Function map generation method and system

Abstract

To provide a method of and a system for generating a function map using an artificial intelligence model.SOLUTION: A function map generating method according to the present invention has step S310 of acquiring a difficult point and first data relating to the difficult point from a first response content of a first query corresponding to the difficult point, step S320 of acquiring a use target and second data relating to the use target from a second response content of a second query corresponding to the user target, function guidance step S331 of acquiring an implemented function and a kind of the target based on the difficult pint and the use target, function associating step S332 of acquiring a required function based on the kind of the target, function comparison step S333 of comparing meanings of the implemented function with the requested function to acquire a content of difference set, and function map generating step S340 of generating a function map based on the kind of the target, the implemented function and the content of the difference set.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a human-computer interaction mechanism, and particularly to a method and system for generating a function map based on an artificial intelligence model.

Background Art

[0002] In today's era of comprehensive digital transformation, the user interface (UI) and user experience (UX) of websites have become extremely important. UI and UX not only directly affect the operation experience but also are directly related to efficiency. For example, in a management website, it is necessary to display data according to importance and be able to quickly link to related data. Therefore, websites with convenient operation procedures and intuitive and easy-to-manage dashboards have become the main theme of web page design. However, to achieve the above objectives, it is necessary for UI / UX designers to intervene, repeatedly conduct hearings with customers, accurately grasp the needs, and there is no doubt that the time cost of website development will increase.

[0003] In general UI / UX design and development procedures, UI / UX designers conduct pre-needs hearings with customers, create a function map (Function Map) of the website according to the type of functions such as image monitoring, device control, event notification, etc., and create a first version of a wireframe so that customers can view the functions. The wireframe has no visual elements and aims to show the structure, functions, and procedures of the website. If there are new needs or opinions from customers, UI / UX designers conduct hearings again, update the function map, remake the wireframe, and repeat the procedure until an agreement is reached. Finally, the website development team enters the UI design stage based on the wireframe and creates mockups and prototypes.

Summary of the Invention

[0004] Throughout the entire design and development cycle, needs assessment is a crucial and unavoidable step in the development process. However, because customers may not fully understand their own needs, multiple assessments are necessary to uncover core needs and supplement them with relevant information. Before proceeding with UI design and development, differences in understanding between both parties often necessitate revisions and reconfirmations until agreement is reached. Consequently, needs assessment takes up a significant amount of time in the development cycle, and its complexity and lack of standardization make it difficult to streamline. [Means for solving the problem]

[0005] The present invention provides a method and system for generating function maps based on artificial intelligence (AI) that can effectively reduce the time cost associated with needs assessment.

[0006] The function map generation method based on an artificial intelligence model includes the following steps performed through a processor: In the first query step, the artificial intelligence model is used to obtain the difficulty and the first data related to the difficulty from the first response of the first query corresponding to the difficulty. In the second query step, the artificial intelligence model is used to obtain the target and the second data related to the target from the second response of the second query corresponding to the target. The completeness determination step includes the following steps: In the function induction step, the artificial intelligence model is used to obtain the implemented function and the type of target based on the difficulty and the target. In the function association step, the artificial intelligence model is used to obtain the requested function based on the type of target. In the function comparison step, the artificial intelligence model is used to perform a semantic comparison between the implemented function and the requested function and obtain the contents of the difference set. In the function map generation step, a function map is generated based on the type of target, the implemented function, and the contents of the difference set.

[0007] In one embodiment of the present invention, the first query step includes generating a prompt corresponding to a first response content, executing a noun capture program via an artificial intelligence model based on the prompt corresponding to the first response content and obtaining a first noun response result, and executing a semantic capture program via an artificial intelligence model and obtaining the difficulty and first data related to the difficulty based on the first response content and the first noun response result.

[0008] In one embodiment of the present invention, the second query step includes generating a prompt corresponding to the second response content, executing a noun capture program via an artificial intelligence model based on the prompt corresponding to the second response content to obtain a second noun response result from the second response content, and executing a semantic capture program via an artificial intelligence model to obtain a usage target and second data related to the usage target based on the second response content and the second noun response result.

[0009] In one embodiment of the present invention, the function map generation method further includes a third query step, the third query step includes receiving a third response to a third query corresponding to a data source, generating a prompt corresponding to the third response, executing a noun capture program via an artificial intelligence model based on the prompt corresponding to the third response and obtaining a third noun response result from the third response, and executing a semantic capture program via an artificial intelligence model to obtain source information based on the third response and the third noun response result.

[0010] In one embodiment of the present invention, the function map generation step further includes updating the function map based on source information.

