Place name and address translation method based on intelligent splitting and matching of proper name database and common name database

Abstract

The present application relates to the technical field of place name and address translation, and discloses a place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library. The method first constructs a common name library containing standard translated names and a proper name library containing transliterated proper names. After multiple rounds of segmentation of the place name and address to be translated, the word segments in each segmentation scheme are type-labeled through dual-library matching, and the optimal segmentation scheme is determined based on the labeled character coverage. Contextual semantic association analysis is further performed on the unlabeled word segments in the optimal segmentation scheme. After inferring the attribute type of the word segment, a large language model is called to generate an adapted translation. Finally, a rule engine is used to structurally reorganize the dual-library labeled translations and the inferred translations to form a standardized standard address translation. Through multiple rounds of word segmentation and labeling and optimal segmentation scheme screening, the present application can effectively solve the segmentation ambiguity problem, accurately identify the core components of the address, and thus improve the translation quality.

Description

Technical Field

[0001] The present application relates to the technical field of place name and address translation, and more specifically, to a place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library. Background Art

[0002] With the deepening of globalization and the increasing frequency of international exchanges, the accurate and efficient translation of place names and addresses, as key elements that carry geographic location information, is crucial for cross-language communication, international trade, cultural exchange, navigation, emergency response, and many other fields. The significant differences between Chinese and Western language structures, expression conventions, and place name formation rules make accurate and standardized translation a challenge. Improper translation often leads to communication barriers, logistics delays, and even financial losses.

[0003] Existing methods for translating place names and addresses rely on either general-purpose machine translation systems or simple rule matching and dictionary lookups. While general-purpose machine translation systems perform well in general text translation, they often lack domain knowledge when dealing with place names and addresses with highly structured and specific terms, resulting in inaccurate and non-standard translation results. These problems include confusion between transliteration and translation of proper nouns, improper handling of common noun suffixes, and unclear expression of the hierarchical relationships between place name elements. Traditional rule-based and dictionary-based methods, while able to guarantee the accuracy of specific vocabulary translations to a certain extent, have limited coverage and struggle to cope with the complex and ever-changing nature of new place names. Furthermore, rule-making and maintenance costs are high, and generalization capabilities are weak. Especially for Chinese place names and addresses, since there is no clear space separation and unified naming rules, accurate segmentation of each component and accurate attribute identification of each component become bottlenecks in translation. Traditional word segmentation models are trained on general corpus and lack targeted recognition capabilities for the unique combinations of proper nouns and general nouns in place names and addresses. They are prone to unconventional segmentation errors and cannot effectively solve the translation problems of address segmentation ambiguity and out-of-vocabulary (OOV) words, which leads to uneven overall translation quality.

[0004] Therefore, it is necessary to provide a place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library 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 translating place names and addresses based on intelligent splitting and matching of a proper name library and a common name library is provided, which includes:

[0007] Constructing a proper name database and a common name database, wherein the common name database stores common suffixes and type words in place names and addresses and their corresponding standard translations, and the proper name database stores common proper nouns in place names and addresses and their corresponding transliteration results or standard translations;

[0008] Get the Chinese place name and address to be translated entered by the user;

[0009] Based on the proper name database and the common name database, multiple rounds of word segmentation and tagging and optimal segmentation scheme screening are performed on the Chinese place name address to be translated to obtain the optimal splitting and matching result;

[0010] Performing attribute inference and translation generation based on context association constraints on the unlabeled Chinese place name and address word segments to be translated in the optimal splitting and matching results to obtain translation results of the unlabeled place name and address word segments;

[0011] The translation results of the unlabeled place name and address word segments and the translation results of the labeled Chinese place name and address word segments to be translated in the optimal split matching results are assembled and normalized based on a rule engine to obtain a complete Chinese place name and address translation result.

[0012] Beneficial effect: Compared with the prior art, the place name and address translation method provided by the present application is based on the intelligent splitting and matching of the proper name library and the common name library. It first constructs a common name library containing standard translated names and a proper name library of transliterated proper names. After multiple rounds of segmentation of the place name and address to be translated, the word fragments in each segmentation scheme are marked with types through dual-library matching, and the optimal segmentation scheme is determined based on the coverage of the marked characters. On this basis, a context association inference mechanism is further introduced to perform context semantic association analysis on the unmarked word fragments in the optimal segmentation scheme. After inferring the attribute type of the word segment, a large language model is called to generate an adapted translation. Finally, the dual-library marked translation name and the inferred translation are structurally reorganized through a rule engine to form a standardized standard address translation. Through multiple rounds of word segmentation and labeling and optimal segmentation scheme screening, the present application can effectively solve the segmentation ambiguity problem, accurately identify the core components of the address, and thus improve the translation quality. 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 The present invention is a flowchart of a place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to an embodiment of the present application.

[0015] Figure 2 The present invention provides a data flow diagram of a place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to an embodiment of the present application.

[0016] Figure 3 This is a flowchart of sub-step S3 of the place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library according to an embodiment of the present application.

[0017] Figure 4 This is a flowchart of sub-step S4 of the place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library according to an embodiment of the present application.

[0018] Figure 5 This is a flowchart of sub-step S43 of the place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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.

[0023] It should be noted in advance that all data processing in this application is carried out in compliance with the relevant local data protection laws and policies and with the authorization of the relevant authority administrator.

[0024] Figure 1 The present invention is a flowchart of a place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library according to the embodiment of the present application. Figure 1 and Figure 2 As shown, the place name and address translation method based on the intelligent splitting and matching of the proper name library and the common name library includes the following steps: S1, constructing a proper name library and a common name library, wherein the common name library stores common suffixes and type words in place name addresses and their corresponding standard translations, and the proper name library stores common proper nouns in place name addresses and their corresponding transliteration results or standard translations; S2, obtaining the Chinese place name and address to be translated input by the user; S3, based on the proper name library and the common name library, performing multiple rounds of word segmentation and annotation and optimal segmentation scheme screening on the Chinese place name and address to be translated to obtain the optimal splitting and matching results; S4, performing attribute inference and translation generation based on context association constraints on the unannotated Chinese place name and address word segments to be translated in the optimal splitting and matching results to obtain the translation results of the unannotated place name and address word segments; S5, assembling the translation results of the unannotated place name and address word segments and the translation results of the annotated Chinese place name and address word segments to be translated in the optimal splitting and matching results and performing normalization processing based on a rule engine to obtain the complete Chinese place name and address translation results.

