Real-time place name and address multi-language translation system and method based on neural network model

Abstract

The invention discloses a real-time place name and address multi-language translation system and method based on a neural network model, and relates to the technical field of multi-language translations, and the method comprises the steps: firstly extracting a topological relation of address components through structured analysis, and then carrying out the deep modeling of cross-component context dependence through employing a deep learning-based neural network model, deep semantic representation of the address text is mined through cross-component semantic association learning, and place name disambiguation based on dynamic semantic reasoning is realized based on the deep semantic representation; according to the method, the problem of semantic deviation in complex scenes such as homonymy, different places and different names is solved, the accuracy of judging the ambiguity of geographic entities is remarkably improved, the translation result can dynamically adapt to place name expression habits of different languages, translation ambiguity caused by limitation of a static knowledge base or imperfect rules is effectively eliminated, and the translation efficiency is improved. And a more accurate multi-language address conversion effect is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of multilingual translation, and more specifically, to a real-time multilingual translation system and method of place names and addresses based on a neural network model. Background Art

[0002] In the multilingual translation process of place names and addresses, place name ambiguity is a very core and common problem. Existing technologies usually rely on rule matching, static geographic knowledge base or shallow semantic analysis to deal with place name ambiguity. However, these methods have significant limitations in practical applications. Traditional rule matching methods are difficult to adapt to complex and changing contextual scenarios, especially when place names are ambiguous. Relying solely on keyword matching can easily lead to mistranslation, which directly leads to inaccurate translation and affects the user's ability to obtain correct information.

[0003] In addition, the existing technology uses address context information at the shallow feature matching level and fails to establish a cross-component semantic reasoning mechanism, which leads to semantic deviation in translation results when facing complex situations such as homonymous places or homonymous places. These problems together lead to bottlenecks in the accuracy of eliminating place name ambiguity and multilingual adaptability of the existing technology.

[0004] Therefore, an optimized real-time multilingual translation method of place names and addresses based on a neural network model is needed to solve the above technical problems. Summary of the invention

[0005] According to one aspect of the present application, a method for real-time multilingual translation of place names and addresses based on a neural network model is provided, which includes: Get the address text entered by the user; Use the address parsing model to perform structured parsing on the address text to obtain a set of address components; Selecting an address component to be enhanced from a set of address components and using other address components in the set of address components as context information of the address component to be enhanced; Performing semantic embedding coding on the address component to be strengthened to obtain the semantic embedding coding features of the address component to be strengthened, and semantically enhancing the semantic embedding coding features of the address component to be strengthened based on the context information of the address component to be strengthened to obtain the semantic enhancement coding features of the address component to be strengthened; Based on the semantic enhancement encoding features of the address components to be enhanced, the target language address text is obtained.

[0006] According to another aspect of the present application, a real-time multilingual translation system of place names and addresses based on a neural network model is provided, which includes: An address text acquisition module is used to acquire the address text input by the user; An address parsing module, used for performing structured parsing on an address text using an address parsing model to obtain a set of address components; A semantic encoding module, used for semantically encoding the address components to be strengthened and the context information of the address components to be strengthened in the set of address components respectively to obtain semantic embedding encoding features of the address components to be strengthened and semantic encoding features of the context information of the address components to be strengthened; A semantic enhancement module, for performing semantic enhancement on the semantic embedding coding features of the address component to be enhanced based on the semantic coding features of the context information of the address component to be enhanced to obtain the semantic enhancement coding features of the address component to be enhanced; The target language address text generation module is used to obtain the target language address text based on the semantic enhancement coding features of the address components to be enhanced.

[0007] Beneficial effects of this application: Compared with the prior art, this application first extracts the topological relationship of address components through structured analysis, and then uses a neural network model based on deep learning to deeply model the contextual dependencies across components, so as to mine the deep semantic representation of the address text through semantic association learning across components, and based on this, realize place name disambiguation based on dynamic semantic reasoning. This application solves the problem of semantic deviation in complex scenarios such as the same name in different places and the same place with different names, significantly improves the accuracy of distinguishing the ambiguity of geographic entities, enables the translation results to dynamically adapt to the place name expression habits of different languages, effectively eliminates translation ambiguity caused by the limitations of static knowledge bases or imperfect rules, and achieves a more accurate multilingual address conversion effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used 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 accompanying drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 It is a flowchart of a method for real-time multi-language translation of place names and addresses based on a neural network model according to an embodiment of the present application; Figure 2 A data flow diagram of a real-time multi-language translation method for place names and addresses based on a neural network model according to an embodiment of the present application; Figure 3 It is a flowchart of sub-step S4 of the method for real-time multi-language translation of place names and addresses based on a neural network model according to an embodiment of the present application; Figure 4 It is a flowchart of sub-step S42 of the method for real-time multi-language translation of place names and addresses based on a neural network model according to an embodiment of the present application; Figure 5 It is a block diagram of a real-time multi-language translation system of place names and addresses based on a neural network model according to an embodiment of the present application. DETAILED DESCRIPTION

