuration for the Residual Heat Removal (RRA) system: residual heat removal pump + inlet isolation valve + outlet isolation valve + heat exchanger + outlet check valve + thermometer + pressure gauge; Typical configuration for the Chemical and Volumetric Control (CVC) system: volumetric control box + makeup water pump + inlet valve + outlet valve + safety valve + level gauge. The co-occurrence correlation strength between the legends in the above nuclear power-specific combination templates is higher than that of general combinations, and they have a larger weighting coefficient in the co-occurrence relationship matrix.

[0032] In addition, when performing context-aware recommendations independently, the following forms are also included: location-based recommendations, which recommend snap-to equipment types, such as valves and instruments, when the user's mouse hovers over a pipeline; system tag-based recommendations, which increase the weight of commonly used symbols of the current drawing based on the system to which it belongs, such as RCV; and current drawing type-based recommendations, which recommend process equipment for process drawings and electrical symbols for electrical drawings.

[0033] In addition to contextual information, this embodiment also incorporates semantic similarity for comprehensive recommendations. The calculation of semantic similarity is explained in detail below.

[0034] First, multimodal user intent capture is implemented. Multiple input methods are provided in the drawing interface to capture user intent in real time. Text input: Users enter device names, types, or codes in the search box, such as "electric valve" or "RCV001VP." The system supports fuzzy pinyin matching and auto-completion. Voice input: Users speak the device name through the microphone, and the system performs speech recognition and converts it to text, supporting mixed Chinese and English. Recognition results are displayed and searchable in real time. Shortcut key triggering: Shortcut points are associated with different legends. Pressing a preset shortcut key recommends related legends. Shortcut keys can also be associated with contextual recommendation functions; for example, the spacebar is associated with contextual recommendation. When a user presses the spacebar, contextual recommendations are triggered, and the system automatically pops up a contextual recommendation list based on the current mouse position. Gesture recognition: Mouse gestures are supported, such as drawing a circle to indicate "find pump." The system recognizes the gesture and executes the corresponding operation. Image input: Users can upload reference images, such as hand-drawn sketches or photos. The system triggers image recognition to match the most similar legend. Vectorization technologies, such as Word2Vec and BERT, are used to convert the text descriptions in the user input information into semantic vectors, facilitating subsequent fuzzy matching and semantic similarity calculation.

[0035] Secondly, a semantic index for legends is constructed. A multi-dimensional semantic index for legends is established to provide a foundation for intelligent recommendations. This includes: a basic naming index (including standard legend names such as "gate valve" and "centrifugal pump"), aliases (common names used in various power plants), and English abbreviations such as MOV and GATE VALVE); an equipment type coding index (corresponding to type codes in the equipment master data, such as "P-101" corresponding to the pump class); a usage scenario tag index (including the system to which it belongs, such as RCV, RCP, and RIS; the medium type, such as water, gas, oil, and steam; the pressure level, such as low pressure, medium pressure, and high pressure; and the installation method, such as horizontal and vertical); a visual feature vector (extracting image features of legends, such as shape, lines, and fill, used for sketch-based similarity matching); a co-occurrence matrix (statistically calculating the frequency of legend pairings in historical drawings, such as "pump" often appearing alongside "inlet valve," "outlet valve," and "pressure gauge"); and usage frequency statistics (recording the usage frequency of each legend globally and for the current user, used for subsequent personalized sorting). Vectorization techniques, such as Word2Vec and BERT, are used to convert the text descriptions contained in the legend attributes into vector form, resulting in a multi-dimensional attribute feature vector for each legend. By associating the different attribute indices, a legend semantic index is constructed for fuzzy matching and semantic similarity calculation.

[0036] Finally, based on the semantic vectors and the legend semantic index, the semantic similarity between the semantic vectors and the multidimensional attribute feature vectors of each legend is calculated.

[0037] In addition to semantic similarity and co-occurrence scores, this embodiment also adds individual usage frequency scores and global popularity scores. Based on user history, recommendations analyze current users' historical drawing habits, calculate the usage frequency of each legend, and determine individual usage frequency scores, prioritizing recommendations of frequently used legends. Based on the historical drawing habits of all users, the global usage frequency of each legend is calculated, and a global popularity score is determined, prioritizing recommendations of globally popular legends.

[0038] Based on semantic similarity, contextual association score, personal usage frequency score, and global popularity score, a multi-factor weighted ranking is adopted. In this embodiment, the weights of each part are allocated as follows: co-occurrence frequency score, i.e., contextual association score 40%, semantic similarity score 30%, personal usage frequency score 20%, and global popularity score 10%.

[0039] Recommended results are displayed as a list of thumbnails near the mouse cursor, in a floating window, and can be quickly selected using keyboard shortcuts such as the number keys 1-9.

