本創作是關於資訊處理技術，特別是關於一種設計素材搜尋系統。This work is about information processing technology, particularly about a design material search system.
隨著數位產品開發流程日趨複雜，設計團隊常需使用設計協作平台(例如 Figma)進行使用者介面與互動流程之協同設計。在此類平台中，設計師會持續產出大量設計素材，並進行多版本修改與共享。然而，隨著專案內容擴增，設計素材數量龐大且版本繁雜，設計師及其他協作者往往難以快速搜尋並定位所需之特定素材。As digital product development processes become increasingly complex, design teams often rely on collaborative design platforms (such as Figma) to collaborate on user interfaces and interactive flows. On these platforms, designers continuously generate large amounts of design assets, modifying and sharing them in multiple versions. However, as project scope grows, the sheer volume and complexity of these assets often makes it difficult for designers and other collaborators to quickly search for and locate the specific assets they need.
傳統的設計素材管理方式大多仰賴設計檔案命名、資料夾分類、時間戳記(timestamp)或編輯者資訊進行初步篩選和整理，但無法有效處理語意導向的素材查詢，例如依據功能、版面結構或設計意圖等自然語言描述進行查找。此外，現有平台亦缺乏即時提示查詢關鍵字、輔助使用者輸入查詢內容之機制，導致素材定位效率低落，進而影響設計流程之整體協作與追蹤。Traditional design resource management methods rely heavily on file naming, folder categorization, timestamps, or editor information for initial screening and organization. However, these methods are ineffective for semantically driven resource searches, such as those based on natural language descriptions such as function, layout structure, or design intent. Furthermore, existing platforms lack mechanisms to instantly prompt search keywords or assist users in entering search results, resulting in inefficient resource locating and impacting overall collaboration and tracking within the design process.
因此，需要一種設計素材搜尋系統，以解決上述問題。Therefore, a design material search system is needed to solve the above problems.
本創作之實施例提供一種設計素材搜尋系統，其包含顯示裝置、儲存裝置及處理裝置。儲存裝置儲存設計協作平台之外掛程式。處理裝置耦接儲存裝置及顯示裝置，及設置以從儲存裝置載入外掛程式以執行下文所述步驟。處理裝置接收使用者輸入之查詢語句，並將該查詢語句輸入大型語言模型，以取得該大型語言模型所輸出之分析結果。接著，處理裝置從設計協作平台擷取該分析結果對應的設計素材資訊，然後使顯示裝置呈現該設計素材資訊。An embodiment of the present invention provides a design material search system, which includes a display device, a storage device, and a processing device. The storage device stores a plug-in for a design collaboration platform. The processing device is coupled to the storage device and the display device, and is configured to load the plug-in from the storage device to execute the steps described below. The processing device receives a query statement input by a user, and inputs the query statement into a large language model to obtain an analysis result output by the large language model. Then, the processing device extracts the design material information corresponding to the analysis result from the design collaboration platform, and then causes the display device to present the design material information.
在一實施例中， 大型語言模型是經由一微調訓練程序所建立，該微調訓練程序包含將多筆設計素材查詢語句及其對應之設計素材資料作為訓練資料，以及使用所述訓練資料進行微調訓練，以增強大型語言模型對設計語意之理解與解析能力。In one embodiment, a large language model is established through a fine-tuning training process, which includes using multiple design material query statements and their corresponding design material data as training data, and using the training data to perform fine-tuning training to enhance the large language model's ability to understand and parse design semantics.
在一實施例中， 設計素材資訊包含一或多筆設計素材資料，每筆設計素材資料又包含設計素材之一或多個代表縮圖，以及用以導向該設計素材於設計協作平台中之對應位置的導引連結。In one embodiment, the design material information includes one or more design material data, each design material data includes one or more representative thumbnails of the design material, and a navigation link for navigating to the corresponding location of the design material in the design collaboration platform.
在一實施例中， 處理裝置更在接收查詢語句時，根據使用者輸入之部分字元，自動使顯示裝置顯示一或多個與該部分字元相關之候選關鍵字，供使用者選擇以補全查詢語句。In one embodiment, when receiving a query statement, the processing device automatically causes the display device to display one or more candidate keywords related to the partial characters input by the user for the user to select to complete the query statement.