[0010] 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 here.

[0011] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprises" and "includes" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0012] Although the present application makes various references to certain modules in the system according to the 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 only illustrative, and different aspects of the system and method can use different modules.

[0013] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps may be removed from these processes.

[0014] 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 here.

[0015] 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 the relevant national data protection laws and policies and with the authorization of the authority manager.

[0016] In the technical solution of the present application, a real-time multi-language translation method of place names and addresses based on a neural network model is proposed. Figure 1 The present invention is a flowchart of a method for real-time multi-language translation of place names and addresses based on a neural network model according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of a real-time multi-language translation method of place names and addresses based on a neural network model according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the real-time multilingual translation method of place names and addresses based on a neural network model includes the following steps: S1, obtaining an address text input by a user; S2, using an address parsing model to perform structured parsing on the address text to obtain a set of address components; S3, selecting an address component to be strengthened from the set of address components and using other address components in the set of address components as context information of the address component to be strengthened; S4, performing semantic embedding coding on the address component to be strengthened to obtain a semantic embedding coding feature of the address component to be strengthened, and semantically enhancing the semantic embedding coding feature of the address component to be strengthened based on the context information of the address component to be strengthened to obtain a semantic enhancement coding feature of the address component to be strengthened; S5, obtaining the target language address text based on the semantic enhancement coding feature of the address component to be strengthened.

[0017] In particular, S1 and S2 obtain the address text entered by the user, and use the address resolution model to perform structured resolution on the address text to obtain a set of address components. Specifically, the original address text entered by the user often presents nonlinear arrangement characteristics, and may contain nested expressions, mixed language symbols or regional abbreviation rules. These non-standardized expressions directly hinder the effectiveness of subsequent semantic analysis and ambiguity elimination. By performing structured resolution on the address text, continuous natural language strings can be cut into address components with clear type labels (such as country, province, city, street, house number, etc.), so that the system can establish spatial logical relationships and semantic dependency paths between components. It is worth mentioning that the address resolution model usually adopts a hybrid neural network architecture that integrates a multi-head attention mechanism and a conditional random field. In one example, the model first performs multi-granular word segmentation on the address text input by the user, where the system can adaptively adjust the word segmentation strategy for different language characteristics (such as character-level processing based on BERT for Chinese and subword segmentation for English); then the boundary features and type features of the address components are captured through a bidirectional Transformer encoder; in addition, the model integrates a geographic entity recognition module that can distinguish between common words and geographic entities, and dynamically links to the geographic knowledge graph to verify the components. In particular, when the user input has format errors or missing components, the model will learn robust features through synthetic data generated by adversarial training, automatically complete necessary components or trigger an interactive verification mechanism. The set of address components generated by structured parsing not only provides discrete processing units for subsequent semantic reinforcement encoding, but also builds a semantic reasoning channel across components.

[0018] In particular, the S3 selects the address component to be strengthened from the set of address components and uses other address components in the set of address components as the context information of the address component to be strengthened. Specifically, the semantics of each component in the address text does not exist in isolation, but constitutes a complete spatial semantic network through the topological relationship with other components. The traditional method is difficult to establish such cross-level semantic dependencies by treating the address components as parallel sets. In the technical solution of the present application, the i-th address component is extracted from the set of address components as the address component to be strengthened, and the other address components in the set of address components are used as the context information of the address components to be strengthened. That is, by dynamically constructing the "target component-context" association framework, the neural network can flexibly capture cross-level semantic dependencies for each address component, thereby providing a computable reasoning unit for subsequent semantic enhancement and ambiguity resolution.