[0011] In one embodiment of the present invention, the function map generation step further includes generating prompts corresponding to difficulties and target users, and using an artificial intelligence model to execute a function guidance program based on the prompts and obtain the types of functions to be performed and the target users.

[0012] In one embodiment of the present invention, the function association step includes generating a prompt corresponding to the type of target, and using an artificial intelligence model to execute a function association program based on the prompt and obtain the requested function.

[0013] In one embodiment of the present invention, the function map generation step includes generating a function map based on the implemented function and the type of target after the function induction step, and updating the function map based on the contents of the difference set after the function comparison step.

[0014] The function map generation system based on the artificial intelligence model of the present invention includes an artificial intelligence model, an interactive interface, and a processor coupled to the artificial intelligence model and the interactive interface. In response to the interactive interface receiving a first response to a first query corresponding to a difficulty, the processor is configured to use the artificial intelligence model to obtain the difficulty and first data related to the difficulty from the first response. In response to the interactive interface receiving a second response to a second query corresponding to a target, the processor is configured to use the artificial intelligence model to obtain second data representing the target and related to the target from the second response. The processor is further configured to use the artificial intelligence model to obtain the implemented function and the type of target based on the difficulty and the target, use the artificial intelligence model to obtain the requested function based on the type of target, use the artificial intelligence model to perform a semantic comparison between the implemented function and the requested function, obtain the contents of the difference set, and generate a function map based on the type of target, the implemented function, and the contents of the difference set. [Effects of the Invention]

[0015] Based on the above, by using an artificial intelligence model to capture important information and organizing it into a function map, it is possible to quickly and accurately grasp customer needs, reduce the number of interviews between designers and customers, and effectively reduce the time cost associated with needs assessment. [Brief explanation of the drawing]

[0016] [Figure 1] This is a block diagram of a function map generation system according to one embodiment of the present invention. [Figure 2] This is a schematic diagram of a robot structure according to one embodiment of the present invention. [Figure 3] This is a flowchart of a method for generating a function map according to one embodiment of the present invention. [Figure 4] This is a schematic diagram of the functional map generation process according to one embodiment of the present invention. [Figure 5A] This is a schematic diagram of multiple prompt templates according to one embodiment of the present invention. [Figure 5B] This is a schematic diagram of multiple prompt templates according to one embodiment of the present invention. [Figure 5C] This is a schematic diagram of multiple prompt templates according to one embodiment of the present invention. [Figure 6] This is a schematic diagram of a problem generation program according to one embodiment of the present invention. [Figure 7] This is a schematic diagram of a noun capture program based on one embodiment of the present invention. [Figure 8] This is a schematic diagram of a semantic capture program based on one embodiment of the present invention. [Figure 9] This is a schematic diagram of a functional induction program according to one embodiment of the present invention. [Figure 10] This is a schematic diagram of a function association program according to one embodiment of the present invention. [Figure 11A] This is a schematic diagram of the content presented by the interactive interface of one embodiment of the present invention. [Figure 11B]Schematic diagram of content presented by an interactive interface according to an embodiment of the present invention.

Embodiment for Carrying out the Invention

[0017] FIG. 1 is a block diagram of a function map generation system according to an embodiment of the present invention. Please refer to FIG. 1. The function map generation system includes a processor 110, an interactive interface 120, and an artificial intelligence (AI) model 130. The processor 110 is coupled to the interactive interface 120 and the AI model 130.

[0018] The processor 110 is, for example, a central processing unit (CPU), a physical processing unit (PPU), a programmable microprocessor, an embedded control chip, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other similar devices.

[0019] The interactive interface 120 is a user interface (UI) that is a medium for interaction and information exchange between the processor 110 and the user. For example, the interactive interface 120 includes a human-computer interaction (HCI) and a graphical user interface (GUI). In one embodiment, the interactive interface 120 may be implemented by a display and an input device, or may be implemented by a touch screen.

[0020] The AI ​​model 130 may be located within the same electronic device as the processor 110. Alternatively, the AI ​​model 130 and the processor 110 may be located in different electronic devices and connected to each other via wired or wireless means. The AI ​​model 130 is, for example, a Large-Scale Language Model (LLM), composed of an artificial neural network with many parameters, and trained on a large amount of unlabeled text using self-supervised or semi-supervised learning. Examples of LLMs include ChatGPT, GPT4, LLaMA-65B, and PaLM-62B. All of these models have achieved very good results in various Natural Language Processing (NLP) evaluations. Evaluation items include common sense reasoning, closed-book question answering, reading comprehension, mathematical reasoning, program code generation, and Massive Multitask Language Understanding (MMLU). In recent years, leading technology leaders have also been actively reducing the number of LLMs while incorporating more powerful capabilities.

