Intelligent translation method and device for foreign language of printed circuit board and terminal equipment

Abstract

The invention relates to the technical field of computers, and discloses an intelligent translation method and device for foreign languages of a printed circuit board and terminal equipment. The method comprises the following steps: acquiring foreign language document data containing professional vocabularies of the printed circuit board, wherein the foreign language document data contains at least one target language except a specified language; for each foreign language sentence in the foreign language document data, converting the foreign language sentence into structured text data based on the format type of the foreign language document data; extracting a plurality of foreign language reference words matched with the structured text data from a printed circuit board language library; and translating the plurality of foreign language reference words through the foreign language translation model to obtain a translated text of the specified language. By adopting the method, a small language or multilingual mixed scene can be flexibly handled, and the problems of information loss and structure disorder caused by a complex document format in traditional translation are solved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an intelligent translation method, apparatus, and terminal device for foreign text on printed circuit boards. Background Technology

[0002] The PCB (Printed Circuit Board) industry is currently characterized by frequent global collaboration, with a large number of design requirements, specifications, and process documents presented in foreign languages ​​such as English, Japanese, and German. These documents contain PCB-specific terminology, complex parameter formats, and contextual information, directly impacting the accuracy of the design and manufacturing processes.

[0003] However, traditional machine translation (such as general-purpose translation software) lacks specialized corpora in the PCB field, resulting in low accuracy in terminology translation and frequent "literal translation errors," such as mistranslating "stiffener" (a "reinforcing plate" in PCB) as "reinforcing material." Furthermore, human translation relies on professionals, leading to inefficiency and high costs, and is ill-suited for urgent needs or large-scale document translation scenarios. In addition, differences in translation styles among individuals can result in inconsistent terminology. Therefore, there is an urgent need for an intelligent translation method that deeply integrates artificial intelligence technology and is tailored to specialized PCB scenarios. Summary of the Invention

[0004] In view of this, embodiments of this application provide an intelligent translation method, apparatus, and terminal device for foreign language text in printed circuit boards, which can effectively solve the problem of low accuracy in translating professional terms in the PCB field.

[0005] In a first aspect, embodiments of this application provide an intelligent translation method for foreign text on printed circuit boards, comprising: Acquire foreign language document data containing printed circuit board terminology, wherein the foreign language document data contains at least one target language other than the specified language; For each foreign language sentence in the foreign language document data, the foreign language sentence is converted into structured text data based on the format type of the foreign language document data; Multiple foreign language reference words that match the structured text data are extracted from the printed circuit board language library through vector similarity retrieval; The foreign language translation model translates multiple foreign language reference words based on the contextual relationships between them, thereby obtaining the translated text in the specified language.

[0006] Secondly, embodiments of this application provide an intelligent translation device for foreign text on a printed circuit board, comprising: The acquisition module is used to acquire foreign language document data containing printed circuit board terminology, wherein the foreign language document data contains at least one language other than the target language; The conversion module is used to convert each foreign language sentence in the foreign language document data into structured text data based on the format type of the foreign language document data. The extraction module is used to extract multiple foreign language reference words that are similar to the structured text data from the printed circuit board language library through vector similarity retrieval; The translation module is used to translate multiple foreign language reference words based on the contextual relationships between the foreign language reference words using a foreign language translation model, so as to obtain the translated text in the specified language.

[0007] Thirdly, embodiments of this application provide a terminal device, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described intelligent translation method for foreign text on printed circuit boards.

[0008] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed on a processor, implements the aforementioned intelligent translation method for foreign text on printed circuit boards.