[0019] In particular, the S4 performs semantic embedding coding on the address component to be strengthened to obtain the semantic embedding coding feature of the address component to be strengthened, and semantically enhances the semantic embedding coding feature of the address component to be strengthened based on the context information of the address component to be strengthened to obtain the semantic enhancement coding feature of the address component to be strengthened. Figure 3 As shown, the S4 includes: S41, using the embedding matrix of the address component to be strengthened to perform semantic embedding coding on the address component to be strengthened, so as to obtain the semantic embedding coding vector of the address component to be strengthened and use it as the semantic embedding coding feature of the address component to be strengthened; S42, based on the context information of the address component to be strengthened, semantically enhancing the semantic embedding coding feature of the address component to be strengthened, so as to obtain the semantic enhancement coding feature of the address component to be strengthened.

[0020] Specifically, the S41 uses the embedding matrix of the address component to be strengthened to perform semantic embedding coding on the address component to be strengthened, so as to obtain the semantic embedding coding vector of the address component to be strengthened and use it as the semantic embedding coding feature of the address component to be strengthened. Considering that the traditional method cannot effectively capture its potential multi-level semantic features when directly inputting the address component in the form of a string into the translation model, such geographic entities can be mapped to a high-dimensional vector space through semantic embedding coding, so that they form a distributed representation associated with administrative divisions and geographical levels in the semantic space. Therefore, in the technical solution of the present application, the embedding matrix of the address component to be strengthened is used to perform semantic embedding coding on the address component to be strengthened, so as to obtain the semantic embedding coding vector of the address component to be strengthened and use it as the semantic embedding coding feature of the address component to be strengthened. Among them, by mapping the discrete address components to the continuous vector space defined by the pre-trained embedding matrix of the address component to be strengthened, the system can capture its implicit geographical attributes, functional classification and cultural association characteristics. In one example, first, the embedding matrix of the address component to be strengthened is obtained based on the pre-training of a massive multilingual address corpus, and each row vector of the embedding matrix of the address component to be strengthened corresponds to the projection of a specific address component in the multi-layer semantic space. When processing address components to be enhanced (such as "Chaoyang District"), the system obtains its basic embedding vector through matrix lookup operations, and then superimposes attribute embedding based on geographic knowledge graph (such as administrative division level, population density characteristics) and context-aware location encoding (such as the strong association between "Chaoyang District" and business district characteristics in Beijing addresses). The semantic embedding encoding vector of the address component to be enhanced generated by semantic embedding encoding not only carries the lexical information of the address component, but also contains its spatial level, functional type and cross-language alignment clues. This deep semantic encoding mechanism provides a high-information-density feature basis for subsequent contextual semantic enhancement, significantly improving the system's ability to resolve the ambiguity of complex place names.

[0021] Specifically, the S42 semantically enhances the semantic embedding coding features of the address component to be enhanced based on the context information of the address component to be enhanced, so as to obtain the semantic enhancement coding features of the address component to be enhanced. Figure 4 As shown, the S42 includes: S421, semantically encoding the context information of the address component to be strengthened to obtain a semantic encoding vector of the context information of the address component to be strengthened and use it as a semantic encoding feature of the context information of the address component to be strengthened; S422, based on the semantic encoding feature of the context information of the address component to be strengthened, semantically enhancing the semantic embedding encoding feature of the address component to be strengthened to obtain the semantic enhancement encoding feature of the address component to be strengthened.

[0022] More specifically, the S421 semantically encodes the context information of the address component to be strengthened to obtain the semantic encoding vector of the context information of the address component to be strengthened and serves as the semantic encoding feature of the context information of the address component to be strengthened. In an embodiment of the present application, the specific steps of semantically encoding the context information of the address component to be strengthened are as follows: semantic embedding encoding is performed on each address component in the context information of the address component to be strengthened to obtain a context sequence of the semantic embedding encoding vector of the place name component; the context sequence of the semantic embedding encoding vector of the place name component is contextually semantically encoded based on a bidirectional LSTM model to obtain a semantic encoding vector of the context information of the address component to be strengthened. Specifically, the context information of the address text is not a simple set of adjacent components, but contains complex spatial topological logic and geographical affiliation. The traditional unidirectional recurrent neural network can only capture dependencies in a single direction, and cannot synchronously perceive the jurisdictional constraints of the superior administrative division on the subordinate place name and the reverse verification of the subordinate component on the superior semantics, resulting in directional deviations in semantic modeling. Therefore, in the technical solution of the present application, the context information of the address component to be strengthened is contextually semantically encoded based on a bidirectional LSTM model to obtain a semantic encoding vector of the context information of the address component to be strengthened. That is, through bidirectional time series modeling, the bidirectional influence of contextual components in spatial hierarchy and grammatical structure is dynamically integrated to construct contextual semantic representation with global perception ability. Specifically, the bidirectional LSTM captures the progressive spatial hierarchy of address components (such as "province→city→district→street") through forward propagation, and extracts the reverse geographical affiliation features (such as "street←district←city←province") through back propagation. This bidirectional information fusion enables the model to establish a geographical hierarchy attenuation relationship and achieve accurate geographical location disambiguation, fundamentally breaking through the limitation of traditional methods' reliance on local window information.

