Intelligent place name and address translation method based on multilingual syllable segmentation

Abstract

The present application relates to the field of intelligent translation technology, and specifically discloses a method for intelligent place name and address translation based on multilingual syllable segmentation, which introduces a deep learning algorithm to make adaptive translation strategy decisions for each vocabulary unit in the source language place name and address text, so as to automatically distinguish between proper nouns that should be transliterated and common vocabulary that should be translated using a dictionary, and performs fine-grained syllable-level segmentation and target language syllable mapping on proper nouns marked as transliterated to achieve accurate transliteration of proper nouns; at the same time, for common vocabulary marked as dictionary translation, a dictionary matching algorithm is used for accurate translation, and then the transliteration results of proper nouns and the translation results of common vocabulary are standardized and reorganized to obtain the place name and address translation results. The present application can effectively improve the translation effect of proper place names when mixed with common words, and enhance the translation accuracy and flexibility of place names and addresses in a multilingual environment.

Description

Technical Field

[0001] The present application relates to the field of intelligent translation technology, and more specifically, to an intelligent place name and address translation method based on multilingual syllable segmentation. Background Art

[0002] With the accelerating pace of internationalization, cross-language translation of place names and addresses is increasingly becoming a core requirement for information exchange in multiple fields, including logistics, map services, cross-border e-commerce, and international tourism. Place names and addresses contain not only conventional common vocabulary but also a large number of proper nouns (such as landmarks, villages, and road names). These proper nouns often span multiple language systems and exhibit significant differences in pronunciation and reading and writing patterns. Simply relying on global dictionaries or traditional rule mapping often makes it difficult to accurately restore the pronunciation or geographical semantics of place names, resulting in ambiguity in information understanding and confusion in actual scenarios. To ensure the accurate transmission of place names and addresses in a multilingual environment, it is necessary to build intelligent translation technology that takes into account semantics, phonology, and structure to achieve accurate and natural conversion of place names and addresses between different languages.

[0003] In the existing technology, the translation of place names and addresses is mainly based on rule-based dictionary translation or end-to-end neural machine translation. The dictionary translation method can provide relatively accurate matching for common vocabulary, but its processing ability for proper nouns is limited, especially when faced with emerging landmarks or unpopular place names. Although end-to-end neural machine translation has a certain generalization ability, it is difficult to balance the pronunciation restoration and semantic consistency of place name proper nouns. The phenomenon of "literal translation" or "mistranslation" of place names frequently occurs, which seriously affects the recognizability and usability in actual applications, and it is difficult to meet the actual cross-language place name and address information accurate translation needs.

[0004] Therefore, it is necessary to provide an optimized intelligent place name and address translation method based on multilingual syllable segmentation to solve the above technical problems. Summary of the Invention

[0005] In order to solve the above technical problems, this application is proposed.

[0006] According to one aspect of the present application, a method for intelligent place name and address translation based on multilingual syllable segmentation is provided, which includes:

[0007] Get the place name and address text in the source language and the target language specified by the user;

[0008] Performing translation strategy decisions on each lexical unit in the source language place name and address text to obtain a sequence of source language place name and address text words containing translation strategy annotations, wherein the translation strategy includes transliteration and dictionary translation;

[0009] Based on the target language, calling the target language general vocabulary translation dictionary to perform standard translation on the source language place name and address text words marked as dictionary translation in the sequence of source language place name and address text words containing translation strategy annotations, so as to obtain a sequence of place name and address general vocabulary translation results;

[0010] Based on the target language, performing syllable segmentation mapping on the source language place name and address text words marked as transliterated in the sequence of source language place name and address text words containing the translation strategy annotations, so as to obtain a sequence of transliteration results of place name and address proprietary vocabulary;

[0011] The sequence of the place name and address general vocabulary translation results and the sequence of the place name and address specific vocabulary transliteration results are structurally reconstructed to obtain the target language place name and address translation results.

[0012] Beneficial effects: Compared with the existing technology, the intelligent place name and address translation method based on multilingual syllable segmentation provided by this application makes adaptive translation strategy decisions for each vocabulary unit in the source language place name and address text by introducing a deep learning algorithm, so as to automatically distinguish between proper nouns that should be transliterated and common vocabulary that should be translated using a dictionary, and then performs fine-grained segmentation at the syllable level and target language syllable mapping on the proper nouns marked as transliterated to achieve accurate transliteration of the proper nouns; at the same time, for common vocabulary marked as dictionary translation, a dictionary matching algorithm is used for accurate translation. This application obtains place name and address translation results by normalizing and reorganizing the transliteration results of proper nouns and the translation results of common vocabulary, which can effectively improve the translation effect of proper place names when mixed with common words, and enhance the translation accuracy and flexibility of place names and addresses in a multilingual environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0014] Figure 1 This is a flowchart of a method for intelligent place name and address translation based on multilingual syllable segmentation according to an embodiment of the present application.

[0015] Figure 2 This is a data flow diagram of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application.

[0016] Figure 3This is a flowchart of sub-step S2 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application.

[0017] Figure 4 This is a flowchart of sub-step S4 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application.

[0018] Figure 5 This is a flowchart of sub-step S42 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application.

[0019] Figure 6 This is a flowchart of sub-step S422 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application.

[0020] Figure 7 This is a flowchart of sub-step S4222 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0022] Although the present application makes various references to certain modules in the system according to embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are illustrative only, and different aspects of the system and method can use different modules.

[0023] Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0025] It should be noted in advance that the acquisition and processing of all information or data in this application are carried out in compliance with relevant data protection laws and policies and with the authorization of the relevant authority managers.

[0026] Figure 1 This is a flowchart of a method for intelligent place name and address translation based on multilingual syllable segmentation according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the intelligent place name and address translation method based on multilingual syllable segmentation according to the embodiment of the present application. Figure 1 and Figure 2 As shown, the intelligent place name and address translation method based on multilingual syllable segmentation includes the following steps: S1, obtaining a source language place name and address text and a user-specified target language; S2, making a translation strategy decision on each vocabulary unit in the source language place name and address text to obtain a sequence of source language place name and address text words containing translation strategy annotations, wherein the translation strategy includes transliteration and dictionary translation; S3, based on the target language, calling a target language general vocabulary translation dictionary to perform standard translation on the source language place name and address text words marked as dictionary translation in the sequence of source language place name and address text words containing translation strategy annotations to obtain a sequence of place name and address general vocabulary translation results; S4, based on the target language, performing syllable segmentation mapping on the source language place name and address text words marked as transliteration in the sequence of source language place name and address text words containing translation strategy annotations to obtain a sequence of place name and address specific vocabulary transliteration results; S5, structurally reconstructing the sequence of place name and address general vocabulary translation results and the sequence of place name and address specific vocabulary transliteration results to obtain a target language place name and address translation result.