在一實施例中， 候選關鍵字的擷取來源包含設計協作平台中儲存的設計素材之文字資訊、使用者的歷史查詢紀錄，及/或預先定義之共用詞庫。In one embodiment, the sources of candidate keywords include text information of design materials stored in the design collaboration platform, user's historical query records, and/or pre-defined shared vocabulary.
本創作所提供之設計素材搜尋系統，透過大型語言模型支援以自然語言描述的查詢語句進行設計素材搜尋，可協助設計師或產設計團隊在設計協作平台中更輕鬆有效且準確地找到不同版本的設計素材，從而促進跨部門溝通協調、專案追蹤和管理設計素材。The design resource search system provided by this project uses a large language model to support search queries expressed in natural language. This system can help designers or production design teams more easily, effectively, and accurately find different versions of design resources within a design collaboration platform, thereby facilitating cross-departmental communication and coordination, project tracking, and design resource management.
以下敘述列舉本創作的多種實施例，但並非意圖限制本創作內容。實際的保護範圍，是由申請專利範圍所界定。The following description lists various embodiments of this invention and is not intended to limit its content. The actual scope of protection is defined by the scope of the patent application.
在以下所列舉的各實施例中，將以相同的標號代表相同或相似的元件或組件。In the various embodiments listed below, the same reference numerals will be used to represent the same or similar elements or components.
圖1是根據本創作之一實施例的一種設計素材搜尋系統10之系統架構圖。如圖1所示，設計素材搜尋系統10包含儲存裝置11、處理裝置12及顯示裝置13。下文將一一詳述這些元件及其特徵。FIG1 is a system architecture diagram of a design material search system 10 according to one embodiment of this invention. As shown in FIG1 , the design material search system 10 includes a storage device 11, a processing device 12, and a display device 13. These components and their features are described in detail below.
設計素材搜尋系統10可以是任何具備運算能力的電腦系統，例如可以是單獨一台電腦裝置，也可以是多台協同運作的電腦裝置所組成的電腦叢集(computer cluster)。所述電腦裝置可以是個人電腦(如桌上型電腦或筆記型電腦)或伺服器電腦，又或者是例如平板電腦或智慧型手機等行動裝置，惟本創作並不對此加以限制。The design resource search system 10 can be any computing system, such as a single computer or a computer cluster composed of multiple computers operating in coordination. The computer can be a personal computer (such as a desktop or laptop), a server computer, or a mobile device such as a tablet or smartphone, though this invention is not limited thereto.
儲存裝置11可以是任何一種包含非揮發性記憶體(如唯讀記憶體(read only memory)、電子抹除式可複寫唯讀記憶體(electrically-erasable programmable read-only memory；EEPROM)、快閃記憶體、非揮發性隨機存取記憶體(non-volatile random access memory；NVRAM))的裝置，諸如硬碟(HDD)、固態硬碟(SSD)或光碟，惟本創作並不對此加以限制。The storage device 11 can be any device including a non-volatile memory (such as read-only memory, electrically-erasable programmable read-only memory (EEPROM), flash memory, or non-volatile random access memory (NVRAM)), such as a hard disk drive (HDD), a solid-state drive (SSD), or an optical disk, but this invention is not limited thereto.
處理裝置12可以包含任何用於執行指令的一種或多種通用或專用處理器及其組合，例如可包含中央處理單元(central processing unit；CPU)及/或圖形處理單元(graphics processing unit；GPU)。處理裝置12可更包含揮發性記憶體(volatile memory)，例如動態隨機存取記憶體(dynamic random-access memory；DRAM)及/或靜態隨機存取記憶體(static random-access memory；SRAM)，以儲存運行中的程式及其所需數據。Processing device 12 may include any one or more general-purpose or special-purpose processors, or any combination thereof, for executing instructions. For example, it may include a central processing unit (CPU) and/or a graphics processing unit (GPU). Processing device 12 may further include volatile memory, such as dynamic random-access memory (DRAM) and/or static random-access memory (SRAM), to store running programs and the data they require.
顯示裝置13可包含但不限於顯示器、觸控螢幕或其他類型的視覺化顯示設備，用以顯示使用者介面(user interface；UI)以與使用者101進行互動。The display device 13 may include but is not limited to a display, a touch screen, or other types of visual display devices for displaying a user interface (UI) for interaction with the user 101.