[0009] The embodiments of this application have the following beneficial effects: This application constructs a universal input interface by defining the "target language" as any unspecified language, enabling it to handle multiple languages ​​and enhancing its applicability and practicality. Furthermore, it addresses the issues of information loss and structural errors caused by complex document formats in traditional translation. Simultaneously, utilizing vectorized retrieval, document understanding, and large language models from artificial intelligence technologies, it extracts multiple foreign language reference words matching structured text data from a printed circuit board language database. Then, through a constructed foreign language translation model, it translates these multiple foreign language reference words. In summary, this application, by deeply integrating artificial intelligence technology into the PCB field, constructs a knowledge-enhanced and automation-driven intelligent translation method for a professional domain. This is not only reflected in a significant improvement in translation efficiency and accuracy but also in fully demonstrating the professional capabilities of artificial intelligence technology at the understanding, reasoning, and generation levels. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This diagram illustrates an application environment for an intelligent translation method of foreign text on a printed circuit board according to an embodiment of this application. Figure 2 This paper illustrates a flowchart of an intelligent translation method for foreign text on a printed circuit board according to an embodiment of this application. Figure 3 This paper illustrates a schematic diagram of a framework for an intelligent translation structure of foreign text on a printed circuit board according to an embodiment of this application. Detailed Implementation

[0012] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0013] The components of the embodiments of this application described and illustrated in the accompanying drawings can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0014] In the following text, the terms "comprising," "having," and their cognates, which may be used in various embodiments of this application, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more combinations thereof. Furthermore, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0015] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this application pertain. Terms (such as those defined in a generally used dictionary) shall be interpreted as having the same meaning as in the context of the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this application.

[0016] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0017] The following describes the intelligent translation method, apparatus, and terminal equipment for foreign text on printed circuit boards using specific embodiments.

[0018] The intelligent translation method for foreign text on printed circuit boards provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed in the cloud or on other network servers. Terminal 102 generates intelligent translation requests for foreign language text on printed circuit boards and sends these requests to server 104, enabling server 104 to translate multiple foreign language reference words using a foreign language translation model to obtain translated text in the specified language. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0019] Figure 2 A flowchart illustrating an intelligent translation method for foreign text on printed circuit boards according to an embodiment of this application is shown. Exemplarily, this intelligent translation method for foreign text on printed circuit boards includes the following steps: Step S202: Obtain foreign language document data containing printed circuit board terminology. The foreign language document data contains at least one target language other than the specified language.

[0020] Among them, professional terms are terms and parameters with specific technical meanings that are widely used in the design, manufacturing, testing and process flow of printed circuit boards (PCBs), including but not limited to: structural terms: such as blind vias; performance parameter terms: such as impedance; and process terms: such as drilling.

[0021] Foreign language document data refers to technical electronic documents written in a non-Chinese language and containing PCB professional terminology.

[0022] The specified language refers to the language in which the user expects the translation results to be in. For example, Simplified Chinese is used to output professional translations that conform to the reading habits of Chinese engineers. This specified language serves as the unified output target of the translation system, ensuring consistency in terminology and standardization of expression, and is suitable for domestic PCB design review, production communication, quality control, and other application scenarios.

[0023] The target language refers to the language used in the original foreign language document, and is not any language that falls under the "specified language" category. Examples include Japanese and English.

[0024] Specifically, the system first receives an external input request, whereby the user uploads the foreign language document data to a pre-defined directory of documents to be processed via a web interface. The foreign language document data undergoes preliminary analysis to identify its basic attributes, including the specified language and the target language. For example, if the specified language is pre-configured as Chinese, the foreign language document data is confirmed to meet the condition of "containing the target language (German) but not the specified language."

[0025] Then, the foreign language document data is marked as "verified PCB foreign language document", the path, language type, and format category of the foreign language document data are recorded, and the foreign language document data is added to the task queue, ready to enter the next step "structured parsing based on format type".

[0026] Step S204: For each foreign language sentence in the foreign language document data, convert the foreign language sentence into structured text data based on the format type of the foreign language document data.

[0027] In this context, a foreign language sentence refers to the smallest linguistic unit that expresses complete semantic or technical information in foreign language document data. Examples include technical phrases, natural language statements, field-value pairs in tables, and non-paragraph text fragments such as figure captions, footnotes, and annotations.

[0028] Format type refers to the electronic file format and content organization characteristics used in foreign language documents. It is mainly divided into two categories: editable format and non-editable format.

[0029] Editable formats refer to file types where document content is stored in native text or structured tags, allowing programs to directly extract text and layout information, such as DOC, PPT, and Excel formats.