[0027] In the above-mentioned intelligent place name and address translation method based on multilingual syllable segmentation, described step S1, obtains the source language place name and address text and the target language specified by the user.It should be understood that the translation demand of place name and address in actual application has obvious diversified characteristics. Users need to carry out targeted conversion for different source language place name and address in scenes such as cross-border e-commerce, international logistics, and map services, and display them in the target language required by the target user or business scenario.Therefore, in order to ensure the pertinence and adaptability of translation service, this application is based on customized input mechanism, first collects source language place name and address text, and allows the user to specify the target language of desired translation, and provides basic input for subsequent translation process.In the specific implementation process, the required original place name and address information and target language instruction can be obtained through user input interface, one-stop interface or batch import mode to realize the standardized startup of translation process and the flexibility of multilingual coverage.

[0028] In the specific implementation process, the collection of source language place name address text and target language instructions can be achieved through various methods. For example, in a cross-border e-commerce platform, the system can embed a multilingual address input component, allowing users to fill in the delivery or shipping address in natural language, and automatically identify the language system used in the address when the form is submitted, and extract it as the source language place name address text. At the same time, when the user selects the delivery destination area, the system will automatically match the official language or commonly used language of the area and record it as the target language instruction. This intelligent guidance mechanism based on user behavior not only improves the convenience of data collection, but also enhances the ability to recognize the language attributes of address information.

[0029] In international logistics scenarios, address information often comes from customers, suppliers, or transportation nodes in different countries, and their formats and language styles vary greatly. In order to uniformly manage and efficiently process these heterogeneous data, the system can provide a batch import interface, allowing users to upload files containing multiple place names and addresses (such as Excel spreadsheets or CSV files), and perform language detection on each line of address content during the file parsing phase, automatically marking the source language type to which it belongs. At the same time, users can manually specify the target language into which the imported data needs to be translated in the upload interface, or automatically recommend target language options based on preset rules (such as based on the recipient country). This method is particularly suitable for large-scale address translation tasks, which can significantly improve data processing efficiency and reduce errors caused by manual intervention.

[0030] Map service applications are another typical application environment. In such systems, users may need to convert a specific place name address into other language versions when navigating, searching for places, or sharing locations. At this time, the system can receive the user's spoken address information through the voice recognition module, and use the language recognition model to determine the language type used in the text after the voice is converted to text, as the source language place name address text. Subsequently, the user can switch the language display mode on the map interface, and the system will determine the target language accordingly, and include the address information currently entered by voice into the translation queue. This real-time interactive address collection method not only improves the user experience, but also provides instant and accurate language input for subsequent translation processing.

[0031] Furthermore, in some highly customized business systems, API calls can be used to automatically retrieve source language place name address text and target language instructions. For example, if an enterprise's internal information management system needs to connect to an external multilingual address translation service, it can configure a standard RESTful API interface to automatically send a request to the translation server each time address data is added or modified. The request body should contain the original address text and its corresponding source language identifier, as well as the target language parameters set by the user or system. This method is suitable for highly integrated information systems and helps to achieve automated address translation across platforms and systems.

[0032] At the technical implementation level, the acquisition of source language place name and address text relies on language recognition technology in natural language processing. Deep learning-based language classification models, such as FastText and BERT, are usually used to identify the language of the input text. After being trained on large-scale multilingual corpora, such models can complete the language identification determination of any text in milliseconds with high accuracy, making them particularly suitable for complex scenarios with mixed language or multilingual input. The designation of the target language relies more on the design of user interaction logic, including drop-down menu selection, automatic positioning matching, default language setting, etc., which can all serve as effective input channels.

[0033] It's worth noting that during actual deployment, the integrity and standardization of address texts must also be considered. For example, some non-standardized addresses may lack necessary administrative division information, or may have irregular spelling and confusing grammatical structures, which can increase the difficulty of language recognition. To this end, while obtaining the source language place name address text, the address standardization module can be combined to perform preliminary cleaning and structuring of the original text, such as removing redundant characters, completing missing fields, and unifying unit expressions, to improve the stability of the subsequent translation process.

[0034] In the above-mentioned intelligent place name and address translation method based on multilingual syllable segmentation, the step S2 makes a translation strategy decision for each vocabulary unit in the source language place name and address text to obtain a sequence of source language place name and address text words containing translation strategy annotations, and the translation strategy includes transliteration and dictionary translation. That is, considering that place names and addresses often contain both proprietary vocabulary that need to retain phonetic features (such as landmarks, road names, community names), and common vocabulary that must strictly reflect semantics and conform to local expression habits (such as "street", "road", "district", etc.), a single translation strategy cannot fully meet the requirements of information fidelity and ease of use. Therefore, in order to flexibly adopt different processing strategies at each vocabulary granularity, the present application introduces a translation strategy adaptive decision mechanism based on a deep neural network model, which automatically identifies and annotates the translation strategy that should be adopted for each vocabulary unit by performing deep semantic analysis and context perception on the source language place name and address text. Among them, Figure 3 Flowchart of sub-step S2 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application. Figure 3 As shown, step S2 includes the following steps: S21, performing word segmentation processing on the source language place name and address text to obtain a sequence of source language place name and address text words; S22, performing semantic embedding encoding based on the mBERT model on each source language place name and address text word in the sequence of source language place name and address text words to obtain a sequence of source language place name and address word granularity semantic embedding encoding vectors; S23, inputting the sequence of source language place name and address word granularity semantic embedding encoding vectors into a translation strategy classifier based on the Transformer architecture to obtain the sequence of source language place name and address text words containing translation strategy annotations.

[0035] Specifically, the step S21 performs word segmentation processing on the source language place name and address text to obtain a sequence of source language place name and address text words. It should be understood that the place name and address text is usually composed of multiple vocabulary units, which include both common vocabulary that requires dictionary translation and proper nouns or landmarks that require transliteration. In order to achieve granular subdivision of the place name and address text in subsequent translation strategy decisions, this application is based on a word segmentation algorithm, and performs word segmentation pre-processing on the input source language place name and address text to cut the continuous character string into semantically independent vocabulary units to form a sequence of source language place name and address text words. In the specific implementation process, the system selects an adaptive word segmentation tool based on the structural characteristics of different languages. For example, a word segmentation method based on word frequency statistics and dictionary matching is used in the Chinese environment, while space and punctuation segmentation is used in English or other Latin-based languages. In this way, a sequence of source language place name address text words can be established at the smallest semantic unit level of the address, providing a standardized data input format for the subsequent translation strategy annotation process, fundamentally eliminating the problems of blurred word boundaries and the mixing of proprietary and common words, and greatly improving the sophistication and intelligence of the overall translation processing flow.