雖然圖1未示出，但設計素材搜尋系統10可更包含一輸入裝置。輸入裝置可包含但不限於鍵盤、滑鼠、觸控筆、觸控面板或麥克風，允許使用者101在使用者介面上進行輸入、選取及編輯之類的互動操作，及/或透過打字、書寫或語音等方式輸入查詢語句102。Although not shown in FIG1 , the design resource search system 10 may further include an input device. The input device may include, but is not limited to, a keyboard, a mouse, a stylus, a touch panel, or a microphone, allowing a user 101 to perform interactive operations such as input, selection, and editing on the user interface, and/or input a query statement 102 by typing, writing, or speaking.
在一實施例中，顯示裝置13與輸入裝置與圖1中的儲存裝置11和處理裝置12均位於同一台電腦主機中，包含於設計素材搜尋系統10。在另一實施例中，設計素材搜尋系統10中的儲存裝置11與處理裝置12設置於遠端伺服器電腦上，而顯示裝置與輸入裝置則位於客戶端電腦中，客戶端電腦可包含於系統內或者在系統外部。此實施例中，外掛程式115可透過網頁形式提供使用者介面，使用者可透過瀏覽器訪問該使用者介面，進行設計素材的搜尋和管理。伺服器端則負責處理使用者請求並返回搜尋到的設計素材資訊106，而客戶端則負責顯示互動過程及設計素材資訊106。In one embodiment, the display device 13 and input device, along with the storage device 11 and processing device 12 in FIG. 1 , are all located on the same computer host, included in the design material search system 10. In another embodiment, the storage device 11 and processing device 12 in the design material search system 10 are installed on a remote server computer, while the display device and input device are located on a client computer, which can be included within the system or external to the system. In this embodiment, the plug-in 115 can provide a user interface in the form of a web page, which users can access through a browser to search and manage design materials. The server is responsible for processing user requests and returning the searched design material information 106 , while the client is responsible for displaying the interaction process and design material information 106 .
如圖1所示，儲存裝置11儲存有設計協作平台105的外掛程式115。設計協作平台105可以是用以支援使用者101進行數位設計作業之平台，具備多頁面、多元件及多人協作等功能。使用者可透過設計協作平台105建立、修改並管理各種設計素材。設計協作平台105例如可以是Figma、Adobe XD、Sketch或 Framer，惟本創作並不對此加以限制。As shown in Figure 1 , storage device 11 stores a plug-in 115 for a design collaboration platform 105. Design collaboration platform 105 can be a platform designed to support user 101 in digital design work, featuring features such as multiple pages, multiple components, and multi-person collaboration. Users can create, modify, and manage various design assets through design collaboration platform 105. Design collaboration platform 105 can be, for example, Figma, Adobe XD, Sketch, or Framer, though this creation is not limited to these platforms.
外掛程式115則為部署於該設計協作平台105中的擴充應用，其用途為提供查詢輸入、語意解析、資料比對及導引顯示等功能，藉以輔助使用者101於眾多設計素材中進行快速搜尋與導向。外掛程式115包含供電腦系統執行之一序列或一組指令，可由任一種或多種程式語言所撰寫，如 Java、C、C#、C++、Python 等，惟本創作並不對此加以限制。處理裝置12設置以從儲存裝置11載入外掛程式115，以實施下文將會介紹的一系列步驟。Plugin 115 is an extension application deployed within the design collaboration platform 105. Its purpose is to provide functions such as query input, semantic parsing, data matching, and navigation display, thereby assisting user 101 in quickly searching and navigating through a variety of design resources. Plugin 115 comprises a sequence or set of instructions for execution by a computer system and can be written in any one or more programming languages, such as Java, C, C#, C++, Python, etc., but this invention is not limited to such languages. Processing device 12 is configured to load plugin 115 from storage device 11 to implement the series of steps described below.