[0040] Furthermore, the recommended results support real-time preview and quick placement. Thumbnail preview: When the mouse hovers over a recommended legend, its size and appearance are displayed as a semi-transparent shadow on the drawing, making it easy for users to determine its position. Smart snap-in preview: When a legend is close to a pipeline, the snap-in alignment effect is automatically displayed, showing connection point alignment and pipeline interruption previews. One-click placement: After the user clicks on a recommended legend or presses a shortcut key, the legend automatically snaps to the mouse cursor; a second click completes the placement, eliminating the need to return to the legend library for dragging. Continuous placement mode: After selecting a legend, the user can click multiple times to place multiple identical legends, suitable for batch drawing of similar equipment.

[0041] Furthermore, step S3 also includes automatically filling in or recommending attribute values ​​while placing the legend. Specifically, this includes the following methods: automatic filling based on equipment codes: if the user enters an equipment code, such as "RCV001VP", the system automatically matches the equipment master data and fills in attributes such as equipment name, model, and pressure rating; context-based recommendation: based on the system and location of the equipment, default attribute values ​​are recommended, such as the default pressure rating of RCV system valves being Class 1500; attribute validation: after the legend is selected and placed, it automatically checks whether required attributes are filled in and whether the format conforms to specifications, such as using regular expressions for equipment codes, highlighting missing or incorrect attributes; and batch attribute filling: after selecting multiple similar devices, the same attributes can be modified in batches.

[0042] Furthermore, step S3 also includes providing intelligent completion and optimization suggestions during the drawing process. Specifically, this includes: connection completion, automatically displaying suggested connection lines as dashed lines when two device connection points are detected to be very close but not connected, which are generated after user confirmation; branch completion, automatically suggesting branch pipeline directions after placing a tee on the main pipeline; equidistant arrangement suggestions, automatically prompting "equidistant arrangement" when the system detects uneven spacing when the user places a row of devices, allowing for one-click adjustment; and alignment suggestions, displaying alignment reference lines when misaligned devices are detected, allowing the user to align them with one click.

[0043] Step S4, User Feedback and Model Self-Learning: Record all user choices for continuous optimization of the recommendation model. Positive Feedback: When a user accepts a recommendation (e.g., clicks or places a recommendation legend), record the strength of the recommendation's relevance to the context. Negative Feedback: When a user ignores a recommendation and manually searches, record the keywords used to update the legend's alias library and semantic vector. Correction Feedback: When a user adjusts the position or attributes of a recommendation legend, record the adjustments for optimizing the layout algorithm. Recording each recommendation feedback and periodically retraining the recommendation model with new data continuously improves recommendation accuracy. Furthermore, personalized model export is supported, allowing users to synchronize their usage habits across different devices.

[0044] This method offers high drawing efficiency and speed, making it particularly user-friendly for beginners. New users can customize the tutorial and tutorial modes as needed. For new users, interactive tutorials are provided, including: step-by-step prompts (pop-up bubbles show how to use shortcuts and preview functions when using the intelligent recommendations for the first time); drawing suggestions (proactively suggesting possible next steps based on the current drawing progress, such as "You have placed the pump, do you want to add the inlet valve?"); and common error alerts (timely reminders and corrective suggestions when potential user errors are detected, such as equipment not being connected).

[0045] Compared with existing technologies, this method has the following advantages: significantly improved drawing efficiency, with intelligent recommendation and combination generation functions greatly reducing the time for legend searching and placement and the entire drawing process; reduced repetitive work, with combination recommendations and template reuse avoiding the repeated drawing of the same structure and improving work satisfaction; improved standardization, with intelligent attribute filling and verification ensuring complete and accurate device attributes and reducing later error correction costs; personalized and adaptive, with the recommendation model continuously optimized according to usage habits, becoming more user-friendly the more it is used, achieving a personalized drawing experience for each user; and smooth real-time interaction, with shortcut keys and preview functions reducing mouse movement and clicks, making long-term drawing less tiring.

[0046] The following example, using a drafter drawing an RCV system diagram, will be used to illustrate this method in more detail through several specific scenarios.

[0047] Scenario 1: Contextual Recommendation After the drafter placed a centrifugal pump P001, the system detected the context and automatically recommended a "pump set" in the floating window: including an inlet valve, an outlet valve, and a pressure gauge.

[0048] The drafter accepts the recommendation and clicks "Pump Set". The system generates the three devices at once and automatically places them in the appropriate positions upstream and downstream of the pump according to preset rules. The inlet valve is on the left, the outlet valve is on the right, and the pressure gauge is on the outlet pipeline. At the same time, the pipeline is automatically connected.

[0049] The draftsman only needs to fine-tune the position of the pressure gauge, saving a lot of time.

[0050] Scenario 2: Intelligent Attribute Filling When the drafter placed the valve, he entered the equipment code "RCV001VP". The system automatically matched it from the equipment master data and filled in the equipment name "upfill pump inlet isolation valve", pressure rating "Class 1500" and material "stainless steel".

[0051] If the drafter forgets to fill in a certain attribute, the system will highlight "Nominal diameter missing" after placement and recommend the commonly used value "DN50", which the drafter can confirm with one click.