[0036] Specifically, step S22 performs semantic embedding encoding on each source language place name and address text word in the sequence of source language place name and address text words based on the mBERT model to obtain a sequence of granular semantic embedding encoding vectors for the source language place name and address text words. Specifically, since each lexical unit in the place name and address text not only has surface spelling information but also has contextual semantic dependencies with other lexical units, classification and annotation based solely on vocabulary information can hardly accurately identify the specific category attributes of the vocabulary. This is especially true for proprietary vocabulary such as landmarks and place names, whose functions can often only be effectively distinguished within a specific context. Therefore, to capture the deep semantic information of each source language place name and address text word in context, this application introduces the multilingual BERT (mBERT) deep pre-trained semantic model to perform semantic embedding encoding on each source language place name and address text word to obtain a semantic embedding representation for each text word. It should be understood that the mBERT model has strong cross-lingual understanding capabilities, enabling effective semantic encoding of text in multiple languages, and has accumulated rich contextual semantic knowledge through its pre-training mechanism. In this application, based on the powerful cross-language semantic representation capability of the mBERT model, each source language place name and address text word can be mapped to a high-dimensional semantic space, capturing the deep semantic associations and contextual dependencies between words, thereby obtaining a sequence of granular semantic embedding encoding vectors of the source language place name and address words, providing accurate and comprehensive semantic information support for the adaptive decision-making of subsequent translation strategies.

[0037] Specifically, in step S23, the sequence of the granular semantic embedding encoding vectors of the source language place name and address words is input into a translation strategy classifier based on the Transformer architecture to obtain the sequence of source language place name and address text words containing translation strategy annotations. It should be understood that, considering that when making translation strategy classification decisions for each source language place name and address text word, the traditional fully connected network often causes overfitting problems due to fixed structure and too many parameters, and it is difficult to capture long-distance semantic dependencies in the sequence. In this regard, in order to achieve fine-grained context-sensitive classification, this application adopts the Transformer model to construct a translation strategy classifier, and uses the self-attention mechanism and position encoding of the Transformer architecture to process long-distance dependencies in the sequence, while maintaining the model's sensitivity to each position in the input sequence and parallel computing capabilities, so as to achieve in-depth contextual understanding of the semantic information of each source language place name and address text word, and finally outputs the binary label (transliteration / dictionary translation) of each source language place name and address text word through the Softmax layer, thereby adaptively determining the most appropriate translation strategy for each text word, providing strong technical support for multilingual translation services of place names and addresses.

[0038] In the above-mentioned intelligent place name and address translation method based on multilingual syllable segmentation, step S3, based on the target language, calls the target language common vocabulary translation dictionary to perform standard translation on the source language place name and address text words marked as dictionary translation in the sequence of source language place name and address text words containing translation strategy annotations, so as to obtain a sequence of place name and address common vocabulary translation results. Specifically, since common vocabulary often has clear and stable equivalent expressions across different languages, directly using authoritative target language dictionary translation can ensure accurate semantic communication. Therefore, to ensure the standardization, clarity, and universality of common descriptive vocabulary (such as direction, ordinal number, type, etc.) in addresses, this application uses a pre-set multilingual dictionary to call the target language common vocabulary translation dictionary to search and match each text word marked as dictionary translation in the source language place name and address text, directly find its corresponding target language expression, and thus generate the corresponding common vocabulary translation result. Through this standardized process, it is possible to effectively avoid mistranslations or ambiguities that may occur in existing neural translation models and achieve high-quality matching of common vocabulary.

[0039] In the above-mentioned intelligent place name and address translation method based on multilingual syllable segmentation, the step S4, based on the target language, performs syllable segmentation mapping on the source language place name and address text words marked as transliteration in the sequence of source language place name and address text words containing translation strategy annotations, so as to obtain a sequence of transliteration results of place name and address proprietary vocabulary. Specifically, due to the high degree of regionality and phonetic characteristics of proprietary place names, the use of transliteration can best preserve the pronunciation information in the original language, thereby achieving cross-language user recognition and scene restoration. At the same time, it is taken into account that direct transliteration of the entire word will lead to distortion of the pronunciation of the target language (for example, transliterating the Chinese word "Xi'an" as "Xian" can easily be misread as "Xian"). Therefore, in order to ensure the reliability and fluency of the transliteration process, this application further performs fine-grained segmentation of each source language place name and address text word marked as transliterated at the syllable level, breaks it down into the smallest pronunciation unit, and maps it one by one according to the syllable structure characteristics of the target language to ensure that each syllable can find the closest pronunciation expression in the target language, thereby ensuring that the transliteration result is natural, smooth and easy to understand. Among them, Figure 4 Flowchart of sub-step S4 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application. Figure 4 As shown, the step S4 includes the steps of: S41, using the G2P model to perform phoneme conversion on the source language place name and address text words marked as transliteration to obtain a source language place name and address proprietary vocabulary phoneme sequence; S42, performing syllable segmentation on the source language place name and address proprietary vocabulary phoneme sequence to obtain a source language place name and address proprietary vocabulary syllable sequence; S43, performing syllable mapping based on the target language syllable library on each source language place name and address proprietary vocabulary syllable in the source language place name and address proprietary vocabulary syllable sequence to obtain a target language place name and address proprietary vocabulary syllable sequence; S44, using the P2G model to convert the target language place name and address proprietary vocabulary syllable sequence into text to obtain the place name and address proprietary vocabulary transliteration result.