首先，處理裝置12接收使用者101輸入之查詢語句102。所述查詢語句102可透過輸入裝置以多種方式輸入，包括打字、手寫筆輸入、語音辨識輸入、觸控選單輸入等方式，惟本創作並不對此加以限制。使用者101可利用自然語言描述其所欲查詢之設計素材內容，例如「找出包含登入畫面的設計」或「尋找最近修改的結帳頁面」。查詢語句102在進入系統後，將作為搜尋需求之語意來源，後續供大型語言模型(large language model；LLM)103進行處理分析。First, the processing device 12 receives a query statement 102 input by a user 101. The query statement 102 can be input through an input device in a variety of ways, including typing, stylus input, voice recognition input, touch menu input, etc., but this invention is not limited to this. The user 101 can use natural language to describe the design material content they want to query, such as "find designs that include a login screen" or "find the most recently modified checkout page." After entering the system, the query statement 102 will serve as the semantic source of the search requirement and will be subsequently processed and analyzed by the large language model (LLM) 103.
接著，處理裝置12將查詢語句102輸入大型語言模型103，以取得其輸出之分析結果104。大型語言模型103可以是一種已於雲端平台上部署之預訓練語言模型，能夠解析並理解自然語言語句中的語意、上下文、結構與隱含意圖。舉例而言，所述大型語言模型103可以是GPT(Generative Pre-trained Transformer)系列模型、PaLM(Pathways Language Model)或LLaMA(Large Language Model Meta AI)等，惟本創作並不對此加以限制。查詢語句102會作為提示詞(prompt)傳遞至大型語言模型103，大型語言模型103則會對提示詞進行語意理解與語境分析，推論出該語句所欲尋找之設計內容關鍵元素，例如包含特定元件、具特定功能、或符合特定版面特徵者，並將該語意理解結果作為分析結果104輸出。Next, processing device 12 inputs query sentence 102 into large language model 103 to obtain its output analysis result 104. Large language model 103 can be a pre-trained language model deployed on a cloud platform that can parse and understand the semantics, context, structure, and implicit meaning of natural language sentences. For example, large language model 103 can be a GPT (Generative Pre-trained Transformer) series model, PaLM (Pathways Language Model), or LLaMA (Large Language Model Meta AI), but this invention is not limited to such models. The query statement 102 is passed as a prompt to the large language model 103. The large language model 103 performs semantic understanding and contextual analysis on the prompt to infer the key elements of the design content that the statement is looking for, such as those containing specific components, having specific functions, or meeting specific layout characteristics. The semantic understanding results are output as analysis results 104.
然後，處理裝置12根據分析結果104，自設計協作平台105擷取與分析結果104相對應之設計素材資訊106。具體而言，處理裝置12可根據分析結果104中所包含之語意特徵或語義關鍵詞，透過設計協作平台105所提供的應用程式介面(application program interface；API)，於設計協作平台105中執行查詢比對動作，辨識出一或多個符合查詢語句之設計素材。設計素材資訊106可包含與這些設計素材相關之識別資訊與資料位置資訊，後續將用以輔助使用者瀏覽與操作。當處理裝置12取得設計素材資訊106後，即可傳送至顯示裝置13，使顯示裝置13呈現所取得之設計素材資訊106，以供使用者檢視及選取。Then, based on the analysis result 104, the processing device 12 retrieves design material information 106 corresponding to the analysis result 104 from the design collaboration platform 105. Specifically, based on the semantic features or semantic keywords contained in the analysis result 104, the processing device 12 can perform a query comparison operation within the design collaboration platform 105 through the application program interface (API) provided by the design collaboration platform 105 to identify one or more design materials that match the query statement. The design material information 106 may include identification information and data location information related to these design materials, which will be used to assist users in browsing and operating the design material. After the processing device 12 obtains the design material information 106, it can be transmitted to the display device 13, so that the display device 13 presents the obtained design material information 106 for the user to view and select.