[0023] More specifically, the S422, based on the semantic coding features of the context information of the address component to be strengthened, semantically enhances the semantic embedded coding features of the address component to be strengthened, so as to obtain the semantic enhanced coding features of the address component to be strengthened. Specifically, when the traditional method simply concatenates or weighted sums the context features with the target component features, it is unable to effectively capture this implicit association across levels and dimensions, resulting in the loss of key context clues in the semantic enhancement process. Therefore, in the technical solution of the present application, based on the semantic coding features of the context information of the address component to be strengthened, the semantic embedded coding features of the address component to be strengthened are semantically enhanced to obtain the semantic enhanced coding features of the address component to be strengthened. That is, a graph structure topological space of the semantic kernel association between the address component to be strengthened and the context is established in the feature principal component dimension, so as to deeply explore the fine-grained semantic interaction rules of the address component to be strengthened, so as to obtain the semantic enhanced coding features of the address component to be strengthened. Specifically, firstly, based on the principal component kernel association coding network, the principal component vectors of the component to be strengthened (such as the administrative division feature axis of "Hongqiao Road") and the principal component vectors of the context component (such as the economic function axis of "Changning District, Shanghai") are mapped to a high-dimensional association space to generate a kernel association code that reflects the intensity of cross-component interaction. This decoupling operation not only eliminates feature redundancy, but also enables subsequent graph structure modeling to focus on the essential association pattern between address components; then, through the principal component kernel association coding network, the system maps the decoupled principal component vectors of the semantic features of the address components to a high-dimensional association space to break through the dependence of the traditional attention mechanism on explicit similarity calculation and capture the abstract semantic association between address components; then, a dynamic graph structure is constructed through the association topology matrix of the kernel association strength evaluation factor of the address component to be strengthened, in which the nodes represent the principal component feature dimensions (such as the postal code feature axis and the road grade axis), and the edge weights are determined by the kernel association strength; then, through the multi-hop information propagation of the graph convolutional neural network, the system can capture the semantic resonance effect across levels and establish long-distance associations between address components. This semantic enhancement mechanism based on graph topology enables the differentiated features of homonymous address components in different contexts to be explicitly expressed, and generates completely different translation vectors through differential information aggregation through graph convolution, thus achieving accurate semantic mapping in cross-language scenarios.