[0030] Structured text data refers to intermediate representation data generated after format parsing, which retains the semantic structure and spatial logical relationships of the original document.

[0031] In one embodiment, a document parsing component extracts first text from foreign language document data; an image recognition component extracts second text from an image embedded in the foreign language document data and adds the second text to the corresponding position of the image embedded in the foreign language document data; and structured text data is obtained based on the first text and the second text.

[0032] Among them, the document parsing component refers to the software module or tool library used to extract the original text content and basic structural information from editable foreign language documents.

[0033] The first text refers to the raw text content extracted from the non-image areas of an editable document using a document parsing component.

[0034] Image recognition components refer to technical modules used to identify and extract text content from images (such as screenshots, circuit diagrams, process flow diagrams, and scanned illustrations) embedded in documents.

[0035] The second text refers to the text content that is embedded in an image from a foreign language document and identified through an image recognition component.

[0036] Specifically, for editable files such as DOC, PPT, and Excel, the document parsing component (document parsing library) is used to directly extract the text from the foreign language document data while preserving the original structure (titles, lists, etc.), resulting in the first text. If the foreign language document data embeds images, the image recognition component (OCR engine, Optical Character Recognition) is used to recognize the text in the image, and the recognized text is then appended to the end of the document. Finally, the format is converted to obtain the second text.

[0037] Finally, the first and second texts are merged to generate structured text data. For example, the "second text" is inserted into the context near its location in the image. After all pages are processed, unified structured text data is generated, containing complete hierarchical structure, parameter items, and image-text relationship information, which serves as the basic input for the next stage of term matching.

[0038] Through the above embodiments, the "first text" and "second text" are integrated according to spatial logic to form a unified semantic flow, breaking through the technical bottleneck of the separation between the "text layer" and the "image layer" in traditional document processing.

[0039] In one embodiment, a layout analysis model is used to perform region detection on foreign language document data, identifying the spatial bounding boxes and category labels of each region. Each region includes at least a title region, a body paragraph region, a table region, and a circuit icon region. For non-table regions, an image recognition component is used to extract text from each text block in the non-table region to obtain the original non-table text. Based on the position of each non-table region and a first preset reading order, the original non-table text is converted into third text, which is then used as structured text data.

[0040] Non-editable formats refer to electronic file formats in which foreign language document data is primarily stored as images or a mixture of vector and image elements, and whose complete semantic content cannot be directly extracted using conventional text extraction tools. For example, the PDF format.

[0041] Page layout analysis model refers to a computer vision model based on deep learning, used to perform region segmentation and functional classification of foreign language document images, identify the spatial location and category attributes of different semantic regions, and output the bounding box coordinates and label type of each region.

[0042] Each region refers to a group of continuous spatial blocks with the same semantic function that are identified during the layout analysis process, including at least the following four categories: title region; body paragraph region, table region, and circuit icon region, etc.

[0043] A spatial bounding box refers to the physical location of a region on a page, represented by a rectangular coordinate system. It is used to identify the position of each text block or region, facilitating sorting and contextual association.

[0044] Category labels are functional semantic identifiers assigned by the layout analysis model to each identified region, indicating the nature of the content in that region.

[0045] Non-table areas refer to all other areas that are not considered "table areas" after being identified by the layout analysis model. These mainly include: title areas; body paragraph areas; circuit diagram areas; and auxiliary information areas such as figure captions, footnotes, headers, and footers.

[0046] A text block refers to the smallest continuous unit of text that is recognized by an OCR engine within a region. It usually corresponds to an independently typed text segment and has a unique spatial bounding box.

[0047] Non-table raw text refers to the unsorted and unintegrated raw string collection obtained by performing OCR engine recognition on each text block in the non-table area, retaining the recognition result and spatial coordinate information of each text block.

[0048] The first presupposed reading order refers to the rule of rearranging multiple blocks of text in non-table areas according to natural human reading habits. For example, from top to bottom, and from left to right within the same horizontal level.

[0049] The third text refers to the structured paragraph text with a logical order generated by integrating the original non-table text according to the first preset reading order, thus removing the influence of disordered, repetitive, or isolated text blocks.