在一實施例中，大型語言模型103是經由一微調訓練(fine-tune training)程序所建立。所述微調訓練包含將多筆設計素材查詢語句及其對應之設計素材資料作為訓練資料，以及使用所述訓練資料進行微調訓練，以增強大型語言模型103對設計語意之理解與解析能力。訓練資料可由設計團隊或系統管理者事先整理，包含一系列使用者查詢語句與其實際對應之設計素材標註關聯，例如：「找出包含登入畫面的頁面」對應至特定畫面ID或素材描述欄位，並建立其對應關係。所述關聯資料可結構化儲存為一組(查詢語句，目標輸出)之對照項，用於監督式訓練(supervised learning)方式下進行語言模型之微調。經由上述程序微調後，大型語言模型103將更能針對設計相關語境進行語意解析，並提升其判斷使用者輸入內容背後意圖之準確性，進而輸出更貼近設計領域需求之分析結果104。所述微調訓練可由模型服務提供端於雲端平台或遠端伺服器中執行，使用者101可透過上傳訓練資料進行模型微調作業，惟本創作並不對其具體訓練實施方式加以限制。In one embodiment, the large language model 103 is built through a fine-tuning training process. The fine-tuning training includes using multiple design resource query statements and their corresponding design resource data as training data, and using the training data to perform fine-tuning training to enhance the large language model 103's ability to understand and parse design semantics. The training data can be pre-organized by the design team or system administrator and includes a series of user query statements associated with their actual corresponding design resource tags. For example, "Find pages containing login screens" is mapped to specific screen IDs or resource description fields, and their corresponding relationships are established. The associated data can be structured and stored as a set of (query statement, target output) mapping items for fine-tuning the language model under supervised learning. After fine-tuning through the above process, the large-scale language model 103 will be more capable of semantic parsing in design-related contexts and improve its accuracy in determining the intention behind user input content, thereby outputting analysis results 104 that are more closely aligned with design domain requirements. The fine-tuning training can be performed by the model service provider on a cloud platform or remote server. Users 101 can perform model fine-tuning operations by uploading training data, but this invention does not limit the specific training implementation method.
在一實施例中，設計素材資訊106可包含一或多筆設計素材資料，每筆設計素材資料又可包含一設計素材之一或多個代表縮圖，以及用以導向該設計素材於設計協作平台105中之對應位置的導引連結。In one embodiment, the design material information 106 may include one or more design material data, and each design material data may include one or more representative thumbnails of a design material, and a navigation link for navigating to the corresponding location of the design material in the design collaboration platform 105.
圖2是根據本創作之一實施例呈現前述外掛程式之一示例性的使用者介面20。如圖2所示，使用者介面20呈現了使用者輸入的查詢語句21，及其相應的設計素材資訊22。在本實施例中，當使用者輸入查詢語句21「我想要最新版本的置底導覽列」後，系統從該語句所分析出的語意特徵，從設計協作平台中擷取出相關之設計素材資訊22，並將該些資訊呈現在使用者介面20中。所述設計素材資訊22包含一或多筆設計素材資料，例如設計素材資料221與設計素材資料222。每筆設計素材資料可包含一導引連結，例如設計素材資料221與設計素材資料222分別包含導引連結2211與導引連結2221，用以導向設計素材於設計協作平台中之對應位置。每筆設計素材資料可更包含設計素材之一或多個代表縮圖，例如設計素材資料221與設計素材資料222分別包含代表縮圖2212與代表縮圖2222，以提供使用者視覺化辨識之依據。如此配置可輔助使用者在短時間內大致理解各素材之內容與用途，並可透過點擊導引連結進一步檢視各素材之具體細節。Figure 2 illustrates an exemplary user interface 20 for the aforementioned plug-in, according to an embodiment of this invention. As shown in Figure 2, user interface 20 displays a query 21 entered by a user and corresponding design material information 22. In this embodiment, when a user enters query 21, "I want the latest version of the bottom navigation bar," the system analyzes the semantic features of the query, extracts relevant design material information 22 from the design collaboration platform, and displays this information in user interface 20. The design material information 22 includes one or more pieces of design material data, such as design material data 221 and design material data 222. Each design material data entry may include a guide link, for example, design material data 221 and design material data 222 include guide link 2211 and guide link 2221, respectively, to direct users to the corresponding location of the design material in the design collaboration platform. Each design material data entry may also include one or more representative thumbnails of the design material, for example, design material data 221 and design material data 222 include representative thumbnails 2212 and representative thumbnails 2222, respectively, to provide users with a visual basis for identification. This configuration helps users quickly understand the content and purpose of each material, and allows users to further view specific details of each material by clicking on the guide link.
在一實施例中，處理裝置12更在接收查詢語句102時，根據使用者101輸入之部分字元，自動使顯示裝置13顯示一或多個與該部分字元相關之候選關鍵字，供使用者選擇以補全查詢語句102。In one embodiment, when receiving the query statement 102, the processing device 12 automatically causes the display device 13 to display one or more candidate keywords related to the partial characters input by the user 101 for the user to select to complete the query statement 102.