[0025] In the above method for translating geographical names and addresses by intelligent splitting and matching based on a proper name library and a common name library, in step S1, a proper name library and a common name library are constructed. Among them, the common name library stores common suffixes and type words in geographical names and addresses and their corresponding standard translations, and the proper name library stores common proper nouns in geographical names and addresses and their corresponding transliteration results or standard translations. It should be understood that Chinese geographical names and addresses have highly structured characteristics and are usually composed of a proper name part that refers to a specific entity (such as "Zhongguancun", "Xidan") and a common name part that indicates its administrative division, road type or building type (such as "district", "street", "building", "road", "hutong"). And these constituent elements often have established or officially stipulated standard translation methods during translation. Due to the lack of explicit delimiters in Chinese geographical names, it causes difficulties in segmentation and recognition during the translation process. Therefore, to lay the foundation for accurate translation of geographical names and addresses and provide high-quality prior knowledge for subsequent intelligent segmentation and matching. This application pre-collects and organizes the core constituent elements in geographical names and addresses and their standard translations, and constructs a proper name library and a common name library with wide coverage and authoritative content to achieve accurate recognition and translation of known geographical name elements. In the specific implementation process, the common name library will collect various types of administrative division names (such as province, city, district, county, town, township, village), road types (such as road, street, avenue, lane, hutong), point-of-interest types (such as square, park, school, hospital, shopping mall, residential community, building, unit, floor, house number), etc., common suffixes and type words, and equip them with internationally common or officially recommended standardized translations. For example, "road" corresponds to "Road", "district" corresponds to "District", "building" corresponds to "Building", and "number" corresponds to "No.". The construction of the proper name library focuses on collecting common and representative proper nouns, such as city names (Beijing, Shanghai), famous streets (Chang'an Avenue, Nanjing Road), well-known landmarks (the Oriental Pearl Tower, the Forbidden City), etc., and stores their widely accepted transliteration results (such as "Beijing" and "Nanjing Lu" based on Chinese pinyin) or existing standard translations (such as "The Forbidden City"). The establishment of the proper name library and the common name library can rely on official geographical information data, map service provider data, gazetteers, and various methods such as corpus mining and manual proofreading to ensure that the frequently occurring and standard-translated components in geographical names and addresses can be accurately and standardizedly translated, thereby significantly improving the lower limit of the overall translation quality and providing key clues for solving segmentation ambiguity problems.

[0026] In the aforementioned method for translating place names and addresses based on intelligent splitting and matching of a specific name database and a common name database, step S2 obtains the Chinese place name and address to be translated, input by the user. Specifically, due to the diverse sources of place names and addresses in actual application scenarios (such as government forms, logistics order numbers, and navigation input), problems arise, such as confusing formatting, omitting common names, and using colloquial expressions. For example, a user might enter "175 Chaoyangmennei Street, Beijing" or simply "Chaoyangmennei 175." Therefore, to accommodate multi-scenario input and adapt to subsequent structured parsing processes, this application provides standardized input for the subsequent translation process through a unified encoding conversion and noise filtering mechanism. In specific implementation, the system requires a multimodal input interface (supporting text, voice, and image OCR recognition), unified encoding of the input content (e.g., converting full-width to half-width), and address element completion (e.g., adding missing province and city prefixes). Furthermore, a threshold control for address length is established to prevent processing efficiency degradation caused by excessively long addresses. In this way, unstructured user input can be converted into a standardized string suitable for machine processing, laying the foundation for subsequent address segmentation and matching.

[0027] During the specific implementation process, the system receives raw input from different channels through a multimodal interface and performs pre-processing operations such as encoding unification, noise filtering, and address element completion on it, thereby converting unstructured user input into a structured, standardized string format. Among them, the design of the multimodal input interface is crucial. It not only supports traditional text input methods, but is also compatible with voice recognition and image OCR (optical character recognition) technology, allowing users to enter addresses in a variety of ways. For example, in the logistics industry, couriers may enter the delivery address by voice; while in mobile navigation applications, users may upload pictures containing address information by taking photos. In order to adapt to these different input methods, the system needs to have built-in corresponding parsing modules to perform semantic extraction and format normalization on the received voice text or OCR recognition results, respectively, to ensure that all inputs can be processed uniformly.

[0028] After supporting diverse input methods, the next step is to standardize the original input content. This process primarily involves converting full-width characters to half-width characters, normalizing punctuation, and removing duplicate words. For example, a user might use the full-width numeral "175" when entering "Jianguomenwai Street, Chaoyang District, Beijing," or insert extra spaces and line breaks into the address. If left unaddressed, these formatting differences will directly affect the accuracy of subsequent segmentation and matching. To this end, the system must incorporate a comprehensive set of encoding conversion rule libraries to identify various unconventional characters and replace them with standard formats. Furthermore, for some user input containing colloquial expressions, such as "Jianguomen that area" or "Dongzhimen that area," the system must also determine the likely formal address names based on the context and attempt to map them to standardized expressions so that subsequent processes can accurately identify common and proper noun components.

[0029] To improve the completeness and consistency of address recognition, this step also requires address element completion. In real-world applications, users often omit some hierarchical information in addresses for convenience. For example, they might enter "175 Chaoyangmennei Street" without the provincial prefix "Beijing," or simply "Chaoyangmennei 175" without specifying the street level. While this abbreviation is generally unambiguous for humans, it can lead to recognition failures or misjudgments for machine processing. Therefore, after receiving user input, the system should automatically complete missing administrative hierarchical information by combining the geolocation knowledge graph with historical data. For example, when detecting the word fragment "Chaoyangmennei Street," the system can infer the municipal district to which it belongs based on the existing geographic database and automatically add "Dongcheng District, Beijing" as a prefix, thereby constructing the complete address structure. This operation not only helps improve the accuracy of subsequent segmentation and matching but also ensures the standardization of the final translation results.