[0050] Specifically, the layout analysis model is loaded, inference is performed on the foreign language document data, and the spatial bounding boxes and category labels of each area of ​​the non-table area (i.e., including the title, body paragraphs and circuit icon area) are output.

[0051] For each non-table region image sub-block, the OCR engine is invoked to identify the corresponding text block and coordinates, and text extraction is performed to obtain the original non-table text. According to the first preset reading order and coordinates, the text blocks are sorted and merged, and finally the sorted and merged text blocks are converted into structured output, i.e., structured text data.

[0052] Through the above embodiments, a leap from "images" to "understandable semantic structures" has been achieved for the large number of scanned technical documents in the PCB industry.

[0053] In one embodiment, for a table area, the row and column structure of the table area is restored by a table structure recognition algorithm; based on the row and column structure, text is extracted from each text block in the table area to obtain the original table text; according to the row and column structure and a second preset reading order, the original table text is converted into fourth text; based on the third and fourth texts, structured text data is obtained.

[0054] In this context, a table area refers to a data block in foreign language document data that has a clear row and column structure and is used to organize parameterized information.

[0055] Table structure recognition algorithm refers to a technical method based on computer vision and deep learning, used to recover the original row and column logical structure of a table area in the form of an image. It includes: detecting table border lines, dashed lines or implicit separators; determining cell boundaries; reconstructing the correspondence between row headers, data rows and their columns; and outputting in a structured format.

[0056] The original text of the table refers to the unintegrated string set obtained by performing text extraction on each cell after the table structure recognition is completed. It retains the spatial position of each text block and its mapping relationship with the row and column to which it belongs.

[0057] The second pre-defined reading order refers to the rules for linearly reorganizing the original text of the table according to standard table reading habits. For example, row first, then column, and within each row, arranged in ascending order of index (i.e., from left to right).

[0058] The fourth text refers to the target text generated by structurally reorganizing the original table text according to the second preset reading order, which conforms to the expression habits of natural language.

[0059] Specifically, for the table area, the algorithm analyzes the line distribution in the image using a table structure recognition algorithm to identify the number of columns and rows (including the table header); determines that each horizontal line is a row segment and each vertical line is a column segment; constructs a logical grid matrix to divide the image into multiple cells; and outputs the row and column structure metadata.

[0060] The OCR engine is called for each cell sub-image to obtain the recognition result. The result is then linearized according to the second preset reading order (rows first, columns second). For example, each row is traversed, field values ​​are concatenated from left to right, and semantic prompts are added to obtain the fourth text. Finally, this "fourth text" is merged with the "third text" (content from non-table areas) generated in the previous steps to form complete structured text data for subsequent terminology matching and translation.

[0061] In the above embodiments, a table processing paradigm of "structure first, semantic follow" is adopted to first restore the logical structure, then extract the content, and finally output it linearly according to rules. Specifically targeting the large number of parametric table documents in the PCB industry, it realizes the leap from "image table" to "machine-understandable structure", improving the accuracy of automated processing of professional documents.

[0062] Step S206: Extract multiple foreign language reference words that match the structured text data from the printed circuit board language library by means of vector similarity retrieval.

[0063] The Printed Circuit Board Language Library refers to a specially constructed multilingual professional knowledge base for the printed circuit board field, which contains a large amount of cleaned, labeled and vectorized bilingual or multilingual text data.

[0064] Foreign language reference terms refer to one or more manually annotated English (or other target language) words or phrases with specific technical meanings that are associated with a reference text vector in the printed circuit board language library. Their purpose is to guide the translation model to correctly understand and output the corresponding translation in the specified language.

[0065] In one embodiment, structured text data is converted into target text vectors; for each reference text vector in the printed circuit board language library, the similarity between the target text vector and the target reference text vector is calculated; based on the similarity between each reference text vector and the target text vector, multiple target reference text vectors and foreign language reference words corresponding to each target reference text vector are determined.

[0066] The target text vector refers to the representation of a processing unit (such as a sentence, paragraph, or page content) in the current structured text data to be translated, which is converted into a high-dimensional numerical vector through an embedding model and used to express the technical meaning of the text in the semantic space.