[0040] Specifically, the step S41 uses the G2P model to perform phoneme conversion on the source language place name address text words marked as transliterated to obtain the source language place name address proper vocabulary phoneme sequence. It should be understood that this application takes into account the proper nouns in the place name address (such as street names " " or city name" (Seoul)") Its written form (grapheme) varies greatly in different languages, and direct transliteration based on characters is easily affected by spelling rules (e.g., the English pronunciation of "Leicester" is / / rather than the surface pronunciation of letter combinations), and the phonetic systems of multiple languages are significantly different (for example, the Chinese pinyin "x" corresponds to the international phonetic symbol / / , Spanish "j" is pronounced / x / ). Therefore, in order to accurately capture the actual pronunciation characteristics of the source language place names, this application converts the spelling characters into the International Phonetic Alphabet (IPA) by training a multi-task G2P (Grapheme-to-Phoneme) model based on the phonetic phoneme conversion principle. In the specific implementation process, the Transformer architecture G2P model pre-trained for the source language is loaded (such as Chinese uses the Pinyin-to-IPA model based on the decomposition of vowels and vowels, and Russian uses the mapping rules of Cyrillic letters to X-SAMPA symbols), and the source language place name address text words marked as transliteration (such as "Rivoli" in French "Rue de Rivoli") are converted at the phoneme level to generate a standardized proprietary vocabulary phoneme sequence (such as "Rivoli" → / / ), providing an accurate pronunciation benchmark for subsequent syllable segmentation mapping.

[0041] Specifically, step S42 is to segment the phoneme sequence of the source language place name and address to obtain the syllable sequence of the source language place name and address. Specifically, due to the essential differences in syllable division rules of different languages (e.g., English allows complex consonant clusters while Japanese strictly follows the CV structure), and mechanical equal division of phonemes can easily lead to pronunciation distortion (e.g., dividing the Chinese word "Xi'an" into " / "). / Error segmented into / / ). Therefore, to achieve syllable boundary location consistent with speech perception, this application further performs context-dependent analysis on each phoneme in the source language place name and address vocabulary phoneme sequence to accurately identify syllable boundaries, thereby segmenting each syllable unit in the source language place name and address vocabulary. Figure 5 is a flowchart of sub-step S42 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of this application. As shown in Figure 5, step S42 includes the following steps: S421, inputting the source language place name and address proper vocabulary phoneme sequence into a Bi-LSTM model including an embedding layer to obtain a sequence of source language place name and address proper vocabulary phoneme embedding associated coding vectors; S422, extracting the source language place name and address proper vocabulary phoneme embedding associated coding vectors to be analyzed from the sequence of source language place name and address proper vocabulary phoneme embedding associated coding vectors, and performing syllable boundary prediction on the source language place name and address proper vocabulary phoneme embedding associated coding vectors to be analyzed to obtain a syllable boundary annotation prediction result; S423, performing syllable segmentation on the source language place name and address proper vocabulary phoneme sequence according to the syllable boundary prediction result to obtain the source language place name and address proper vocabulary syllable sequence.

[0042] More specifically, in step S421, the source language place name and address proper vocabulary phoneme sequence is input into a Bi-LSTM model including an embedding layer to obtain a sequence of source language place name and address proper vocabulary phoneme embedding associated encoding vectors. It should be understood that since the original phoneme sequence obtained from the G2P model (for example, " / b / , / The discrete phoneme symbols (such as " / , / r / , / k / , / l / , / i / ") themselves lack direct and rich contextual association information, and the demarcation of syllable boundaries is often highly dependent on the combined environment before and after the phoneme and the phonotactics of the language. Therefore, in order to fully understand the specific role and contextual dependencies of each phoneme in its sequence and accurately predict syllable boundaries, this application is based on the principle of deep learning sequence modeling. First, each discrete phoneme symbol is mapped into a low-dimensional dense real vector through an embedding layer (EmbeddingLayer), and the phonological similarity between phonemes is retained in the vector space. Then, the contextual dependency relationship between phonemes is modeled through a bidirectional long short-term memory network (Bi-LSTM), capturing the long-distance dependency features between phonemes to obtain deep-level association information between phonemes. Specifically, the Bi-LSTM layer uses two independent LSTMs The network traverses the phoneme sequence from both the forward and backward passes to independently capture the forward and backward dependencies of the phonemes. It then concatenates the forward and backward hidden states of each phoneme's embedding features to generate a context-aware representation of each phoneme, resulting in a sequence of associated encoding vectors for the phoneme embeddings of the source language's place names, addresses, and proprietary vocabulary. This approach enables a more accurate understanding of the complex interactions between phonemes, providing a strong basis for subsequent syllable boundary prediction.

[0043] More specifically, the step S422 extracts the source language place name and address specific vocabulary phoneme embedded associated coding vectors from the sequence of the source language place name and address specific vocabulary phoneme embedded associated coding vectors, and performs syllable boundary prediction on the source language place name and address specific vocabulary phoneme embedded associated coding vectors to obtain syllable boundary annotation prediction results. Specifically, since syllable segmentation needs to consider both the legitimacy of local phoneme combinations (such as Japanese does not allow "ん" to be followed by a consonant) and the global prosodic structure (such as the tendency of Chinese to be disyllabic), and single-scale feature extraction is prone to over-segmentation or under-segmentation (such as French "aujourd'hui" / / is incorrectly segmented into three syllables instead of the correct four syllables). To achieve accurate syllable boundary prediction, this application extracts the source language place name and address proprietary vocabulary phoneme embedding associated coding vector to be analyzed from the sequence of the source language place name and address proprietary vocabulary phoneme embedding associated coding vector as the analysis target, and measures its syllable boundary adaptability at different granularities by mining its local neighborhood structure and global distribution characteristics in the phoneme embedding space, thereby achieving intelligent prediction of syllable boundaries. Among them, Figure 6 FIG4 is a flowchart of sub-step S422 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application. Figure 6 As shown, the step S422 includes the steps of: S4221, performing syllable boundary prediction on the phoneme embedded associated coding vector of the source language place name and address to be analyzed based on global context association perception, so as to obtain the global fitness of the syllable boundary of the phoneme feature to be analyzed; S4222, performing syllable boundary prediction on the phoneme embedded associated coding vector of the source language place name and address to be analyzed based on local neighborhood association perception, so as to obtain the local fitness of the syllable boundary of the phoneme feature to be analyzed; S4223, determining whether the phoneme corresponding to the phoneme embedded associated coding vector of the source language place name and address to be analyzed is a syllable boundary based on the global fitness of the syllable boundary of the phoneme feature to be analyzed and the local fitness of the syllable boundary of the phoneme feature to be analyzed.