[0030] Furthermore, during address input, some noise is unavoidable, such as spelling errors, redundant words, and irrelevant symbols. For example, users may include explanatory phrases such as "Please deliver as soon as possible" or "Pay attention to the house number" in their addresses. Typing errors may also result in unconventional expressions such as "Chaoyangmen" or "Jianwai Street" appearing in the address. To address these issues, the system needs to integrate a natural language processing module to perform semantic analysis and anomaly detection on the input content. Words that clearly do not belong to the address context, such as greetings and directive phrases, should be automatically removed by the system. For place names suspected of being misspelled, a fuzzy matching algorithm can be used to compare them with a database of candidate words and attempt to correct them to the correct form. For example, if a user enters "Chaoyangmen," the system can recognize that it has a high similarity to "Chaoyangmen" and, based on the context, determine that it should be "Chaoyangmen." It can then correct the word to the correct form before proceeding to the next step.

[0031] Given the complexity and diversity of address input, this step also requires a length control mechanism to prevent processing efficiency degradation caused by extremely long addresses. In some special scenarios, users may enter a detailed address containing multiple levels of information, such as "Area A, Building 3, Intelligent Systems Laboratory, Institute of Automation, Chinese Academy of Sciences, 6A Zhongguancun Street, Haidian District, Beijing." While these addresses are complete, their length far exceeds the typical address range, potentially impacting subsequent processing speed and resource consumption. Therefore, the system needs to set a reasonable address length threshold and, when it exceeds the threshold, trigger automatic truncation or prompts to guide users to enter a more concise version. Address compression can also be used to simplify lengthy descriptive content into a standard format, for example, simplifying "Area A, Building 3, Intelligent Systems Laboratory, Institute of Automation, Chinese Academy of Sciences" to "Area A, Building 3, Institute of Automation, Chinese Academy of Sciences." This improves processing efficiency while preserving key information.

[0032] In the above-mentioned place name and address translation method based on the intelligent splitting and matching of the proper name library and the common name library, the step S3, based on the proper name library and the common name library, performs multiple rounds of word segmentation and annotation and screening of the optimal segmentation scheme on the Chinese place name and address to be translated to obtain the optimal splitting and matching result. Specifically, since Chinese place name and address are usually continuous character strings, lacking natural separators such as spaces, it is easy to cause segmentation ambiguity, which directly affects the recognition and translation of subsequent components; at the same time, different segmentation methods will correspond to different knowledge base matching results, thereby affecting the accuracy of translation. Therefore, in order to solve the segmentation ambiguity of Chinese place name and address, and to make the best use of the high-quality knowledge in the constructed proper name library and common name library, this application is based on the principle of maximum coverage of the knowledge base, through multiple rounds of word segmentation attempts of different granularities on the Chinese place name and address to be translated, and combined with the knowledge base for annotation, and then screens out the segmentation scheme that best reflects the address structure and obtains the most support from the knowledge base. Among them, Figure 3 Flowchart of sub-step S3 of the place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library according to the embodiment of the present application. Figure 3As shown, step S3 includes the steps of: S31, performing multiple rounds of word segmentation on the Chinese place name and address to be translated to obtain a candidate set of sequences of Chinese place name and address word segments to be translated; S32, based on the proper name library and the common name library, performing word segment query and type annotation on each sequence of Chinese place name and address word segments in the candidate set of sequences of Chinese place name and address word segments to be translated to obtain a candidate set of sequences of Chinese place name and address word segments to be translated with annotations; S33, calculating the proportion value of the annotated characters in the total characters in each sequence of Chinese place name and address word segments with annotations in the candidate set of sequences of Chinese place name and address word segments with annotations, and extracting the sequence of Chinese place name and address word segments with annotations with the highest proportion value as the optimal splitting and matching result.

[0033] Specifically, in step S31, multiple rounds of word segmentation are performed on the Chinese place name and address to be translated to obtain a candidate set of sequences of Chinese place name and address word segments to be translated. That is, the obtained Chinese place name and address to be translated (such as "No. 1 Courtyard, Zhongguancun Street, Haidian District, Beijing City") is processed using various word segmentation strategies or algorithms (for example, a word segmenter based on a statistical language model, word segmentation under different dictionary combinations, etc.) to generate a candidate set containing various possible segmentation results. For example, candidate 1: ["Beijing", "City", "Haidian", "District", "Zhongguancun", "Street", "1", "No.", "Courtyard"] may be obtained, candidate 2: ["Beijing City", "Haidian District", "Zhongguancun Street", "No. 1 Courtyard"], candidate 3: ["Beijing City", "Haidian District", "Zhongguancun", "Street", "No. 1", "Courtyard"], etc.

[0034] Specifically, in step S32, based on the proper name library and the common name library, word segment query and type annotation are performed on each sequence of Chinese place name and address word segments in the candidate set of sequences of Chinese place name and address word segments to be translated to obtain a candidate set of sequences of Chinese place name and address word segments to be translated with annotations. Specifically, for each sequence of Chinese place name and address word segments in the candidate set, the proper name library and the common name library are queried word segment by word segment: If the word segment exists in the common name library (such as "City", "District", "Street", "No.", "Courtyard"), it is annotated as "common name" and its standard translation in the target translation language is attached; if it exists in the proper name library (such as "Beijing", "Haidian", "Zhongguancun"), it is annotated as "proper name" and its transliteration or standard translation is attached.