[0067] Reference text vectors refer to high-dimensional numerical vectors obtained by encoding professional corpus fragments (such as terminology definitions, parameter specifications, contextual examples, etc.) pre-extracted and stored from the printed circuit board language library using the same embedding model. Each reference text vector corresponds to one or more knowledge units that can assist in translation.

[0068] Specifically, a structured PCB-specific corpus is first constructed. This corpus contains a large amount of manually annotated bilingual data. An embedding model is then used to convert each text segment (which can be a single term, phrase, or sentence with context) in the corpus into a high-dimensional numerical vector (i.e., a "reference text vector"). These vectors, along with their original text and corresponding technical terms, are stored in a vector database, forming a "Printed Circuit Board Language Library."

[0069] Then, in the translation process, upon receiving a user-uploaded foreign language document containing both editable and non-editable formats, the document is first parsed into structured plain text using an OCR engine and layout analysis model. Assuming the translation task is a key paragraph from the foreign language document, the same embedding model used when building the printed circuit board language library is invoked to convert the paragraph text into a high-dimensional numerical vector, which is the target text vector.

[0070] Finally, the "target text vector" is input into the printed circuit board language library, and an approximate nearest neighbor search is performed. The cosine similarity between the "target text vector" and each "reference text vector" is calculated. A relevance threshold is set, and the number of returned results is limited. Ultimately, a set of "target reference text vectors" with the highest similarity is retrieved. Based on the preset mapping relationship, the original text corresponding to these "target reference text vectors" and their key terms (i.e., foreign language reference words) are obtained.

[0071] Through the above embodiments, the "vectorization + similarity retrieval" mechanism can quickly and accurately filter out the most relevant information for the current translation task from a massive knowledge base. This ensures that the contextual information injected into the model is highly relevant and accurate.

[0072] Step S208: Based on the contextual relationships between the foreign language reference words, the foreign language translation model is used to translate multiple foreign language reference words to obtain the translated text in the specified language.

[0073] The foreign language translation model refers to a core algorithm component based on artificial intelligence technology used to perform the conversion from a source language (such as English, Japanese, or German) to a target language (a specified language, such as Chinese). In this application, the model specifically refers to a large language model that has undergone domain adaptation or is enhanced by external knowledge. Its core function is to generate accurate translations that conform to the professional standards and technical logic of the PCB field after receiving structured contextual information.

[0074] Translated text refers to the translation result in a specified language (such as Chinese) corresponding to the source language content, generated by a foreign language translation model based on input enhanced prompts (including foreign language reference words, contextual information, etc.).

[0075] In one embodiment, based on preset splitting conditions, the foreign language document data is split into multiple sub-documents; for each sub-document, the foreign language reference words corresponding to the sub-document are translated using a foreign language translation model to obtain the translated text in the target language.

[0076] Among them, the preset splitting conditions refer to the rules set in advance to balance translation quality, system performance and resource consumption when performing large-scale document translation. They are used to guide how to divide complete foreign language document data into smaller and more manageable independent processing units (i.e., sub-documents).

[0077] A subdocument is an independent processed segment that is separated from the original foreign language document data according to "preset splitting conditions" and has a complete semantic or structural unit. It is the basic operation object of the system when performing translation tasks.

[0078] Specifically, the backend is configured with preset splitting conditions, such as splitting by physical pages, with every 5 consecutive pages forming a sub-document. The foreign language document data is read and automatically split into multiple sub-documents according to the preset splitting conditions. Each sub-document corresponds to 5 pages of the original text. Then, a batch processing workflow is started to translate the multiple sub-documents using a foreign language translation model.

[0079] In one embodiment, the structured text data is divided into multiple translation processing units according to a preset granularity condition, and each translation processing unit corresponds to a page or paragraph; for multiple translation processing units, the historical translation text obtained by the previous translation processing unit of the target translation processing unit after being translated by the foreign language translation model is obtained; the historical translation text and the foreign language reference words corresponding to the target translation processing unit are input into the foreign language translation model to obtain the translation text of the target language.