[0021] In one embodiment, a UI / UX design team can collect past interview data, organize three key pieces of information (such as challenges, target users, and related data) and the type of website to be ultimately implemented and the functions to be performed, and construct an input corpus and output results (including the output corpus). The input used to train the AI ​​model 130 is the three key pieces of information. The output used to train the AI ​​model 130 is the type of website and the functions to be performed. Results obtained from subsequent analysis using the AI ​​model 130 can also be fed back into the AI ​​model 130 to improve its accuracy.

[0022] The AI ​​model 130 employed in this embodiment has the following functions: sentence decomposition, which can break down the structure of a sentence, such as nouns and verbs; semantic comprehension, which can understand the context, analyze the meaning, and summarize; general expertise for learning knowledge through training using a large amount of teaching materials in a wide range of fields; topic modeling, which can analyze multiple keywords and guide corresponding topics into groups; association, which can associate related things at various levels depending on the topic; word embedding, which can convert words and sentences into a semantic distance space; and highly anthropomorphic sentence generation, which can generate sentences according to the situation, emotion, role, etc.

[0023] The AI ​​model 130 has strong generalization capabilities. When applying the AI ​​model 130 to a different task, retraining is not necessary; only relevant prompts (including task description, examples, output format, etc.) need to be provided to output the corresponding results. This method is called in-context learning, where the output results differ depending on the input prompt. The design of the prompts is also called prompt engineering. Prompt engineering is performed by the processor 110. Through the design, one or more corresponding prompts are selected or generated according to the current status (such as the function induction step S331, the function association step S332, the function comparison step S333, and the third query step S350). Finally, the AI ​​model 130 can be driven through the prompts to obtain the desired results.

[0024] Figure 2 is a schematic diagram of a robot structure according to one embodiment of the present invention. This embodiment includes three roles: user, robot 200, and AI model 130 (see Figure 1). Based on its expertise and experience in UI / UX design, robot 200 guides the user to answer questions and uses the functions of AI model 130 to obtain important information necessary for website design, such as data to display, data sources, usage scenarios, and development functions. The key information is organized into a function map. Finally, after user confirmation, the final version of the function map is generated.

[0025] Please refer to Figure 2. The robot 200 includes a question query module 210, a response analysis module 220, and a map construction module 230. A module is, for example, a software module consisting of one or more program codes stored in memory. The memory is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk, or other similar device, or a combination thereof. The question query module 210, the response analysis module 220, and the map construction module 230 can also be implemented by hardware chips or circuits.

[0026] The processor 110 drives the question query module 210, the response analysis module 220, and the map building module 230 to perform the corresponding functions. The question query module 210 is configured to call the AI ​​model 130 to generate a corresponding question according to key information. The response analysis module 220 is configured to call the AI ​​model 130 to analyze the content of the response to the query. The map building module 230 is configured to call the AI ​​model 130 to find a list of functions and build a function map based on the results obtained by the response analysis module 220.

[0027] To facilitate understanding, Figures 3 through 11B are shown below and explained together with Figures 1 and 2.

[0028] Figure 3 is a flowchart of a function map generation method according to one embodiment of the present invention. Figure 4 is a schematic diagram of function map generation according to one embodiment of the present invention. Figures 5A to 5C are schematic diagrams of multiple prompt templates according to one embodiment of the present invention. Figure 6 is a schematic configuration diagram of a problem generation program according to one embodiment of the present invention. Figure 7 is a schematic diagram of a noun capture program according to one embodiment of the present invention. Figure 8 is a schematic diagram of a semantic capture program according to one embodiment of the present invention. Figure 9 is a schematic diagram of a function guidance program according to one embodiment of the present invention. Figure 10 is a schematic diagram of a function association program according to one embodiment of the present invention. Figures 11A and 11B are schematic diagrams of content presented by an interactive interface according to one embodiment of the present invention.

[0029] First, please refer to Figure 3. The function map generation method includes a first query step S310, a second query step S320, a completeness determination step S330, and a function map generation step S340. The completeness determination step S330 includes a function induction step S331, a function association step S332, and a function comparison step S333. The function map generation method may also include a third query step S350. The third query step S350 is not a required step and may be added or omitted depending on the situation.