[0044] In a specific example of the present application, step S4221 includes: first, inputting the sequence of the source language place name and address specific vocabulary phoneme embedding associated coding vector and the source language place name and address specific vocabulary phoneme embedding associated coding vector into a global context adaptation learning module based on a converter structure to obtain the global context adaptation associated coding vector of the phoneme feature syllable boundary to be analyzed, which is expressed as follows:

[0045] ;

[0046] ;

[0047] in, A sequence of phoneme embeddings associated with the encoding vectors of the source language place names and addresses, 、 、 and They represent the first, second, and third phonemes in the sequence of the source language place name and address specific vocabulary embedded in the associated coding vector. and The source language place name address specific vocabulary phoneme embedding associated encoding vector, represents the phoneme-embedded associated coding vector of the place name and address proper vocabulary of the source language to be analyzed, represents the Transformer encoder, Represents the global context adaptation associated encoding vector of the phoneme feature syllable boundary to be analyzed.

[0048] That is, the Transformer architecture is used to capture the complex dependencies between the phonemes of the source language place names and addresses to be analyzed and the overall sequence of the phonemes in the source language, generating a highly context-aware global context-adaptive association encoding vector of the phoneme feature syllable boundaries to be analyzed. The global context-adaptive association encoding vector of the phoneme feature syllable boundaries to be analyzed not only contains the attribute information of the phoneme to be analyzed itself, but also encapsulates its structural role in the entire distribution of phonemes in the proprietary vocabulary, its interaction relationship with all other phonemes, and its position in the global phoneme ecosystem, providing an information basis for the subsequent accurate demarcation of syllable boundaries and the realization of cross-language syllable mapping.

[0049] Then, the global context adaptation associated coding vector of the phoneme feature syllable boundary to be analyzed is explicitly decoded to obtain the global adaptation degree of the phoneme feature syllable boundary to be analyzed, which is expressed as follows:

[0050] ;

[0051] in, represents the sigmoid activation function, represents transpose, is the learnable weight parameter, represents the bias term, Indicates the global adaptability of the syllable boundary of the phoneme feature to be analyzed.

[0052] Specifically, through an explicit decoding process, the global contextual adaptation-associated encoding vectors of the syllable boundaries of the phoneme features to be analyzed are projected from a high-dimensional space into a low-dimensional scalar value space. The signals most relevant to the global adaptability of the syllable boundaries are extracted and made explicit, quantifying the degree of fit between the phoneme features to be analyzed and the macro-pattern or distribution defined by the entire source language place-name address proper vocabulary phoneme set. This information compression and abstraction process yields the global adaptability of the syllable boundaries of the phoneme features to be analyzed, directly reflecting the rationality of syllable boundary division. This effectively guides subsequent syllable segmentation and mapping decisions, improving the accuracy of pronunciation restoration during proper noun transliteration and the accuracy of cross-language phonetic mapping.

[0053] Figure 7 Flowchart of sub-step S4222 of the intelligent place name and address translation method based on multilingual syllable segmentation according to an embodiment of the present application. Figure 7As shown, the step S4222 includes the steps of: S42221, extracting the local neighborhood context information of the source language place name and address proprietary vocabulary phoneme embedded associated coding vector in the sequence of the source language place name and address proprietary vocabulary phoneme embedded associated coding vector to obtain the local neighborhood set of the source language place name and address proprietary vocabulary phoneme embedded associated coding vector; S42222, inputting the source language place name and address proprietary vocabulary phoneme embedded associated coding vector and the local neighborhood set of the source language place name and address proprietary vocabulary phoneme embedded associated coding vector into the local context adaptation learning module to obtain the local adaptation of the syllable boundary of the phoneme feature to be analyzed.

[0054] In a specific example of the present application, step S42221 is expressed as follows:

[0055] ;

[0056] in, represents the local neighborhood feature extraction radius, Indicates Sequence of phoneme embeddings associated with encoding vectors of place names and addresses in the source language The position in is the center and the radius is extracted The source language place name address proper vocabulary phoneme embedding associated encoding vector, The local neighborhood set representing the source language place name address vocabulary phoneme embedding associated encoding vector, The dimension of the vector in the local neighborhood set of the associated encoding vector of the source language place name and address proper vocabulary phoneme embedding, The number of vectors in the local neighborhood set of the source language place name and address proper vocabulary phoneme embedding associated encoding vector.

[0057] That is, we define and isolate the microenvironment or small range that is most directly related to the phonemes of the specific vocabulary of the place name and address to be analyzed in the source language, providing a data basis for the subsequent evaluation of its local adaptability with the surrounding phonemes, thereby evaluating the phoneme characteristics in different local environments and enhancing the robustness of the evaluation. In this way, we can focus on the local contextual information around the phoneme to be analyzed, capture the short-distance dependencies and local speech patterns between phonemes, and provide support for the subsequent more accurate identification of syllable boundaries and the realization of cross-language syllable mapping, thereby improving the accuracy and naturalness of the transliteration results.

[0058] In a specific example of the present application, step S42222 is expressed as follows:

[0059] ;

[0060] ;

[0061] in, The local neighborhood set of the source language place name address proper vocabulary phoneme embedding associated coding vector is represented by The source language place name address specific vocabulary phoneme embedding associated encoding vector, express Relative to The syllable boundary monomer phoneme association fitness, is the learnable weight parameter, 、 and is the local neighborhood weight matrix, is the activation function, yes and The characteristic scale of represents the hyperbolic tangent function, Indicates the local adaptability of the syllable boundary of the phoneme feature to be analyzed.

[0062] Specifically, the degree of fit between phoneme features and syllable boundaries is assessed from the perspective of local context. By quantitatively mining the syllable boundary fit between the phoneme feature to be analyzed and each phoneme feature in each local neighborhood, and aggregating the contextual fit information of all local neighborhood phonemes, the phoneme's localized speech feature association pattern is captured, resulting in the local fit of the phoneme feature to be analyzed. This helps accurately identify syllable boundaries that conform to the pronunciation patterns of the source language, effectively improving the accuracy of proper noun segmentation. Furthermore, by incorporating local context information, the subsequent syllable mapping process can better preserve the pronunciation characteristics of the source language, ultimately enhancing the recognizability and phonetic fidelity of proper place name translation in multilingual environments.

[0063] In particular, in a preferred example of the present application, the step S42221 includes: taking the source language place name and address specific vocabulary phoneme embedding associated coding vector to be analyzed as the center, based on the initial local neighborhood feature extraction radius, extracting the local neighborhood set of the initial source language place name and address specific vocabulary phoneme embedding associated coding vector; based on the mixed state information entanglement strength between the local neighborhood set of the initial source language place name and address specific vocabulary phoneme embedding associated coding vector and the source language place name and address specific vocabulary phoneme embedding associated coding vector to be analyzed, iteratively optimizing the initial local neighborhood feature extraction radius to obtain an optimized local neighborhood feature extraction radius; based on the optimized local neighborhood feature extraction radius, taking the source language place name and address specific vocabulary phoneme embedding associated coding vector to be analyzed as the center, extracting the local neighborhood set of the source language place name and address specific vocabulary phoneme embedding associated coding vector.