[0035] Specifically, step S33 calculates the ratio of annotated characters to the total characters in each annotated Chinese place name and address word fragment sequence in the candidate set of annotated Chinese place name and address word fragment sequences to be translated, and selects the annotated Chinese place name and address word fragment sequence with the highest ratio as the optimal splitting and matching result. That is, after the annotating is completed, the total number of characters contained in all successfully annotated word fragments (i.e., matched in the proper name database or the general name database) in each Chinese place name and address word fragment sequence to be translated is calculated, and then divided by the total number of characters in the original input Chinese place name and address to be translated to obtain the ratio of annotated characters to the total characters, i.e., the knowledge base coverage. Next, from all candidate sequences, the annotated Chinese place name and address word fragment sequence to be translated with the highest ratio is extracted and selected as the optimal splitting and matching result. In this way, the limitations that may be brought about by a single word segmentation algorithm can be effectively overcome. By comparing multiple schemes and evaluating based on the knowledge base, the segmentation scheme that best conforms to the structural characteristics of place names and addresses and can be best explained by known knowledge can be screened out, thereby greatly improving the recognition accuracy and translation quality of the core components of place names and addresses, laying a solid foundation for the subsequent processing of unlabeled fragments.

[0036] In the above-mentioned place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library, step S4 performs attribute inference and translation generation based on context-related constraints on the unlabeled Chinese place name and address word segments to be translated in the optimal splitting and matching results to obtain translation results for the unlabeled place name and address word segments. It should be understood that due to the limited coverage of the proper name library and the common name library, they cannot include all emerging place names, specific store names, uncommon building names, or descriptive attributives, etc. As a result, in the optimal splitting and matching results, there will always be word segments that are not successfully labeled by the proper name library or the common name library (i.e., unregistered words or OOV segments). If a general translation engine is simply used for translation processing, due to the general engine's lack of understanding of the specific context of the place name and address, it cannot accurately determine whether these segments should be transliterated, translated, or used as descriptive additions, often resulting in poor translation results. Therefore, in order to break through the dictionary coverage limitation and achieve semantically coherent translation, this application further uses advanced natural language processing technology to perform contextual semantic association analysis on the optimal split matching results. By using the position of the unlabeled word fragment in the overall address, the semantic relationship of the context address word fragment and the attribute annotation information, the possible semantic category and attribute of the unlabeled word fragment are inferred, and on this basis, the unlabeled Chinese place name address word fragment to be translated is translated. Figure 4 Flowchart of sub-step S4 of the place name and address translation method based on intelligent splitting and matching of the proper name library and the common name library according to the embodiment of the present application. Figure 4As shown, the step S4 includes the steps of: S41, performing word-granularity semantic embedding coding on the optimal split matching result to obtain a sequence of semantic embedding coding vectors of place name and address word fragments to be translated; S42, extracting the semantic embedding coding vectors of the place name and address word fragments to be translated corresponding to the unlabeled Chinese place name and address word fragments to be translated from the sequence of semantic embedding coding vectors of the place name and address word fragments to be translated as the attribute query place name and address word fragment semantic embedding coding vectors; S43, performing semantic query response coding based on context association constraints on the sequence of the attribute query place name and address word fragment semantic embedding coding vectors and the place name and address word fragment semantic embedding coding vectors to be translated to obtain the attribute query place name and address word fragment context semantic query response coding vector; S44, determining the type inference result of the unlabeled Chinese place name and address word fragment to be translated based on the attribute query place name and address word fragment context semantic query response coding vector; S45, inputting the target translation language and the unlabeled Chinese place name and address word fragment to be translated and its type inference result into a large language model to obtain the translation result of the unlabeled place name and address word fragment.

[0037] Specifically, in a specific example of the present application, the step S41 includes: inputting the optimal split matching result into a BERT model containing position encoding to obtain a sequence of semantic embedding coding vectors of the place name and address word fragments to be translated. Specifically, considering that the original Chinese place name and address word fragments themselves are discrete text symbols, it is difficult for computers to directly understand their deep semantics and the complex connections between fragments. Therefore, in order to accurately capture the meaning of each place name and address word fragment to be translated in the specific address context, thereby providing high-quality feature representation for subsequent attribute inference and translation generation, the present application is based on the powerful semantic understanding ability principle of the pre-trained language model, by inputting the optimal split matching result (i.e., the sequence of Chinese place name and address word fragments to be translated with annotations) into a BERT model containing position encoding to achieve contextual awareness and semantic representation of each Chinese place name and address word fragment to be translated, and obtain a sequence of semantic embedding coding vectors of the place name and address word fragments to be translated. During its implementation, the BERT model first converts each place name and address word fragment to be translated and its annotated attribute information (if any) in the optimal split matching result into a low-dimensional dense word vector representation through a word embedding layer, and adds a position code to each word vector through a position embedding layer, so that the model can distinguish the relative positions of different word fragments in the address. Subsequently, the word vector incorporating the position information is input into the Transformer structure of the BERT model for multi-layer self-attention mechanism processing. In each layer, the model dynamically adjusts the representation of each word fragment by calculating the attention weights between different word fragments, so that it can capture the semantic associations with other word fragments. After being processed by the multi-layer Transformer structure, the semantic embedding representation of each place name and address word fragment to be translated not only incorporates its own semantics, but also its position in the overall address and its semantic relationship with surrounding word fragments, providing a rich feature foundation for subsequent attribute inference and translation generation.

[0038] Specifically, step S42 extracts the semantic embedding coding vectors of the unlabeled Chinese place-name and address word segments to be translated from the sequence of semantic embedding coding vectors of the place-name and address word segments to be translated as the attribute query place-name and address word segment semantic embedding coding vectors. That is, to focus on the unlabeled segments requiring attribute inference and translation generation, this application, based on the principle of target-driven data screening, extracts the semantic embedding coding vectors of the place-name and address word segments to be translated corresponding to the unlabeled segments from the complete sequence of semantic embedding coding vectors of the place-name and address word segments to be translated, thereby achieving targeted capture of target features and facilitating subsequent targeted processing.