[0030] Please refer to Figure 4. The first query Q1, the second query Q2, and the third query Q3 used in the first query step S310, the second query step S320, and the third query step S350 are generated by the question query module 210. The type of question for the first query Q1 is difficulty, the type of question for the second query Q2 is target audience (including purpose of use and target audience), and the type of question for the third query Q3 is data source. Specifically, the question query module 210 selects a corresponding prompt template according to the type of question. The AI ​​model 130 then generates the corresponding first query Q1, second query Q2, or third query Q3. The questions generated by the AI ​​model 130 may be dynamic, informal, polite, and anthropomorphic, including a service-oriented tone.

[0031] Please refer to Figure 6. Taking the first query Q1, which addresses a difficulty, as an example, the question query module 210 adopts a prompt template 510 according to the type of question, "Difficulties the website needs to solve," and introduces key information corresponding to the "difficulties." At the same time, it can also supplement previously collected information, such as proper nouns and fields, in context to generate the final prompt. After deciding to adopt the prompt template 510, the content of the question type "Difficulties the website needs to solve" is entered into area 511 of the prompt template 510 to obtain the prompt. Next, the question generation program P0 is executed using the AI ​​model 130 to generate the corresponding first query Q1 based on the prompt. For example, as shown in the first query Q1 in Figure 11A, it presents, "To better support website development, we need to understand the difficulties you are facing so that we can provide the most effective support. Please share some specific issues that the website needs to solve so that we can provide more specific advice and support." The second query Q2 and the third query Q3 can be inferred.

[0032] After generating a query, the query module 210 presents the query (first query Q1, second query Q2, or third query Q3) on the interactive interface 120, and the interactive interface 120 receives the corresponding response content (first response content R1, second response content R2, or third response content R3). Taking Figure 11A as an example, after the interactive interface 120 presents the first query Q1, the first response content R1 is received via the interactive interface 120.

[0033] Referring to Figures 3 and 4, in the first query step S310, the processor 110 uses the AI ​​model 130 to obtain the difficulty and the first data related to the difficulty from the first response content R1 of the first query Q1 corresponding to the difficulty, and stores the first data in the data and source database D1.

[0034] Specifically, referring to Figure 7, the response analysis module 220 selects a prompt template 520 based on the first response content R1. Then, it embeds the first response content R1 into area 521 of the prompt template 520 to generate a prompt corresponding to the first response content R1. Next, it executes the noun capture program P1 via the AI ​​model 130 based on the prompt corresponding to the first response content R1 to obtain the first noun response result.

[0035] The noun capture program P1 has the function of calling the AI ​​model 130 via a prompt, capturing proper nouns, odd nouns, and branched nouns that appear in the first response content R1, and explaining their meanings. Next, in the "parsing" process, the response parsing module 220 determines whether the noun list listed in the first noun response result contains known nouns (nouns that the AI ​​model 130 knows and can interpret) or unknown nouns. For example, whether a noun is an unknown noun can be determined by searching whether strings such as "online search" or "search online" appear in the first noun response result (as shown in Example 1-1 below). If the noun is an unknown noun, a search engine may be used to perform a further search and obtain the corresponding explanation. After that, the meaning of the noun and its corresponding explanation are confirmed. For example, the noun and explanation are presented to the interactive interface 120 for the user to confirm. Finally, the results are stored in the noun database D2. If no relevant explanation can be found, the user can also receive an explanation provided by the user through the interactive interface 120.

[0036] Example 1-1

[0037] [Table 1]

[0038] In other words, if AI model 130 encounters an unknown noun that it cannot understand, it searches for the corresponding meaning through a search engine, confirms the meaning, and provides the user with further confirmation. Regardless of whether the final noun is known or unknown, it displays its meaning, including any corrections, to the user for confirmation.

[0039] Furthermore, during the "analysis" process, if the response analysis module 220 confirms that proper nouns, non-uniform nouns, or strange nouns do not currently exist by searching for default strings such as "does not appear," "not found," "did not appear," or "not found," the process terminates without proceeding to the next step. This means that the semantic verification step is not entered, and the list in the noun database D2 is not updated. For example, referring to Example 2-1, the first noun response result indicates "does not appear," so it is determined that proper nouns, non-uniform nouns, or strange nouns do not exist.

[0040] Example 2-1

[0041] [Table 2]

[0042] Next, referring to Figure 8, the response analysis module 220 executes the semantic capture program P2 via the AI ​​model 130 and acquires the difficulty and first data related to the difficulty based on the first response content R1 and the first noun response result. The semantic capture program P2 aims to capture key information through semantic understanding by the AI ​​model 130. Different results can be obtained in "prompt generation" and semantic capture program P2 based on the "information type" input. For example, a prompt template is selected according to the "information type," and the AI ​​model 130 analyzes the output semantic response result according to the "information type."