[0080] The preset granularity condition refers to the rules or standards set in advance to divide "structured text data" into "units to be translated" when performing continuous contextual translation. This condition determines the fineness and scope of context inheritance.

[0081] A translation processing unit refers to the smallest working unit with independent processing boundaries, which is divided from structured text data according to "preset granularity conditions". It is the basic object for the system to perform the operations of "acquiring historical translation text" and "injecting context".

[0082] Historical translation texts refer to the translated texts in a specific language (such as Chinese) that have been successfully translated and confirmed during the translation process and belong to the "previous unit to be translated." They are actively used as a contextual reference for generating the current unit's translation to ensure terminological consistency and semantic coherence.

[0083] Specifically, based on preset granularity conditions, such as "each unit to be translated corresponds to one page", the parsed structured text data is divided into multiple "units to be translated".

[0084] Processing the first page: This is the first processing unit, therefore there is no "previous" unit. Skip the "retrieve historical translation text" step, directly retrieve the foreign language reference words for the first page, and input them into the foreign language translation model. The foreign language translation model generates and stores the translated text for the first page.

[0085] Processing the second page: Identify the previous unit to be translated as the first unit to be translated and obtain the historical translation text corresponding to that first unit. Then, retrieve the foreign language reference words for the second page. Input the historical translation text and the foreign language reference words from the second page into the foreign language translation model. Under the influence of the context of the historical translation text, the foreign language translation model generates and stores the semantically coherent translation text for the first page.

[0086] Processing subsequent units (pages 3 to N): This process continues sequentially, processing each unit in turn. When translating page N, the translated text from page N-1 is always retrieved first as the historical translation text, and this text, along with the foreign language reference words from page N, is input into the foreign language translation model. After all units to be translated have been completed, the translated texts from page 1 to page N are merged sequentially to form a complete Chinese translation with consistent terminology and fluent writing—that is, the translation text in the specified language.

[0087] Through the above embodiments, a powerful state memory capability is established by forcibly inheriting the "historical translation text." Once a term is accurately translated early in the document, its translation will be used throughout the entire document. This eliminates the need for tedious terminology verification in the final output, allowing it to directly meet deliverable standards.

[0088] Figure 3 A schematic diagram of a printed circuit board (PCB) foreign language translation device 300 according to an embodiment of this application is shown. Exemplarily, the PCB foreign language translation device 300 includes: The acquisition module 302 is used to acquire foreign language document data containing printed circuit board terminology, wherein the foreign language document data contains at least one target language other than the specified language; The conversion module 304 is used to convert each foreign language sentence in the foreign language document data into structured text data based on the format type of the foreign language document data. Extraction module 306 is used to extract multiple foreign language reference words that match the structured text data from the printed circuit board language library; Translation module 308 is used to translate multiple foreign language reference words using a foreign language translation model to obtain translated text in the specified language.

[0089] It is understood that the device in this embodiment corresponds to the intelligent translation method for foreign text on printed circuit boards in the above embodiments. The options in the above embodiments are also applicable to this embodiment, so they will not be described again here.

[0090] This application also provides a terminal device, exemplary of which includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to enable the terminal device to perform the above-described method or the functions of the various modules in the above-described intelligent translation device for foreign text on printed circuit boards.

[0091] The processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including at least one of a Central Processing Unit (CPU), Graphics Processing Unit (GPU), Network Processor (NP), Digital Signal Processor (DSP), Application-Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0092] Memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), and Electrically Erasable Programmable Read-Only Memory (EEPROM). Memory is used to store computer programs, and the processor can execute these programs upon receiving execution instructions.

[0093] This application also provides a computer-readable storage medium for storing computer programs used in the aforementioned terminal devices. For example, the computer-readable storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0094] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that, as an alternative implementation, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0095] In addition, the functional modules or units in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0096] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a smartphone, personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0097] The above description is merely a specific embodiment of this application, but the scope of protection of this application 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 this application should be included within the scope of protection of this application.