[0052] Example 2

[0053] This embodiment provides a context-based digital flowchart auxiliary drawing device, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, it implements the context-based digital flowchart auxiliary drawing method described in Embodiment 1.

[0054] The context-based digital flowchart auxiliary drawing device provided in this embodiment is used to implement the context-based digital flowchart auxiliary drawing method. Therefore, the context-based digital flowchart auxiliary drawing device also has the same technical effects as the context-based digital flowchart auxiliary drawing method, and will not be described again here.

[0055] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of the present invention.

Claims

1. A context-based digital flowchart-assisted drawing method, characterized in that, Includes the following steps: Step S1: Obtain the context information of the current drawing action of the digital flowchart; Step S2: Count the frequency of use of legends in all historical drawings to obtain the co-occurrence matrix; Step S3: Calculate the co-occurrence frequency score of each legend with the context information based on the context information and the co-occurrence relationship matrix, and dynamically generate a legend recommendation list based on the co-occurrence frequency score.

2. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Step S1 specifically involves: The user's most recent legend placement action is obtained as the current drawing action. Multiple legend placement actions prior to the current drawing action are obtained as the context information. The context information includes the legend type identifier, spatial location coordinates, and attribute parameter configuration corresponding to the current drawing action, as well as the legend type, position, attributes, and pipeline connection topology relationships between the multiple legend placement actions.

3. The context-based digital flowchart-assisted drawing method according to claim 2, characterized in that, Step S1 further includes: The recording of legend placement actions adopts an event-driven mechanism. The system triggers event recording every time a legend is added or deleted, and appends the operation events to the context sequence in the order of the operation occurrence. The length of the context extraction window is configurable, and the legend placement actions with the most recent preset number of times are retained as valid context.

4. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Step S2 specifically involves: The number of times multiple legends are used in the same order in historical drawing is counted. Legends with a higher number of times they are used together are grouped into legend groups. The co-occurrence frequency of each legend group is calculated based on the number of times they are used together. The co-occurrence relationship matrix is ​​obtained by combining the co-occurrence frequencies of different legend groups. Each element of the co-occurrence relationship matrix represents the association strength of the two legends used together in all historical drawing records.

5. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Step S2 further includes: During the construction of the co-occurrence relationship matrix, global usage frequency statistics and individual usage frequency statistics are maintained simultaneously. The global usage frequency statistics record the cumulative number of times each legend is used in the historical drawings of all users, and the individual usage frequency statistics record the cumulative number of times each legend is used in the historical drawings of the current user. The global usage frequency is used to calculate the global popularity score, and the individual usage frequency is used to calculate the personalized preference score.

6. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Step S3 specifically involves: The context information is paired with each legend to construct different legend groups. The co-occurrence frequency of each legend group is obtained based on the co-occurrence relationship matrix. The co-occurrence frequency score is calculated based on the co-occurrence frequency. A legend recommendation list is generated for the user based on the co-occurrence frequency score. If a user places a new legend after the current legend recommendation list is generated, the system retrieves new context information based on the new legend placement action and generates a new legend recommendation list, thus achieving dynamic updating of the legend recommendation list.

7. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Step S3 further includes: The system calculates the individual usage frequency of each legend and its score based on that frequency. It also calculates the global usage frequency of each legend and its global popularity score. The system then performs a weighted average of the co-occurrence frequency score, the individual usage frequency score, and the global popularity score to obtain a total recommendation score. Finally, it dynamically generates a legend recommendation list based on the total recommendation score.

8. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Also includes: Step S4: Record the user's selection behavior for each recommendation; If the user accepts the recommendation, the correlation strength between the corresponding recommended legend and the context is recorded; if the user ignores the recommendation and searches manually, the keywords manually searched by the user are recorded, and the alias library and semantic vector of the legend are updated based on the keywords; if the user accepts the recommendation but adjusts the position or attributes of the legend, the adjustments are recorded to optimize the corresponding layout algorithm.

9. The context-based digital flowchart-assisted drawing method according to claim 1, characterized in that, Also includes: Intelligent recommendations are made based on the user's real-time actions while drawing. Specifically, the intelligent recommendations include recommending legends related to the legends where the mouse position is located, based on the user's mouse position information. A typical configuration library is constructed, which stores predefined legend combination templates. The legend combination templates include legend combinations and their connection rules. When a user places a legend, the complete combination is automatically recommended based on the co-occurrence relationship. After the user confirms the recommended content, all devices in the corresponding combination are automatically generated and automatically connected according to preset rules. After the user selects and places a recommended legend, the system checks the required attributes of the selected legend, provides prompts for missing or incorrect attributes, and automatically fills in the relevant or recommended attribute values ​​based on the selected legend.

10. A context-based digital flowchart-assisted drawing device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the context-based digital flowchart-assisted drawing method as described in any one of claims 1-9.

Images