[0043] For example, referring to Example 2-2 below, the response analysis module 220 selects a prompt template 530 based on the information type. Then, it enters the first response content R1 into area 531 of the prompt template 530, enters the type of question (e.g., "difficulties the website should solve") into area 532, and generates a prompt for semantic capture. Subsequently, via the AI ​​model 130, based on the generated prompt... Meaning Capture Program P2The process is executed to obtain a first semantic response result (including "difficulties:...", "data:...", etc.). Then, the meaning of the first semantic response result is confirmed. For example, the first semantic response result is presented to the interactive interface 120 for the user to review.

[0044] Example 2-2

[0045] [Table 3]

[0046] In another embodiment, referring to Example 1-2 below, a first response content R1 is entered in area 551 of prompt template 550, the type of question (e.g., "Difficulties the website must solve") is entered in area 553, and a proper noun and its description (e.g., "aaaaaa: refers to the cyclic growth rate of a tumor, and the size is calculated once a month") is entered in area 552, thereby generating a prompt for semantic capture. Then, it is assumed that a noun capture program P1 is executed via AI model 130 based on the generated prompt to obtain a first semantic response result (including "Difficulties:...", "Data:...", etc.). The purpose of adding the collected proper nouns to the prompt is to avoid them being miscaptured or missed.

[0047] Example 1-2

[0048] [Table 4]

[0049] Next, referring to Figures 3 and 4, in the second query step S320, the processor 110 uses the AI ​​model 130 to obtain the target and related second data from the second response content R2 of the second query Q2 corresponding to the target, and stores the second data in the data and source database D1.

[0050] The response analysis module 220 generates a prompt corresponding to the second response content R2, and then, via the AI ​​model 130, executes the noun capture program P1 based on the prompt corresponding to the second response content R2, and obtains the second noun response result. Subsequently, the response analysis module 220 executes the semantic capture program P2 via the AI ​​model 130, and obtains second data related to the target of use and the target of use based on the second response content R2 and the second noun response result. Here, the target of use includes the user and the purpose of use.

[0051] The processes of the noun capture program P1 and the semantic capture program P2 executed in the second query step S320 are the same as in the first query step S310. First, the noun capture program P1 determines whether or not there are proper nouns in the second response content R2. For example, suppose the second response content R2 is "The website is used for internal monitoring purposes only and is not open to the public," as shown in Example 3 below. The noun capture program P1 determines that there are no proper nouns, strange nouns, or non-unique nouns in the provided content. The response analysis module 220 adopts the prompt template 540, enters the second response content R2 into area 541 of the prompt template 540, and enters the type of question (e.g., "Who uses the website? What is the purpose?") into area 542 to generate the corresponding prompt. Then, the semantic capture program P2 is executed via the AI ​​model 130 to obtain the target users and the second data.

[0052] Example 3

[0053] [Table 5]

[0054] In the third query step S350, the processor 110 uses the AI ​​model 130 to obtain the data source from the third response content R3 of the third query Q3 corresponding to the data source. After receiving the third response content R3 of the third query Q3 corresponding to the data source, it generates a prompt corresponding to the third response content R3. Through the AI ​​model 130, it executes the noun capture program P1 based on the corresponding prompt and obtains the third noun response result from the third response content R3. Through the AI ​​model 130, it executes the semantic capture program P2, obtains source information based on the third response content R3 and the third noun response result, and stores the source information in the data and source database D1.

[0055] In the third query step S350, the processor 110 queries the user for the data source according to the first data obtained in the first query step S310 and the second data obtained in the second query step S320. After receiving a response from the user, it determines whether the capture of the source is complete. If the capture is not complete, it first stores the information in the database, and then generates another query using the query module 210, as shown in Examples 4 and 5 below, and continues this cycle in this order until the capture of all data sources is complete.

[0056] Example 4

[0057] [Table 6]

[0058] Example 5

[0059] [Table 7]

[0060] The processes of the noun capture program P1 and the semantic capture program P2 executed in the third query step S350 are the same as those in the first query step S310. First, the noun capture program P1 determines whether or not a proper noun exists in the second response content R2. Then, in the semantic capture program P2, the response analysis module 220 adopts the prompt template 550, enters the third response content R3 into area 561 in the prompt template 560, and enters the first data and the second data into area 562 to generate a corresponding prompt. Subsequently, based on this prompt, the noun capture program P1 is executed via the AI ​​model 130 to obtain source information.