[0039] Specifically, in step S43, semantic query response encoding based on context association constraints is performed on the sequence of the semantic embedding encoding vectors of the attribute query place name and address word segments and the semantic embedding encoding vectors of the to-be-translated place name and address word segments to obtain the context semantic query response encoding vectors of the attribute query place name and address word segments. It should be understood that since the attribute speculation of the unlabeled place name and address word segments needs to comprehensively consider the synergistic effect of their own semantics and the context semantic field (for example, when "county" in "Saga County" is used as a common name, "Saga" is more likely to be the overall proper name). Therefore, in order to construct a quantitative propagation channel for the semantic influence between place name and address word segments to model the semantic conduction relationship of place names and addresses, the present application further models the semantic association structure in the sequence of the semantic embedding encoding vectors of the to-be-translated place name and address word segments based on the graph encoding technology, and queries in the global semantic association graph of place names and addresses using the semantic embedding encoding vectors of the attribute query place name and address word segments to挖掘出未标注地名地址词片段与周围已标注地名地址词片段的潜在语义联系,从而利用图谱编码的语义传导机制来增强未标注地名地址词片段的语义表示,得到属性查询地名地址词片段上下文语义查询响应编码向量。其中, Figure 5 FIG. is a flowchart of sub-step S43 of the place name and address translation method based on intelligent splitting and matching of proper name libraries and common name libraries according to an embodiment of the present application. As Figure 5 shown, step S43 includes the steps of: S431, performing context association analysis between word segments on the sequence of the semantic embedding encoding vectors of the to-be-translated place name and address word segments to construct an association topology matrix between the semantic features of the to-be-translated place name and address word segments; S432, based on the association topology matrix between the semantic features of the to-be-translated place name and address word segments, performing graph structure context association perception on the sequence of the semantic feature vectors of the to-be-translated place name and address word segments to obtain an association encoding matrix of the semantic features of the to-be-translated place name and address word segments similar to a graph; S433, performing semantic query response encoding on the semantic embedding encoding vectors of the attribute query place name and address word segments and the association encoding matrix of the semantic features of the to-be-translated place name and address word segments similar to a graph to obtain the context semantic query response encoding vectors of the attribute query place name and address word segments.

[0040] More specifically, step S431 includes: First, performing semantic concentration encoding on each semantic feature vector in the sequence of the semantic feature vectors of the to-be-translated place name and address word segments to obtain a sequence of semantic feature concentration encoding vectors of the to-be-translated place name and address word segments, which is represented by the formula:

[0041] ;

[0042] ;

[0043] It should be noted that there seems to be some incomplete or unclear parts in the original text, especially the part "挖掘出未标注地名地址词片段与周围已标注地名地址词片段的潜在语义联系" which is not fully translated as there may be some missing words or incorrect grammar in the original. You may want to check and correct the original text for a more accurate translation. ;

[0044] in, A sequence representing the semantic feature vectors of the place name and address word segments to be translated, 、 、 and They represent the first, second, and third place names in the sequence of semantic feature vectors of the place name and address word segments to be translated. and semantic feature vectors of place name and address word segments to be translated, is the number of vectors in the sequence of semantic feature vectors of the place name and address word fragment to be translated, Indicates the calculation of the 2-norm of the vector, represents the ReLu activation function, and Represent the weight matrix and bias term of the semantic enrichment network respectively, A sequence representing the semantic feature condensed encoding vectors of the place name and address word segments to be translated, 、 、 and They represent the first, second, and third place names in the sequence of semantic feature condensed encoding vectors of the place name and address word fragments to be translated. and The semantic feature condensed encoding vector of the place name and address word fragments to be translated.

[0045] Specifically, semantic condensation coding removes irrelevant information, projects and refines the semantic feature vectors of each place-name or address word segment to be translated into a low-dimensional semantic space, extracts its most representative core semantics, and thus improves the accuracy of subsequent semantic relevance calculations. The resulting sequence of semantic feature condensation encoding vectors for the place-name or address word segments to be translated achieves low-dimensionality and semantic refinement, preserving the core semantic information of the word segments in a more concise and efficient form, providing a higher-quality feature foundation for subsequent semantic relevance analysis.

[0046] Then, the semantic correlation between any two semantic feature condensed coding vectors of the place-name and address word segments to be translated in the sequence of semantic feature condensed coding vectors of the place-name and address word segments to be translated is calculated to obtain a topological matrix of semantic feature correlation between the place-name and address word segments to be translated composed of multiple semantic correlations, which is expressed as follows:

[0047] ;

[0048] in, The first in the sequence of semantic feature condensed encoding vectors of place name and address word segments to be translated The semantic feature condensed encoding vector of the place name and address word fragment to be translated, and Represent the semantic association feature weight matrix and semantic association feature bias term respectively, express and The semantic association encoding vector between Represents the semantic association encoding vector The characteristic scale value of Represents the semantic association encoding vector The index of the eigenvalues at each position in , express and The semantic correlation between the first and second place names and address words in the topological matrix of the semantic features of the place name and address word segments to be translated is The element value at position.

[0049] Specifically, by quantifying the semantic relevance between the coded vectors of the semantic features of the place-name and address word segments to be translated, an initial relationship graph is explicitly constructed. The resulting topological matrix of semantic features between the place-name and address word segments to be translated captures the deep semantic relationships between word segments, improving the ability to recognize unique combinations of proper and general nouns in addresses. This provides a structured semantic relationship foundation for subsequent graph-structured contextual perception, addressing the problem that traditional word segmentation models lack the ability to specifically recognize unique combinations of place-name and address words.

[0050] More specifically, step S432 includes: first, inputting the correlation topology matrix between semantic features of the place-name and address word segments to be translated into a gated mask network to obtain a sparse correlation topology matrix between semantic feature nodes of the place-name and address word segments to be translated, which is expressed as follows:

[0051] ;

[0052] ;

[0053] in, represents the sigmoid activation function, represents the transpose of a vector, means dividing by position, and represent the mask weight vector and mask bias weight parameters respectively, express and The semantic association mask weight of the place name and address word segments to be translated, The first one in the sparse correlation topology matrix between the semantic feature nodes of the place name and address word segment to be translated The element value at position.

[0054] Specifically, by introducing a gated mask network, we adaptively sparsify the topological matrix of semantic features of place-name and address word segments to be translated, extracting the most critical semantic connections and optimizing the graph structure. The resulting sparse topological matrix of semantic feature nodes significantly reduces the computational complexity of subsequent graph convolutions, improves the model's robustness to noise, effectively prevents excessive information propagation (oversmoothing), and highlights truly important semantic relationships, making the graph's topology clearer, more robust, and closer to the underlying feature manifold.