Claims

1. An intelligent translation method for foreign text on printed circuit boards, characterized in that, include: Acquire foreign language document data containing printed circuit board terminology, wherein the foreign language document data contains at least one target language other than the specified language; For each foreign language sentence in the foreign language document data, the foreign language sentence is converted into structured text data based on the format type of the foreign language document data; Multiple foreign language reference words that match the structured text data are extracted from the printed circuit board language library through vector similarity retrieval; The foreign language translation model translates multiple foreign language reference words based on the contextual relationships between them, thereby obtaining the translated text in the specified language.

2. The method according to claim 1, characterized in that, The format type includes editable formats; The conversion of the target foreign language sentence into structured text data based on the format type of the foreign language document data includes: The first text in the foreign language document data is extracted using the document parsing component; Using an image recognition component, the second text embedded in the image in the foreign language document data is extracted, and the second text is added to the corresponding position of the embedded image in the foreign language document data; Based on the first text and the second text, structured text data is obtained.

3. The method according to claim 2, characterized in that, The format types also include non-editable formats; The conversion of the target foreign language sentence into structured text data based on the format type of the foreign language document data includes: The foreign language document data is analyzed using a layout analysis model to detect regions and identify the spatial bounding boxes and category labels of each region. Each region includes at least a title region, a body paragraph region, a table region, and a circuit icon region. For non-table areas, the image recognition component extracts text from each text block in the non-table area to obtain the original non-table text. Based on the position of each non-table area and the first preset reading order, the original non-table text is converted into third text, and the third text is used as structured text data.

4. The method according to claim 3, characterized in that, The method further includes: For the table area, the row and column structure of the table area is restored using a table structure recognition algorithm; Based on the row and column structure, text is extracted from each text block in the table area to obtain the original text of the table. Based on the row and column structure and the second preset reading order, the original text of the table is converted into a fourth text; Based on the third and fourth texts, structured text data is obtained.

5. The method according to claim 1, characterized in that, The printed circuit board language library contains multiple reference text vectors, and a foreign language reference word corresponding to each reference text vector; The step of extracting multiple foreign language reference words that match the structured text data from the printed circuit board language library through vector similarity retrieval includes: Convert the structured text data into a target text vector; For each of the reference text vectors in the printed circuit board language library, calculate the similarity between the target text vector and the target reference text vector; Based on the similarity between each of the reference text vectors and the target text vectors, multiple target reference text vectors and foreign language reference words corresponding to each of the target reference text vectors are determined.

6. The method according to claim 1, characterized in that, The step of translating multiple foreign language reference words based on the contextual relationships between them using a foreign language translation model to obtain translated text in the specified language includes: Based on preset splitting conditions, the foreign language document data is split into multiple sub-documents; For each sub-document, the foreign language reference words corresponding to the sub-document are translated using a foreign language translation model to obtain the translated text in the specified language.

7. The method according to claim 6, characterized in that, The step of translating multiple foreign language reference words based on the contextual relationships between them using a foreign language translation model to obtain translated text in the specified language includes: The structured text data is divided into multiple translation processing units according to preset granularity conditions, and each translation processing unit corresponds to a page or paragraph; For multiple units to be translated, obtain the historical translated text obtained by the foreign language translation model of the previous unit to be translated for the target unit; The historical translation text and the foreign language reference words corresponding to the target translation processing unit are input into the foreign language translation model to obtain the translation text in the specified language.

8. An intelligent translation device for foreign text on a printed circuit board, characterized in that, include: The acquisition module is used to acquire foreign language document data containing printed circuit board terminology, wherein the foreign language document data contains at least one target language other than the specified language; The conversion module is used to convert each foreign language sentence in the foreign language document data into structured text data based on the format type of the foreign language document data. The extraction module is used to extract multiple foreign language reference words that match the structured text data from the printed circuit board language library through vector similarity retrieval; The translation module is used to translate multiple foreign language reference words based on the contextual relationships between the foreign language reference words using a foreign language translation model, so as to obtain the translated text in the specified language.

9. A terminal device, characterized in that, The terminal device includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the intelligent translation method for foreign text on printed circuit boards according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed on a processor, implements an intelligent translation method for foreign text on printed circuit boards according to any one of claims 1-7.

Images