[0061] Next, the integrity determination step S330 is performed. In the function guidance step S331, the processor 110 uses the AI ​​model 130 to acquire the function to be performed and the type of target based on the difficulty and the target of use. The map construction module 230 generates prompts corresponding to the difficulty and the target of use, and then uses the AI ​​model 130 to execute the function guidance program P3 based on the prompts to acquire the function to be performed and the type of target.

[0062] Referring to Figure 9, the map construction module 230 selects the prompt template 570, enters the difficulty points in area 571, and enters the target users (purpose and target) in area 572 to generate the corresponding prompts. Next, the AI ​​model 130 is used to execute the function guidance program P3 based on the prompts to obtain the implemented functions and target types. The AI ​​model 130 can effectively guide the implemented functions and analyze the target types (e.g., website types), and the AI ​​model 130 analyzes and outputs the implemented functions and website types using "Functional Requirements:..." and "Website Type," as shown in Example 6.

[0063] Example 6

[0064] [Table 8]

[0065] In the function association step S332, the processor 110 uses the AI ​​model 130 to obtain the requested function based on the type of target. The map construction module 230 generates a prompt corresponding to the type of target, and uses the AI ​​model 130 to execute the function association program P4 based on the prompt to obtain the requested function.

[0066] Referring to Figure 10 and Example 7, the map building module 230 selects a prompt template 580, enters the target type in the domain 581, and generates a corresponding prompt. Then, it utilizes the general domain knowledge capabilities of the AI ​​model 130 to associate other required functions based on the target type, complementing the expertise that the user does not possess.

[0067] Example 7

[0068] [Table 9]

[0069] AI model 130 uses specialized associations to acquire the required functions relevant to implementation. Because AI model 130 has strong associative capabilities, the acquired prompts are described as classifying the acquired required functions according to their importance and outputting only the high-importance required functions, but this is not limited to this and can be determined according to the requirements.

[0070] Subsequently, in the function comparison step S333, the processor 110 executes the function comparison program P5 using the AI ​​model 130 to perform a semantic comparison between the implemented function and the requested function, and obtains the contents of the difference set. For example, after obtaining the "requested function" and the "implemented function," a semantic comparison is performed using the word embedding capability of the AI ​​model 130. The main purpose of this process is to find the common parts by adding the "requested function" that will actually be used in implementation to the "implemented function" proposed by the user. In the function comparison step S333, overlapping items between the "requested function" and the "implemented function" are deleted so that they do not overlap, and the contents of the difference set are obtained and output.

[0071] For example, as shown in Example 6, the requested functionality based on user responses includes integrating park information from various departments and tracking total visitor numbers, total attendance, and average stay duration. The requested functionality obtained through association by AI model 130 is as shown in Example 7 and includes user authentication and access control, real-time monitoring data visualization, alarm and notification systems, and logging and auditing functions. After comparison, the contents of the difference set include user authentication and access control, real-time monitoring data visualization, alarm and notification systems, and logging and auditing functions.

[0072] Subsequently, in step S340, the processor 110 generates a function map based on the type of target, the function to be performed, and the contents of the difference set. Furthermore, after obtaining the type of target and the function to be performed, the processor 110 may pre-construct the function map FM. After obtaining the contents of the difference set, it updates the function map FM by adding further contents to the difference set. Also, after obtaining source information, it updates the function map FM based on the source information.

[0073] A Functional Map (FM) consists of four parts, for example: website type, functions, data, and existing systems and processes. The Functional Map (FM) is presented in the form of a mind map, organized from abstract to concrete. The website type is the top-level core of the entire Functional Map (FM), determining the overall direction of the website. The second level contains the functional items necessary for building the website, obtained based on the Functional Introduction Program (P3) and Functional Association Program (P4).

[0074] Referring to Figures 11A and 11B, this embodiment shows the content presented by an interactive interface 120 with a website design. First, the processor 110 uses the question query module 210 to call the AI ​​model 130 and generate a first query Q1 to query the pain points that the website needs to solve, and presents the first query Q1 to the interactive interface 120. Next, the processor 110 receives a first response R1 corresponding to the first query Q1 via the interactive interface 120. Subsequently, the processor 110 uses the response analysis module 220 to call the AI ​​model 130 to obtain the pain points and first data, and generates corresponding confirmation information A1 for user confirmation based on the pain points and the first data presented to the interactive interface 120.

[0075] Next, the processor 110 uses the question query module 210 to call the AI ​​model 130 and generates a second query Q2 to query the target of use (purpose and target) of the website presented to the interactive interface 120. Then, it receives a second response R2 corresponding to the second query Q2 via the interactive interface 120. Subsequently, the processor 110 uses the response analysis module 220 to call the AI ​​model 130 to obtain the target of use (purpose and target) and the second data, and generates corresponding confirmation information A2 to be presented to the interactive interface 120 based on the target of use and the second data for user confirmation.