[0064] Considering that when extracting the local neighborhood set corresponding to the source language place name and address specific vocabulary phoneme embedded associated coding vector to be analyzed, although the local neighborhood set of the source language place name and address specific vocabulary phoneme embedded associated coding vector limits the relevant microenvironment or small range, the source language place name and address specific vocabulary phoneme embedded associated coding vector to be analyzed and its local neighborhood set still transition from the geometric eigenstate to the associated coupling phase state, thereby causing the measurement of associated coupling when calculating the local fitness of the features therebetween. Therefore, the present application hopes to improve the calculation accuracy of the local fitness of the syllable boundary of the phoneme feature to be analyzed by iteratively optimizing the initial local neighborhood feature extraction radius.

[0065] Specifically, first, let , the adjacent manifold architecture based on the microscopic behavior pattern is used as the probability density reduction distribution, so as to obtain the mixed-state functional index value of the pure-state adaptation functional in the form of the global non-local entropy feature technology:

[0066] ;

[0067] in, represents the logarithmic value with a natural constant as the base, express Relative to The pure state fitness of represents the global mixed state functional index value.

[0068] That is, each In the case of pure state fitness distribution, the global non-local uncertainty or degree of mixing is formally measured based on the statistical mixture of each pure state fitness distribution, so as to quantify the mixed state characterization at different levels.

[0069] At the same time, the geometric regularization factor is introduced , calculate the manifold measure expansion value:

[0070] ;

[0071] in, represents the logarithm with base 2, represents the mixed state information entanglement strength, Indicates the A geometric regularization factor.

[0072] That is, the global non-negative eigendecomposition of the pure state fitness distribution is used to expand the paradigm based on entropy correlation, thereby expanding the mixed state representation so that the information entropy is associated with the degree of entanglement. In this way, the initial local neighborhood feature extraction radius can be adjusted by modulating the neighborhood parameter. It is minimized, thereby reducing the measurement correlation coupling. Finally, based on the local neighborhood feature extraction radius after modulation optimization, with the source language place name and address specific vocabulary phoneme embedding associated coding vector as the center, the local neighborhood set of the source language place name and address specific vocabulary phoneme embedding associated coding vector is re-extracted, and then the source language place name and address specific vocabulary phoneme embedding associated coding vector and the local neighborhood set of the source language place name and address specific vocabulary phoneme embedding associated coding vector are input into the local context adaptation learning module to calculate the local adaptation of the phoneme feature syllable boundary to be analyzed, thereby improving the calculation accuracy of the local adaptation of the phoneme feature syllable boundary to be analyzed.

[0073] In a specific example of the present application, step S4223 includes: first, based on the global fitness of the phoneme feature syllable boundary to be analyzed and the local fitness of the phoneme feature syllable boundary to be analyzed, determining the comprehensive fitness of the phoneme feature syllable boundary to be analyzed, which is expressed as follows:

[0074] ;

[0075] ;

[0076] in, and represent the weighted fusion weights of the global adaptation degree of the syllable boundary of the phoneme feature to be analyzed and the local adaptation degree of the syllable boundary of the phoneme feature to be analyzed, respectively. represents the normalized exponential function, represents the comprehensive adaptation weight parameter matrix, represents the comprehensive adaptation bias term, Indicates the comprehensive fitness of the syllable boundary of the phoneme feature to be analyzed.

[0077] That is, by integrating the global fitness of the syllable boundaries of the phoneme features to be analyzed and the local fitness of the syllable boundaries of the phoneme features to be analyzed, a more comprehensive and robust comprehensive evaluation standard is constructed to address the limitations of single-scale evaluation in the syllable segmentation of cross-language place names, addresses and proper nouns. Among them, the global fitness can capture the overall correlation pattern between phoneme features and syllable boundaries from a macro level, while the local fitness can explore the fine phonetic structure of phonemes in the local context. The combination of the two can give full play to the synergistic effect and accurately locate syllable boundaries that conform to the pronunciation rules of the source language and take into account the phonetic habits of the target language. The generated comprehensive fitness of the syllable boundaries of the phoneme features to be analyzed can effectively overcome the problems of global analysis ignoring local phonetic details and local analysis lacking overall phonetic association, and significantly improve the accuracy and stability of syllable boundary recognition.

[0078] Then, based on the comparison between the comprehensive fitness of the syllable boundaries of the phoneme features to be analyzed and the preset threshold, it is determined whether the phoneme corresponding to the embedded associated coding vector of the source language place name address proper vocabulary to be analyzed is a syllable boundary. That is, by comparing the comprehensive fitness of the syllable boundaries of the phoneme features to be analyzed with the preset threshold, a clear decision basis is provided for each phoneme to determine whether it constitutes a syllable boundary based on a comprehensive consideration of global speech associations and local pronunciation details. The binary results generated by this threshold decision mechanism can accurately filter out redundant phoneme boundaries with weak speech feature correlation, retain effective boundaries that conform to the pronunciation rules of multiple languages, and significantly improve the accuracy and efficiency of proper noun syllable segmentation. When faced with complex scenarios such as emerging landmarks or unpopular place names, it can effectively reduce transliteration deviations caused by syllable segmentation errors and enhance semantic recognizability and information transmission accuracy in a cross-language environment.

[0079] More specifically, the step S423 performs syllable segmentation on the source language place name and address proper vocabulary phoneme sequence according to the syllable boundary prediction result to obtain the source language place name and address proper vocabulary syllable sequence. It should be understood that the syllable boundary prediction result reveals the end position of each syllable unit in the source language place name and address proper vocabulary phoneme sequence. Based on this syllable boundary information, the source language place name and address proper vocabulary phoneme sequence is segmented to obtain a series of separated syllable units, namely the source language place name and address proper vocabulary syllable sequence. For example, if the original source language place name and address proper vocabulary phoneme sequence is / / , and the predicted boundary labels indicate the end positions of the syllable unit as / k / and / i / , then the segmentation operation will produce two syllables: the first syllable is / / , and the second syllable is / li / . In this way, the original continuous stream of phonemes is effectively and accurately converted into an ordered list or sequence consisting of multiple syllables, such as [ / / , / li / ].