圖3是根據本創作之一實施例呈現前述外掛程式之另一示例性的使用者介面30。如圖3所示，使用者介面30包含一查詢輸入欄301，供使用者輸入查詢語句內容。例如，使用者可輸入「我要找各種/pa」，系統可即時解析該輸入內容並進行提示建議。其中「/」用作指令觸發符號(trigger character)，其作用為提示系統於該處開始進行候選關鍵字之自動補全。系統偵測到「/」後，將其後續輸入的字元(如「pa」)視為關鍵字前綴，並於查詢輸入欄301上方呈現候選關鍵字提示區塊302，顯示與該部分字元相關之候選關鍵字，例如「Panel」、「Pagination」、「Password」等。使用者可點選所需項目以快速補全查詢語句，提升操作效率並降低輸入成本。應理解，「/」僅為指令觸發符號之一示例，在其他實施例中，亦可採用其他特殊字元(例如「#」、「@」或「$」)作為指令觸發符號，惟本創作並不對此加以限制。FIG3 illustrates another exemplary user interface 30 of the aforementioned plug-in, according to an embodiment of the present invention. As shown in FIG3 , the user interface 30 includes a query input field 301 for the user to enter a query. For example, a user can enter "I want to find various /pa" and the system will instantly parse the input and provide suggestions. The "/" character serves as a trigger character, prompting the system to begin auto-completion of candidate keywords at that point. After the system detects "/", it regards the characters entered subsequently (such as "pa") as keyword prefixes, and displays a candidate keyword prompt block 302 above the query input field 301, showing candidate keywords related to the characters, such as "Panel", "Pagination", "Password", etc. The user can click on the required item to quickly complete the query statement, improve operational efficiency and reduce input costs. It should be understood that "/" is only an example of a command trigger symbol. In other embodiments, other special characters (such as "#", "@" or "$") can also be used as command trigger symbols, but this invention is not limited to this.
此外，使用者介面30亦包含排序條件欄位304與篩選條件欄位305。排序條件欄位304可供使用者選擇查詢結果之呈現順序，例如依據素材之編輯者、編輯日期、使用頻率或被查詢頻率進行排序。篩選條件欄位305則可用以限定查詢結果之時間範圍，例如僅顯示最近7天內、最近一個月內或最近一年內之素材。上述條件欄位可透過下拉式選單或其他互動元件實作，惟本創作並不對其具體實作方式加以限制。In addition, user interface 30 also includes a sorting condition field 304 and a filtering condition field 305. Sorting condition field 304 allows the user to select the order in which the query results are presented, for example, by editor, edit date, frequency of use, or frequency of query. Filtering condition field 305 can be used to limit the time range of the query results, for example, only displaying materials from the last seven days, the last month, or the last year. These condition fields can be implemented through drop-down menus or other interactive elements, but this invention does not limit their specific implementation methods.
最後，使用者介面30包含一提交查詢按鈕306，使用者於完成查詢語句輸入及條件選擇後，可點擊該提交查詢按鈕306以正式送出查詢請求。該查詢請求將由系統進行語意解析並返回對應之設計素材資訊，進一步於顯示裝置上呈現。Finally, the user interface 30 includes a submit query button 306. After completing the query statement input and condition selection, the user can click the submit query button 306 to formally submit the query request. The query request will be semantically parsed by the system and the corresponding design material information will be returned, which will be further presented on the display device.
在一實施例中，前述候選關鍵字的擷取來源可以包含設計協作平台105中儲存的設計素材之文字資訊、使用者101的歷史查詢紀錄，及/或預先定義之共用詞庫。In one embodiment, the aforementioned candidate keyword extraction sources may include text information of design materials stored in the design collaboration platform 105, historical query records of the user 101, and/or a predefined shared vocabulary.