[0055] Then, the sequence of semantic feature condensed coding vectors of the place-name and address word segments to be translated and the sparse association topology matrix between the semantic feature nodes of the place-name and address word segments to be translated are subjected to graph-like encoding based on a graph convolutional network model to obtain the graph-like association coding matrix of the semantic feature of the place-name and address word segments to be translated, which is expressed as follows:

[0056] ;

[0057] in, Represents the sparse correlation topology matrix between the semantic feature nodes of the place name and address word fragment to be translated, represents a graph convolutional network, It represents the graph-like association coding matrix of the semantic features of the place name and address word fragments to be translated.

[0058] That is, by utilizing the iterative neighborhood information aggregation and transformation mechanism of the graph convolutional network model, a structured deep embedding is performed on the sparse association topology matrix between the semantic feature nodes of the place-name and address word fragments to be translated and the condensed semantic feature encoding vector of the place-name and address word fragments to be translated. This captures the higher-order semantic relationships and global contextual associations between word fragments that go beyond simple similarity, enabling the representation of each word fragment to incorporate its multi-hop neighborhood structure information in the feature manifold. In this way, the generated graph-like association encoding matrix of the semantic features of the place-name and address word fragments to be translated achieves a holistic and structured deep representation of the semantic features of the place-name and address word fragments to be translated. Each row vector in the matrix contains the corresponding word fragment's position and semantic role in context, effectively capturing the high-order semantic dependencies and global structural information between word fragments, and providing a richer contextual semantic foundation for subsequent semantic query response encoding.

[0059] In particular, considering that the graph neural network model has realized the extraction of multi-order associations and overall architectural information under the graph structure for the graph-like association coding matrix of the semantic features of the place name and address word fragments to be translated, but due to the low-density characteristics introduced by the gated mask network into the graph topology, for each row vector in the graph-like association coding matrix of the semantic features of the place name and address word fragments to be translated, it is still expected that it will show convergent statistical characteristics as much as possible on the high-dimensional macro-distribution scale outside the graph structure, that is, it is expected to achieve a macro-comprehensive association of each row vector. Based on this, in a preferred example of the present application, the step S422123 includes: first, performing coupling optimization based on the global semantic field balance maintenance on each row vector in the graph-like association coding matrix of the semantic features of the place name and address word fragments to be translated to obtain an optimized graph-like association coding matrix of the semantic features of the place name and address word fragments to be translated.

[0060] Specifically, first, the measure semiring space is used to calculate the neighborhood rigid mapping function of each row vector in the graph-like association encoding matrix of the semantic features of the place name and address word fragment to be translated, so as to capture the complex nonlinear association represented by the quasi-transcendental covariance:

[0061] ;

[0062] in, The first one in the semantic feature-like graph association coding matrix of the place name and address word segment to be translated is represented by row vectors, Indicates the calculation of absolute value, represents the mean vector obtained by averaging the row vectors in the graph-like association coding matrix of the semantic features of the place name and address word fragment to be translated according to their positions, express The corresponding neighborhood rigid mapping vector.

[0063] Then, calculate the neighborhood rigid mapping vector Time-varying cumulative index of:

[0064] ;

[0065] in, express No. The eigenvalues of the position, represents the logarithmic function with base e, express The time-varying cumulative index of .

[0066] Finally, the time-varying cumulative index For the row vector Perform weighted optimization:

[0067] ;

[0068] in, Indicates the optimization of semantic features of place name and address word segments to be translated in the graph-like association coding matrix. A row vector.

[0069] That is, this application takes into account that the macro-statistical distribution law is dominated by high-order cumulative quantities. In order to achieve consistency in the scale of high-dimensional macro-distribution between different row vectors, it is necessary to constrain the global symmetric dynamics through dynamic accumulation measures under the rigid neighborhood framework, thereby ensuring the uniform transmission of macro-statistical characteristics under the graph structure and avoiding effects such as statistical loss of connection under the graph structure.

[0070] Then, the semantic embedding encoding vector of the attribute query place name and address word fragment and each row vector in the semantic feature imitation graph association encoding matrix of the optimized place name and address word fragment to be translated are subjected to feature interactive attention aggregation based on the feature query response strength to obtain the context semantic query response encoding vector of the attribute query place name and address word fragment, which is expressed as follows:

[0071] ;

[0072] ;

[0073] in, Represents the semantic embedding encoding vector of the attribute query place name and address word fragment, Indicates the optimization of semantic features of place name and address word segments to be translated in the graph-like association coding matrix. row vectors, express and The semantic interaction between them responds to the attention weights, Indicates point multiplication by position, Represents the encoding vector of the contextual semantic query response of the attribute query place name and address word fragment.

[0074] Specifically, an attention mechanism is introduced to enable interaction between the semantic embedding encoding vectors for attribute query place-name and address word fragments and the optimized semantic feature graph-like association encoding matrix for the to-be-translated place-name and address word fragments. This involves weighted aggregation of relevant node embeddings based on the strength of feature query responses, integrating query information with the semantic information of the to-be-translated word fragments through deep relationship modeling, to accurately locate and express the association between query intent and the to-be-translated content in a complex semantic space. Based on this, the resulting contextual semantic query response encoding vectors for attribute query place-name and address word fragments accurately reflect the location and relationship of the attribute query word fragments within the structured semantic space defined by the to-be-translated word fragments, dynamically capturing the contextual dependencies between word fragments and providing a precise semantic basis for attribute inference and translation generation of unlabeled word fragments.