[0076] Subsequently, the processor 110 uses the question query module 210 to call the AI ​​model 130, generates a third query Q3-1 to query the data source, and presents the third query Q3-1 to the interactive interface 120. Then, it receives the corresponding response content R3-1 via the interactive interface 120. After that, the processor 110 uses the response analysis module 220 to call the AI ​​model 130 to obtain source information, and generates corresponding confirmation information A3-1 to be presented to the interactive interface 120 based on the source information for user confirmation.

[0077] Furthermore, since response content R3-1 does not fully answer all data sources, processor 110 uses the question query module 210 to call AI model 130 and generate a fourth query Q3-2 to query the data sources, and presents the fourth query Q3-2 to the interactive interface 120. Subsequently, it receives the corresponding reply content R3-2 via the interactive interface 120. After that, processor 110 uses the response analysis module 220 to call AI model 130 to obtain source information and generates corresponding confirmation information A3-2, which is presented to the interactive interface 120 based on the source information for user confirmation.

[0078] Subsequently, the processor 110 uses the map construction module 230 to call the AI ​​model 130, generates function map information A4 corresponding to the function map, and presents the function map information A4 to the interactive interface 120 for user confirmation.

[0079] The above embodiment can be divided into the following four stages in order.

[0080] In the first stage, in the first query step S310, the robot 200 queries and captures the difficulties that the website needs to solve, as well as keywords for data and files mentioned in the response. In the second query step S320, the robot 200 queries and captures the target audience and purpose of the website. Subsequently, in the function guidance step S331, the robot 200 analyzes the type of website (target audience) via the AI ​​model 130 based on the information captured in the first query step S310 and the second query step S320, obtains the website type and the corresponding implementation functions, and constructs a preliminary function map (FM).

[0081] In the second stage, in the function association step S332, based on the type of website, the robot 200 calls the AI ​​model 130 to associate the required functions common to that type of website, divides them according to importance, acquires the functions with the highest importance, and compares the completeness of the required functions with the previously acquired implemented functions. If it is found that the implemented functions acquired in the function guidance step S331 are missing, the function map FM is updated. At this point, the capture of function and function map FM data is complete.

[0082] In the third stage, in addition to the key information captured in the second query step S320 of the first stage, the robot 200 captures source information to update the function map FM. It then captures keywords for the data and files and sends them to the third query step S350 to query the source information for the data and files.

[0083] In the fourth stage, the generated feature map (FM) is presented to the user for review. If there are no problems, the feature map (FM) becomes the final version. UI / UX designers continue to receive the feature map (FM) for website design.

[0084] In one embodiment, the function map FM generated by the processor 110 may be imported into a prototype website development platform with AI capabilities to generate prototype website program code. The processor 110 then packages the website program code, deploys it to a cloud environment through a compilation engine, and records the browsing URL. The browsing URL is then returned to the user via the robot 200 so that the user can further browse the web page. For example, after the robot 200 records the browsing URL, it packages the browsing URL (digital signal) into a set of logical transmission data (i.e., a data frame), drives a communication chip or communication circuit, and transmits the transmission data to a pre-specified electronic device associated with the user via email, short message service (SMS), push technology, etc.

[0085] In summary, the system of the present invention can operate 24 hours a day, improving the work efficiency of designers. Furthermore, the present invention uses an AI model with deep conversational capabilities and humanized speech patterns to provide a superior customer experience compared to conventional robots. Based on the designer's expertise, the AI ​​model can effectively capture important information. The proper noun understanding mechanism can quickly resolve discrepancies in recognition between the two parties, improving the efficiency of understanding needs. By directly generating function maps, designers can avoid the initial needs assessment period and directly address clear needs, accelerating the entire design and development process. [Industrial applicability]

[0086] The function map generation method and system based on artificial intelligence models can be applied to UI / UX design and development procedures, enabling the creation of mockups and prototypes. [Explanation of Symbols]

[0087] 110: Processor 120: Interactive Interface 130: AI Model 200: Robot 210: Question Query Module 220: Response Analysis Module 230: Map Construction Module 510~580: Prompt templates 511, 521, 531, 532, 541, 542, 551, 552, 553, 561, 562, 571, 572, 581: Area A1, A2, A3-1, A3-2: Confirmation information A4: Function Map Information D1: Data and source databases D2: Noun Database FM: Function Map Q1: First query Q2: Second query Q3: Third query P0: Question generation program P1: Noun acquisition program P2: Semantic acquisition program P3: Functional Induction Program P4: Function association program P5: Feature Comparison Program R1: First response R2: Second response R3: Third response S310: First query step S320: Second query step S330: Steps of the integrity determination step S331: Function induction step S332: Function association step S333: Functional Comparison Step S340: Function map generation step S350: Third query step