在其中一種情境下，處理裝置12可分析設計協作平台105中各設計素材之屬性欄位，擷取與文字描述相關之資訊，例如圖層名稱、元件命名、註解文字、標籤或樣式標題等。這些資訊通常由設計師於創建素材時自行命名或標註，具備高度語意指標性，因此可用以作為候選關鍵字之來源。舉例而言，若某素材元件命名為「Primary Button」或「Error Modal」，則系統可擷取「Button」與「Modal」作為提示字詞，協助其他使用者進行快速查詢。In one scenario, the processing device 12 can analyze the property fields of each design asset in the design collaboration platform 105 and extract information related to the textual description, such as layer names, component names, annotations, tags, or style titles. This information is typically named or annotated by the designer when creating the asset and is highly semantically specific, making it a useful source of candidate keywords. For example, if a component is named "Primary Button" or "Error Modal," the system can extract "Button" and "Modal" as prompts to help other users quickly search for the item.
在另一種情境下，處理裝置12亦可記錄使用者101之歷史查詢紀錄，包含其曾經輸入過的查詢語句、選取之候選關鍵字、實際點擊之素材結果等互動資訊。系統可根據這些紀錄統計使用頻率較高或近期曾用之關鍵字，並於使用者再次輸入查詢語句時作為提示依據，提升使用者體驗並強化查詢建議的個人化程度。In another scenario, processing device 12 may also record user 101's historical query history, including query phrases entered, candidate keywords selected, and actual clicked results, among other interaction information. Based on these records, the system can compile statistics on frequently used or recently used keywords and use them as prompts when the user enters a query again, thereby enhancing the user experience and personalizing query suggestions.
再者，候選關鍵字亦可來自一預先定義之共用詞庫，該詞庫可由系統管理者或設計團隊預先建立，內容可包含特定產品、功能模組、常用元件名稱、品牌術語或企業內部設計用語等。此類詞庫可依照不同產品別或專案類型進行配置，使得系統能提供更具上下文語意之候選關鍵字建議，並強化團隊設計溝通與命名一致性。Furthermore, keyword candidates can be sourced from a predefined shared vocabulary. This vocabulary can be created by the system administrator or design team and may include specific products, functional modules, commonly used component names, brand terms, or internal corporate design terminology. This vocabulary can be configured based on different product or project types, allowing the system to provide more contextual keyword candidate suggestions and strengthen team design communication and naming consistency.
本創作所提供之設計素材搜尋系統，透過大型語言模型支援以自然語言描述的查詢語句進行設計素材搜尋，可協助設計師或產設計團隊在設計協作平台中更輕鬆有效且準確地找到不同版本的設計素材，從而促進跨部門溝通協調、專案追蹤和管理設計素材。The design resource search system provided by this project uses a large language model to support search queries expressed in natural language. This system can help designers or production design teams more easily, effectively, and accurately find different versions of design resources within a design collaboration platform, thereby facilitating cross-departmental communication and coordination, project tracking, and design resource management.
以上段落採用多種態樣作敘述。顯然地，本文之教示可以多種方式實現，而在範例中所揭露之任何特定架構或功能僅是一種代表性的情況。根據本文之教示，本領域應理解，可獨立實作本文所揭露之各個態樣，或者合併實作兩種以上之態樣。The above paragraphs describe various aspects. Obviously, the teachings herein can be implemented in a variety of ways, and any specific architecture or functionality disclosed in the examples is merely representative. Based on the teachings herein, it should be understood by those skilled in the art that each aspect disclosed herein can be implemented independently, or two or more aspects can be combined.
雖然本創作已以實施例敘述如上，然其並非用以限定本創作，任何熟習此技藝者，在不脫離本創作之精神和範圍內，當可作些許之更動與潤飾，因此新型之保護範圍當視後附之申請專利範圍所界定者為準。Although this invention has been described above with reference to embodiments, this is not intended to limit the invention. Anyone skilled in the art may make minor changes and modifications without departing from the spirit and scope of this invention. Therefore, the scope of protection of the new invention shall be determined by the scope of the patent application attached hereto.