[0080] Specifically, step S43 performs syllable mapping on each syllable of the source language place name and address proprietary vocabulary in the source language place name and address proprietary vocabulary syllable sequence based on the target language syllable library to obtain the target language place name and address proprietary vocabulary syllable sequence. It should be understood that transliteration needs to take into account both pronunciation similarity and target language orthographic norms (e.g., Japanese katakana cannot directly represent French nasal vowels), and simple phoneme literal translation is prone to produce illegal syllables (e.g., English "Smith" / / Avoid " when mapping to Korean ” and other invalid consonant clusters). Therefore, in order to achieve a balance between pronunciation fidelity and acceptability in the target language, this application, based on the principle of optimal syllable alignment, constructs a target language syllable library and adopts a multi-level syllable mapping method to find the best mapping path. Specifically, first load the target language syllable list (such as the Japanese JIS X4063 Katakana syllable table and the Arabic ISO 233 transliteration rules), and then perform multiple candidate matching on each source syllable: for example, the Chinese syllable "zhang" / / When mapped to Russian, calculate its value with " ”( / / ), " ”( / / ) and other candidate phoneme edit distances, and combined with the historical translation frequency (such as "Zhang→ " appears 93% in the official Russian place name database) by comprehensively considering the phoneme edit distance and the frequency of historical translations to select the optimal mapping and ultimately generate the target language place name address vocabulary syllable sequence.

[0081] Specifically, step S44 uses a P2G model to convert the target language place name and address proper vocabulary syllable sequence into text to obtain the place name and address proper vocabulary transliteration result. That is, in order to convert the target language place name and address proper vocabulary syllable sequence into a text form that conforms to the target language writing standard, this application is based on the phoneme-to-grapheme (P2G) principle and uses a P2G model trained for the target language to convert the target language syllable sequence into the final text output. Specifically, the P2G model is also a sequence-to-sequence model based on deep learning. It can select the most appropriate characters or letter combinations based on the syllable sequence, contextual information, and common vocabulary and writing habits of the target language. This not only ensures the accurate pronunciation of the transliteration result, but also generates a text form that conforms to the target language writing standard and is easy to read and understand. This completes the complete conversion from the source language proper name to the target language high-quality transliteration text, so that the final place name and address translation result can not only accurately convey geographical information but also meet the reading and usage habits of users of the target language.

[0082] In the above intelligent place name and address translation method based on multilingual syllable segmentation, in step S5, the sequences of the translation results of the general vocabulary of the place name and address and the transliteration results of the proper vocabulary of the place name and address are structurally reconstructed to obtain the translation result of the target language place name and address. It should be understood that this application takes into account the relatively high structural requirements of place name and address texts. Although individual words have been correctly translated or transliterated, if they are not recombined according to grammar or local habits, problems such as word order confusion and unclear expression are likely to occur. Therefore, in order to finally output a standardized and normalized target language place name and address while comprehensively ensuring accurate meaning expression and readability, this application is based on a structural reconstruction engine to automatically combine and format the output of general translation words and transliterated proper words according to the word order, punctuation, and writing norms of the target language. In the specific implementation process, first, a target language template library is constructed (including English "[Block] [Type], [City] [Suffix]", Japanese "[Prefecture] [City / Town / Village] [Block]", etc.), and the transliteration results and dictionary translation results are sorted according to the target language specifications (English, and necessary prepositions and format symbols are inserted at the same time (such as English commas, black dots in Japanese). In this way, the translation result not only retains the source semantic information but also conforms to the cognitive habits of target users.

[0083] In summary, the intelligent place name and address translation method based on multilingual syllable segmentation according to the embodiments of this application is elucidated. It makes adaptive translation strategy decisions on each lexical unit in the source language place name and address text by introducing deep learning algorithms to automatically distinguish proper nouns that should be transliterated and general vocabulary that should be translated using a dictionary. Then, by performing fine-grained syllable-level segmentation and target language syllable mapping on the proper nouns marked for transliteration, accurate transliteration of proper nouns is achieved; at the same time, for general vocabulary marked for dictionary translation, a dictionary matching algorithm is used for accurate translation. Furthermore, by performing normalized recombination on the transliteration results of proper nouns and the translation results of general vocabulary, the translation result of the place name and address is obtained. This method can effectively improve the translation effect of proper place names when mixed with general words and enhance the translation accuracy and flexibility of place names and addresses in a multilingual environment.

[0084] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and easy understanding, rather than limitations, and the above details do not limit the present invention to necessarily adopt the above specific details to be implemented.

[0085] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0086] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be encompassed therein. Any reference to a figure in a claim should not be construed as limiting the claim to which it relates.

[0087] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0088] Finally, it should be noted that the above description has been provided for purposes of illustration and description. Furthermore, the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to be limiting. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art will appreciate that the technical solutions of the present invention may be modified or replaced with equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent place name and address translation method based on multilingual syllable segmentation, characterized in that: include: Get the place name and address text in the source language and the target language specified by the user; Performing translation strategy decisions on each lexical unit in the source language place name and address text to obtain a sequence of source language place name and address text words containing translation strategy annotations, wherein the translation strategy includes transliteration and dictionary translation; Based on the target language, calling the target language general vocabulary translation dictionary, performing standard translation on the source language place name and address text words marked as dictionary translation in the sequence of source language place name and address text words containing translation strategy annotations, to obtain a sequence of place name and address general vocabulary translation results; Based on the target language, performing syllable segmentation mapping on the source language place name and address text words marked as transliterated in the sequence of source language place name and address text words containing the translation strategy annotations, so as to obtain a sequence of transliteration results of place name and address proprietary vocabulary; The sequence of the place name and address general vocabulary translation results and the sequence of the place name and address specific vocabulary transliteration results are structurally reconstructed to obtain the target language place name and address translation results.

2. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 1 is characterized in that: Performing translation strategy decisions on each lexical unit in the source language place name and address text to obtain a sequence of source language place name and address text words containing translation strategy annotations, wherein the translation strategy includes transliteration and dictionary translation, including: Performing word segmentation processing on the source language place name and address text to obtain a sequence of source language place name and address text words; Performing semantic embedding encoding based on the mBERT model on each source language place name and address text word in the sequence of source language place name and address text words to obtain a sequence of source language place name and address word granular semantic embedding encoding vectors; The sequence of the source language place name and address word granular semantic embedding encoding vectors is input into a translation strategy classifier based on the Transformer architecture to obtain the sequence of source language place name and address text words containing translation strategy annotations.

3. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 1 is characterized in that: Performing syllable segmentation and mapping on the source language place name and address text words marked as transliterated in the sequence of source language place name and address text words containing translation strategy annotations to obtain a sequence of transliteration results of place name and address proprietary vocabulary, including: Using the G2P model, performing phoneme conversion on the source language place name and address text words marked as transliterated words to obtain a phoneme sequence of the source language place name and address proprietary vocabulary; Performing syllable segmentation on the source language place name and address proper vocabulary phoneme sequence to obtain the source language place name and address proper vocabulary syllable sequence; Performing syllable mapping based on a target language syllable library on each source language place name and address proprietary vocabulary syllable in the source language place name and address proprietary vocabulary syllable sequence to obtain a target language place name and address proprietary vocabulary syllable sequence; The P2G model is used to convert the target language place name and address proprietary vocabulary syllable sequence into text to obtain the place name and address proprietary vocabulary transliteration result.

4. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 3 is characterized in that: Performing syllable segmentation on the source language place name and address proper vocabulary phoneme sequence to obtain the source language place name and address proper vocabulary syllable sequence, including: Inputting the source language place name and address proper vocabulary phoneme sequence into a Bi-LSTM model including an embedding layer to obtain a sequence of source language place name and address proper vocabulary phoneme embedding associated coding vectors; Extracting the source language place name and address specific vocabulary phoneme embedding associated coding vector to be analyzed from the sequence of source language place name and address specific vocabulary phoneme embedding associated coding vectors, and performing syllable boundary prediction on the source language place name and address specific vocabulary phoneme embedding associated coding vector to obtain a syllable boundary annotation prediction result; The source language place name and address proper vocabulary phoneme sequence is syllable segmented according to the syllable boundary prediction result to obtain the source language place name and address proper vocabulary syllable sequence.

5. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 4 is characterized in that: Perform syllable boundary prediction on the phoneme embedding associated coding vector of the source language place name and address to be analyzed to obtain syllable boundary annotation prediction results, including: Performing syllable boundary prediction based on global context association perception on the phoneme embedding association coding vector of the source language place name and address to be analyzed, so as to obtain the global fitness of the syllable boundary of the phoneme feature to be analyzed; Performing syllable boundary prediction based on local neighborhood association perception on the phoneme embedding association coding vector of the specific vocabulary of the place name and address of the source language to be analyzed to obtain the local fitness of the syllable boundary of the phoneme feature to be analyzed; Based on the global adaptation degree of the syllable boundary of the phoneme feature to be analyzed and the local adaptation degree of the syllable boundary of the phoneme feature to be analyzed, it is determined whether the phoneme corresponding to the phoneme embedding associated coding vector of the specific vocabulary of the source language place name and address to be analyzed is a syllable boundary.

6. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 5 is characterized in that: The syllable boundary prediction of the phoneme embedded associated coding vector of the source language place name and address to be analyzed is performed based on global context association perception to obtain the global fitness of the syllable boundary of the phoneme feature to be analyzed, including: Inputting the sequence of the source language place name and address specific vocabulary phoneme embedding associated coding vector and the source language place name and address specific vocabulary phoneme embedding associated coding vector into a global context adaptation learning module based on a converter structure to obtain the global context adaptation associated coding vector of the phoneme feature syllable boundary to be analyzed; The global context adaptation associated coding vector of the phoneme feature syllable boundary to be analyzed is explicitly decoded to obtain the global adaptation degree of the phoneme feature syllable boundary to be analyzed.

7. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 6 is characterized in that: The syllable boundary prediction is performed on the phoneme embedded association coding vector of the source language place name and address to be analyzed based on local neighborhood association perception to obtain the local fitness of the syllable boundary of the phoneme feature to be analyzed, including: Extracting local neighborhood context information of the source language place name and address proper vocabulary phoneme embedding associated coding vector to be analyzed in the sequence of the source language place name and address proper vocabulary phoneme embedding associated coding vector to obtain a local neighborhood set of the source language place name and address proper vocabulary phoneme embedding associated coding vector; The source language place name and address specific vocabulary phoneme embedding associated coding vector and the local neighborhood set of the source language place name and address specific vocabulary phoneme embedding associated coding vector are input into the local context adaptation learning module to obtain the local adaptation of the phoneme feature syllable boundary to be analyzed.

8. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 7 is characterized in that: Extracting local neighborhood context information of the source language place name and address proper vocabulary phoneme embedding associated coding vector to be analyzed in the sequence of the source language place name and address proper vocabulary phoneme embedding associated coding vector to obtain a local neighborhood set of the source language place name and address proper vocabulary phoneme embedding associated coding vector, including: Taking the source language place name and address specific vocabulary phoneme embedding associated coding vector to be analyzed as the center, extracting a local neighborhood set of the initial source language place name and address specific vocabulary phoneme embedding associated coding vector based on the initial local neighborhood feature extraction radius; Iteratively optimizing the initial local neighborhood feature extraction radius based on the mixed state information entanglement strength between the local neighborhood set of the initial source language place name and address proper vocabulary phoneme embedding associated coding vector and the source language place name and address proper vocabulary phoneme embedding associated coding vector to be analyzed to obtain an optimized local neighborhood feature extraction radius; Based on the optimized local neighborhood feature extraction radius, a local neighborhood set of the source language place name and address proper vocabulary phoneme embedding associated coding vector is extracted with the source language place name and address proper vocabulary phoneme embedding associated coding vector as the center.

9. The intelligent place name and address translation method based on multilingual syllable segmentation according to claim 7 is characterized in that: Determining whether a phoneme corresponding to the phoneme embedding associated coding vector of the to-be-analyzed source language place name and address proper vocabulary is a syllable boundary based on the global adaptation degree of the to-be-analyzed phoneme feature syllable boundary and the local adaptation degree of the to-be-analyzed phoneme feature syllable boundary includes: Determining the comprehensive fitness of the phoneme-featured syllable boundary to be analyzed based on the global fitness of the phoneme-featured syllable boundary to be analyzed and the local fitness of the phoneme-featured syllable boundary to be analyzed; Based on the comparison between the comprehensive fitness of the syllable boundary of the phoneme feature to be analyzed and a preset threshold, it is determined whether the phoneme corresponding to the phoneme embedded associated coding vector of the source language place name and address proprietary vocabulary to be analyzed is a syllable boundary.

Images