[0075] Specifically, in a specific example of the present application, the step S44 includes: inputting the attribute query place name and address word fragment context semantic query response encoding vector into the attribute inference module based on the classifier to obtain the type inference result. It should be understood that the attribute query place name and address word fragment context semantic query response encoding vector contains the semantic features of the unlabeled place name and address word fragment in the overall address context and its association information with the surrounding word fragments. Based on this, in order to achieve accurate attribute inference of the unlabeled place name and address word fragment, the present application designs a classifier based on a neural network architecture to perform feature learning and attribute inference on the attribute query place name and address word fragment context semantic query response encoding vector. The classifier is trained on a large amount of place name and address data, and can effectively learn the semantic features of place name and address word fragments, and establish a mapping relationship between the semantic information of place name and address word fragments and attribute labels. During the attribute inference process, the classifier performs feature extraction and pattern recognition on the contextual semantic query response encoding vector of the input attribute query place name and address word fragment based on the prior knowledge learned through training, thereby accurately determining the type of unlabeled place name and address word fragment, such as proper name / common name and road name / area name / building type, etc., thereby providing key guidance information for subsequent translation processing and ensuring the accuracy of translation.

[0076] Specifically, in step S45, the target translation language, the unannotated Chinese toponym and address word segment to be translated, and its type speculation result are input into the large language model to obtain the translation result of the unannotated toponym and address word segment. Specifically, based on the prompting learning and text generation principle of the large language model (LLM), this application uses the original text of the unannotated segment, its speculated attributes, and the target translation language type as inputs together to drive the LLM to generate the final translation result. In the specific implementation process, assume that the unannotated segment is "Qinghe", its speculated attribute is "proper name - road name", and the target translation language is English. Then it is filled into the preset template prompt to construct a prompt for the large language model: "Please translate the Chinese toponym segment 'Qinghe' into English. The attribute of this segment is 'proper name - road name'. The large language model (such as the GPT series) translates the toponym segment 'Qinghe' according to the input prompt information, utilizes its own language understanding and generation capabilities, and generates a more idiomatic English expression considering its attribute of 'proper name - road name', such as 'Qinghe', thus ensuring the accuracy and authenticity of the translation. In this way, guided by the word segment attribute information and combined with the powerful language generation capabilities of the large language model, the accuracy of translation can be effectively improved, and the problem of ambiguity caused by different attributes of the same toponym in different contexts can be avoided. For example, 'Xingfu' is translated as 'Xingfu' when it is used as a road name, while it may be translated as 'Happiness' when it is used as a community name or scenic spot name.

[0077] In the above-mentioned method for translating geographical names and addresses by intelligent splitting and matching based on a proper name library and a common name library, in step S5, the translation results of the unlabeled geographical name and address word segments and the translation results of the labeled Chinese geographical name and address word segments in the optimal splitting and matching results are assembled and normalized based on a rule engine to obtain the complete translation result of the Chinese geographical name and address. Specifically, in order to meet the requirements of the Geographic Information System (GIS) for the standardized address format, the present application further performs structural reorganization of the translation components, and calibrates the grammar and format of the translation results through a rule engine. In the specific implementation process, the rule engine needs to build in three types of specifications: 1) word order rules (e.g., "No. X, Road Y" in Chinese is translated as "Y Road No. X"); (Like adding a space between proper names and common names, and using "No." for numbers); 3) capitalization rules (e.g., the first letter of the common name is capitalized). For example, for the segmentation result "Haidian District (Haidian District), North West 3rd Ring Road (North West 3rd Ring Road), A2 (A2)", the engine will reorganize it into "A2, North West 3rd Ring Road, Haidian District, Beijing" according to the English expression habit of "from large range to small range". At the same time, the engine will detect and repair the problem of inconsistent abbreviations between "Road" and "Rd.". In this way, it can be ensured that the finally output translation result not only conforms to the grammar norms of the target language (e.g., English addresses are arranged from small to large), but also meets the requirements of the GIS system for standardized naming (e.g., facilitating database indexing and spatial matching), significantly improving machine readability.

[0078] In summary, the method for translating geographical names and addresses by intelligent splitting and matching based on a proper name library and a common name library according to the embodiments of the present application is elucidated. First, a common name library containing standard translations and a proper name library of transliterated proper names are constructed. After multiple rounds of segmentation of the geographical names and addresses to be translated, the word segments in each segmentation scheme are type-labeled through dual-library matching, and the optimal segmentation scheme is determined based on the labeled character coverage. On this basis, a context association inference mechanism is further introduced to perform context semantic association analysis on the unlabeled word segments in the optimal segmentation scheme. After inferring the word segment attribute type, a large language model is called to generate an adapted translation. Finally, the rule engine performs structural reorganization on the dual-library labeled translations and the inferred translations to form a standardized standard address translation. Through multiple rounds of word segmentation and labeling and screening of the optimal segmentation scheme, this method can effectively solve the problem of segmentation ambiguity, accurately identify the core components of the address, and thus improve the translation quality.

[0079] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in the present invention are merely illustrative and non-limiting, and should not be construed as necessarily possessed by each embodiment of the present invention. Furthermore, the specific details of the above embodiments are provided for illustrative purposes and to facilitate understanding, and are not intended to be limiting. These details do not necessarily limit the present invention to being implemented using these specific details.

[0080] 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.

[0081] 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.

[0082] 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.

[0083] 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. A place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library, characterized in that: include: Constructing a proper name database and a common name database, wherein the common name database stores common suffixes and type words in place names and addresses and their corresponding standard translations, and the proper name database stores common proper nouns in place names and addresses and their corresponding transliteration results or standard translations; Get the Chinese place name and address to be translated entered by the user; Based on the proper name database and the common name database, multiple rounds of word segmentation and optimal segmentation scheme screening are performed on the Chinese place name address to be translated to obtain the optimal splitting and matching result, which includes: Performing multiple rounds of word segmentation processing on the Chinese place name and address to be translated to obtain a candidate set of word segment sequences of the Chinese place name and address to be translated; Based on the proper name database and the common name database, performing word fragment query and type annotation on each Chinese place name and address word fragment sequence to be translated in the candidate set of Chinese place name and address word fragment sequences to be translated to obtain a candidate set of annotated Chinese place name and address word fragment sequences to be translated; Calculating the ratio of the annotated characters to the total characters in each annotated Chinese place name and address word fragment sequence to be translated in the candidate set of the annotated Chinese place name and address word fragment sequences to be translated, and extracting the annotated Chinese place name and address word fragment sequence to be translated with the highest ratio as the optimal splitting and matching result; Performing attribute inference and translation generation based on context association constraints on the unlabeled Chinese place name and address word segments to be translated in the optimal splitting and matching results to obtain translation results of the unlabeled place name and address word segments; The translation results of the unlabeled place name and address word segments and the translation results of the labeled Chinese place name and address word segments to be translated in the optimal split matching results are assembled and normalized based on a rule engine to obtain a complete Chinese place name and address translation result.

2. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 1 is characterized in that: Performing attribute inference and translation generation based on contextual association constraints on the unlabeled Chinese place name and address word segments to be translated in the optimal splitting and matching results to obtain translation results of the unlabeled place name and address word segments, including: Performing word-granularity semantic embedding coding on the optimal splitting and matching results to obtain a sequence of semantic embedding coding vectors of place name and address word segments to be translated; Extracting the semantic embedding coding vector of the place name and address word segment to be translated corresponding to the unlabeled Chinese place name and address word segment to be translated from the sequence of semantic embedding coding vectors of the place name and address word segment to be translated as the semantic embedding coding vector of the attribute query place name and address word segment; Performing semantic query response encoding based on context association constraints on the sequence of the semantic embedding encoding vector of the attribute query place name and address word fragment and the semantic embedding encoding vector of the place name and address word fragment to be translated to obtain a context semantic query response encoding vector of the attribute query place name and address word fragment; Determining a type inference result of the unlabeled Chinese place name and address word segment to be translated based on the attribute query place name and address word segment context semantic query response encoding vector; The target translation language and the unlabeled Chinese place name and address word segments to be translated and the type inference results thereof are input into a large language model to obtain a translation result of the unlabeled place name and address word segments.

3. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 2 is characterized in that: Performing word-granular semantic embedding coding on the optimal split matching result to obtain a sequence of semantic embedding coding vectors of the place name and address word segments to be translated, including: The optimal split matching result is input into a BERT model including position encoding to obtain a sequence of semantic embedding encoding vectors of the place name and address word fragment to be translated.

4. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 3 is characterized in that: Performing semantic query response encoding based on contextual association constraints on the sequence of the semantic embedding encoding vector of the attribute query place name and address word fragment and the semantic embedding encoding vector of the place name and address word fragment to be translated to obtain the attribute query place name and address word fragment context semantic query response encoding vector, including: Performing contextual association analysis between word segments on the sequence of semantic embedding coding vectors of the place-name and address word segments to be translated to construct a topological matrix of association between semantic features of the place-name and address word segments to be translated; Based on the topological matrix of the association between the semantic features of the place-name and address word segments to be translated, performing graph structure context association perception on the sequence of the semantic feature vectors of the place-name and address word segments to be translated to obtain a graph-like association coding matrix of the semantic features of the place-name and address word segments to be translated; The semantic embedding coding vector of the attribute query place name and address word fragment and the semantic feature imitation graph association coding matrix of the place name and address word fragment to be translated are subjected to semantic query response coding to obtain the context semantic query response coding vector of the attribute query place name and address word fragment.

5. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 4 is characterized in that: Performing contextual association analysis between word segments on the sequence of semantic embedding coding vectors of the place-name and address word segments to be translated to construct a topological matrix of association between semantic features of the place-name and address word segments to be translated, including: Performing semantic condensation coding on each semantic feature vector of the place-name and address word fragment to be translated in the sequence of semantic feature vectors of the place-name and address word fragment to be translated to obtain a sequence of semantic feature condensation coding vectors of the place-name and address word fragment to be translated; The semantic correlation between any two semantic feature condensed coding vectors of the place name and address word fragments to be translated in the sequence of semantic feature condensed coding vectors of the place name and address word fragments to be translated is calculated to obtain a topological matrix of semantic feature correlation between the place name and address word fragments to be translated composed of multiple semantic correlations.

6. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 5 is characterized in that: Based on the topological matrix of the association between the semantic features of the place-name and address word segments to be translated, a graph structure context association perception is performed on the sequence of the semantic feature vectors of the place-name and address word segments to be translated to obtain a graph-like association coding matrix of the semantic features of the place-name and address word segments to be translated, including: Inputting the correlation topology matrix between semantic features of the place-name and address word segments to be translated into a gated mask network to obtain a sparse correlation topology matrix between semantic feature nodes of the place-name and address word segments to be translated; The sequence of semantic feature condensed coding vectors of the place name and address word fragments to be translated and the sparse association topology matrix between the semantic feature nodes of the place name and address word fragments to be translated are subjected to graph-like encoding based on a graph convolutional network model to obtain the graph-like association coding matrix of the semantic feature of the place name and address word fragments to be translated.

7. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 6 is characterized in that: Performing semantic query response encoding on the semantic embedding encoding vector of the attribute query place name and address word segment and the semantic feature imitation graph association encoding matrix of the place name and address word segment to be translated to obtain the context semantic query response encoding vector of the attribute query place name and address word segment, including: Performing coupling optimization based on global semantic field balance preservation on each row vector in the semantic feature graph-like association coding matrix of the place name and address word segments to be translated to obtain an optimized graph-like association coding matrix of the semantic feature of the place name and address word segments to be translated; The semantic embedding coding vector of the attribute query place name and address word fragment and each row vector in the optimized semantic feature imitation graph association coding matrix of the place name and address word fragment to be translated are subjected to feature interaction attention aggregation based on feature query response strength to obtain the context semantic query response coding vector of the attribute query place name and address word fragment.

8. The place name and address translation method based on intelligent splitting and matching of a proper name library and a common name library according to claim 7 is characterized in that: Determining a type inference result of the unlabeled Chinese place name and address word segment to be translated based on the attribute query place name and address word segment context semantic query response encoding vector includes: The attribute query place name and address word fragment context semantic query response encoding vector is input into the classifier-based attribute inference module to obtain the type inference result.

Images