10:設計素材搜尋系統11:儲存裝置12:處理裝置13:顯示裝置101:使用者102:查詢語句103:大型語言模型104:分析結果105:設計協作平台106:設計素材資訊115:外掛程式20:使用者介面21:查詢語句22:設計素材資訊221,222:設計素材資料2211,2221:導引連結2212,2222:代表縮圖30:使用者介面301:查詢輸入欄302:候選關鍵字提示區塊304:排序條件欄位305:篩選條件欄位306:提交查詢按鈕10: Design Material Search System 11: Storage Device 12: Processing Device 13: Display Device 101: User 102: Query Statement 103: Large Language Model 104: Analysis Results 105: Design Collaboration Platform 106: Design Material Information 115: Plug-in 20: User Interface 21: Query Statement 22: Design Material Information 221, 222: Design Material Data 2211, 2221: Navigation Links 2212, 2222: Representative Thumbnails 30: User Interface 301: Query Input Field 302: Keyword Prompt Area 304: Sorting Criteria Field 305: Filter Criteria Field 306: Submit Query Button
本創作將可從以下示範的實施例之敘述搭配附帶的圖式更佳地理解。此外，應理解的是，在本創作之流程圖中，各區塊的執行順序可被改變，且/或某些區塊可被改變、刪減或合併。圖1是根據本創作之一實施例的一種設計素材搜尋系統之系統架構圖。圖2是根據本創作之一實施例呈現前述外掛程式之一示例性的使用者介面。圖3是根據本創作之一實施例呈現前述外掛程式之另一示例性的使用者介面。This invention can be better understood through the following description of exemplary embodiments, accompanied by accompanying diagrams. Furthermore, it should be understood that in the flowcharts of this invention, the execution order of each block can be changed, and/or certain blocks can be changed, deleted, or merged. Figure 1 is a system architecture diagram of a design material search system according to one embodiment of this invention. Figure 2 is an exemplary user interface for the aforementioned plug-in according to one embodiment of this invention. Figure 3 is another exemplary user interface for the aforementioned plug-in according to one embodiment of this invention.
10:設計素材搜尋系統10: Design material search system
11:儲存裝置11: Storage device
12:處理裝置12: Processing device
13:顯示裝置13: Display device
101:使用者101: User
102:查詢語句102: Query statement
103:大型語言模型103: Large Language Models
104:分析結果104: Analysis Results
105:設計協作平台105: Design Collaboration Site
106:設計素材資訊106: Design material information
115:外掛程式115: Plugins
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW114204195UTWM672805U (en) | 2025-04-28 | 2025-04-28 | Design asset search system |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW114204195UTWM672805U (en) | 2025-04-28 | 2025-04-28 | Design asset search system |
| Publication Number | Publication Date |
|---|---|
| TWM672805Utrue TWM672805U (en) | 2025-07-11 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW114204195UTWM672805U (en) | 2025-04-28 | 2025-04-28 | Design asset search system |
| Country | Link |
|---|---|
| TW (1) | TWM672805U (en) |
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | In-ide code generation from natural language: Promise and challenges | |
| Narechania et al. | NL4DV: A toolkit for generating analytic specifications for data visualization from natural language queries | |
| US10896214B2 (en) | Artificial intelligence based-document processing | |
| US11520800B2 (en) | Extensible data transformations | |
| Treude et al. | Extracting development tasks to navigate software documentation | |
| US12386849B2 (en) | Efficient transformation program generation | |
| US9009664B2 (en) | Structural search of source code | |
| US12298996B2 (en) | Facilitating data transformations | |
| RU2696305C2 (en) | Browsing images through intellectually analyzed hyperlinked fragments of text | |
| US9613003B1 (en) | Identifying topics in a digital work | |
| US20120278302A1 (en) | Multilingual search for transliterated content | |
| Kelley et al. | A framework for creating knowledge graphs of scientific software metadata | |
| US20120030553A1 (en) | Methods and systems for annotating web pages and managing annotations and annotated web pages | |
| US20110184960A1 (en) | Methods and systems for content recommendation based on electronic document annotation | |
| US20130305149A1 (en) | Document reader and system for extraction of structural and semantic information from documents | |
| US11170020B2 (en) | Collecting and annotating transformation tools for use in generating transformation programs | |
| US9182947B2 (en) | Program source code navigation | |
| US20180096067A1 (en) | Creation and optimization of resource contents | |
| US11163788B2 (en) | Generating and ranking transformation programs | |
| US9411880B2 (en) | System and method for dynamically configuring content-driven relationships among data elements | |
| Konda | Elasticsearch in action | |
| Kumar | Apache Solr search patterns | |
| US20160086499A1 (en) | Knowledge brokering and knowledge campaigns | |
| US20160085850A1 (en) | Knowledge brokering and knowledge campaigns | |
| US20240184798A1 (en) | Extensible data transformations |