Claims

1. A method for generating a function map based on an artificial intelligence model, which performs the following multiple steps via a processor, A first query step includes generating a prompt for semantic capture, and via the artificial intelligence model, executing a semantic capture program on the first response content of a first query corresponding to a difficulty point based on the prompt for semantic capture, thereby obtaining the difficulty point and first data related to the difficulty point from the first response content; A second query step includes generating another prompt for semantic capture, and via the artificial intelligence model, executing the semantic capture program on the second response content of a second query corresponding to the target, based on the other prompt, thereby obtaining the target and second data related to the target from the second response content; Integrity determination step, Using the aforementioned artificial intelligence model, a function guidance step is performed to acquire the functions to be implemented and the types of targets based on the aforementioned difficulties and the aforementioned target users. Using the aforementioned artificial intelligence model, a function association step is performed to acquire the requested function based on the type of the target, A function comparison step is performed using the artificial intelligence model to perform a semantic comparison between the implemented function and the requested function, and to obtain the contents of the difference set. A function map generation step includes generating a function map based on the type of target, the function to be implemented, and the contents of the difference set, The integrity determination step, A method for generating a function map, including the above.

2. The first query step described above further, To generate prompts for noun capture, By executing a noun capture program based on the prompt for noun capture via the artificial intelligence model, a first noun response result is obtained from the first response content. By executing the semantic capture program on the first response content based on the prompt for semantic capture via the artificial intelligence model, the difficulties and first data related to the difficulties are obtained based on the first response content and the first noun response result. Includes, The second query step described above is: To generate another prompt for noun capture, The process involves obtaining a second noun response result from the second response content by executing the noun capture program on the second response content based on the other prompt for noun capture via the artificial intelligence model, By executing the semantic capture program on the second response content based on the other prompt for semantic capture via the artificial intelligence model, the user and second data related to the user are obtained based on the second response content and the second noun response result. Further including, The method for generating a function map according to claim 1.

3. A third query step, and further including, The first query step described above is: Receiving the third response content of a third query corresponding to the data source, To generate a prompt corresponding to the third response content, The process involves executing the noun capture program based on the prompt corresponding to the third response content via the artificial intelligence model, and obtaining a third noun response result from the third response content. The process involves executing the semantic acquisition program via the artificial intelligence model and obtaining source information based on the third response content and the third noun response result. Includes, The function map generation step described above is: Further includes updating the function map based on the aforementioned source information, The method for generating a function map according to claim 2.

4. The aforementioned function induction step is, To generate prompts corresponding to the aforementioned difficulties and the aforementioned target users, Using the artificial intelligence model, the function guidance program is executed based on the prompt, and the implemented function and the type of target are obtained. Includes, The aforementioned function association step is, To generate a prompt corresponding to the type of the aforementioned target, Using the aforementioned artificial intelligence model, the function association program is executed based on the prompt, and the requested function is obtained. Includes, The function map generation step described above is: After the function induction step, the function map is generated based on the function to be implemented and the type of target, After the function comparison step, the function map is updated based on the contents of the difference set. including, The method for generating a function map according to claim 1.

5. Artificial intelligence models and, Interactive interface, A processor coupled to the artificial intelligence model and the interactive interface, Includes, In response to the interactive interface receiving a first response to a first query corresponding to a difficulty, the processor is configured to generate a prompt for semantic capture via the artificial intelligence model, and to execute a semantic capture program on the first response based on the prompt for semantic capture, thereby obtaining the difficulty and first data related to the difficulty from the first response. In response to the interactive interface receiving a second response to a second query corresponding to the target object, the processor is configured to generate another prompt for semantic capture via the artificial intelligence model, and to execute the semantic capture program on the second response content based on the other prompt for semantic capture, thereby obtaining the target object and second data related to the target object from the second response content. The aforementioned processor further, Using the aforementioned artificial intelligence model, the functions to be implemented and the types of targets are obtained based on the aforementioned difficulties and the aforementioned target users. Using the aforementioned artificial intelligence model, the requested function is obtained based on the type of the target. Using the aforementioned artificial intelligence model, a semantic comparison is performed between the implemented function and the requested function, and the contents of the difference set are obtained. A system configured to generate a function map based on the type of target, the function to be implemented, and the contents of the difference set. A function map generation system based on an artificial intelligence model.

Images