[0024] Specifically, first, the semantic coding vector of the context information of the address component to be strengthened and the semantic embedding coding vector of the address component to be strengthened are subjected to feature principal component analysis to obtain a set of principal component coding vectors of the context semantic features of the address component to be strengthened and a set of principal component coding vectors of the semantic features of the address component to be strengthened. Specifically, there is multicollinearity in the original feature space between the context semantic coding of the address component (such as the temporal features generated by the bidirectional LSTM) and the semantic embedding coding (such as the distributed representation of the embedding matrix mapping). When the traditional method directly splices these high-dimensional vectors, the collinearity between different semantic dimensions (such as the correlation between geographical location and administrative divisions) will make it difficult for the model to focus on key distinguishing features, and this redundant association will interfere with the subsequent cross-component semantic interaction modeling. Therefore, in the technical solution of the present application, in order to achieve semantic decoupling and information purification of the feature space, the semantic coding vector of the context information of the address component to be strengthened and the semantic embedding coding vector of the address component to be strengthened are subjected to feature principal component analysis to obtain a set of principal component coding vectors of the context semantic features of the address component to be strengthened and a set of principal component coding vectors of the semantic features of the address component to be strengthened. By using the principal component analysis technology to perform orthogonal transformation on the semantic coding vector of the context information of the address component to be strengthened and the semantic embedding coding vector of the address component to be strengthened, the system can separate independent principal component dimensions from the complex feature interweaving state. These dimensions carry the core semantic attributes of the address component (such as the geographic coordinate feature of "Paris" as a city entity) and the context constraint relationship (such as the directionality of "Texas" to the North American geographic reference system). In the specific implementation process, a hierarchical orthogonal feature reconstruction strategy is adopted to perform covariance matrix decomposition on the semantic coding vector of the context information of the address component to be strengthened and the semantic embedding coding vector of the address component to be strengthened, and the principal component axis with the highest variance contribution rate is selected by eigenvalue sorting. In this way, not only the redundant associations between the address component features are eliminated, but also the feature dimensions with clear semantic orientation are reconstructed. In a specific example of the present application, the following principal component analysis formula is used to perform feature principal component analysis on the semantic coding vector of the context information of the address component to be strengthened and the semantic embedding coding vector of the address component to be strengthened, so as to obtain a set of principal component coding vectors of the context semantic features of the address component to be strengthened and a set of principal component coding vectors of the semantic features of the address component to be strengthened; wherein, the principal component analysis formula is: ; ; in, is the semantic encoding vector of the context information of the address component to be strengthened, is the semantic embedding coding vector of the address component to be strengthened, is the characteristic principal component analysis, is a set of principal component encoding vectors of the contextual semantic features of the address component to be strengthened, is a set of principal component encoding vectors of the semantic features of the address components to be strengthened, are respectively the first, second, and third principal component encoding vectors of the contextual semantic features of the address component to be strengthened. and The principal component encoding vector of the contextual semantic features of the address components to be strengthened, are respectively the first, second, and third principal component encoding vectors of the semantic features of the address components to be strengthened. and The principal component encoding vector of the semantic features of the address components to be strengthened, is the diagonal matrix of semantic features related to the scheduling demand time, and They are and The corresponding eigenvalues ​​are is the diagonal matrix of semantic features of the address components to be strengthened, and They are and The corresponding eigenvalues.

[0025] Next, each corresponding set of principal component coding vectors of the contextual semantic features of the address components to be strengthened and the principal component coding vectors of the semantic features of the address components to be strengthened are respectively subjected to principal component kernel association coding to obtain a set of kernel association coding vectors between principal components of the semantic features of the address components to be strengthened. The linear combination of the set of principal component coding vectors of the contextual semantic features of the address components to be strengthened and the set of principal component coding vectors of the semantic features of the address components to be strengthened in a low-dimensional orthogonal space cannot capture complex semantic dependencies. Traditional linear association cannot effectively model the nonlinear coupling relationship between the two in the feature dimension, resulting in the inability to distinguish subtle differences between, for example, multiple cities with the same name. Therefore, in the technical solution of the present application, in order to construct a hyperplane association topology between the main component features of the address, so as to learn the cross-dimensional semantic resonance mode through a deep neural network, in the technical solution of the present application, each corresponding set of the main component coding vectors of the contextual semantic features of the address components to be strengthened and the main component coding vectors of the main component coding vectors of the semantic features of the address components to be strengthened are respectively subjected to the main component kernel association coding, so as to obtain the set of kernel association coding vectors between the main components of the semantic features of the address components to be strengthened. Specifically, the system projects the decoupled principal component pairs (such as the cultural naming feature axis of "Chang'an Avenue" and the administrative division feature axis of "Xi'an City") to the high-dimensional association space through nonlinear mapping, so that the semantic relationship that was originally linearly inseparable in the low-dimensional space (such as the road naming rules under the specific historical background of Beijing) can be explicitly expressed. In one example, a deep coding architecture with heterogeneous feature fusion can be adopted. First, an independent fully connected neural network channel is constructed for each pair of principal components (such as the "landmark function" axis of the component to be strengthened and the "administrative division level" axis of the context component), and the cross-dimensional association pattern is learned through multi-layer nonlinear transformation; then, the system uses the attention gating mechanism to dynamically weight the association encoding, so that the differentiated association patterns of the same-name components in different contexts can be explicitly expressed. This deep association mechanism enables the system to handle extremely complex scenarios of the same name in different places (such as the 28 "San Jose" in the world), and learns the unique association pattern between the context principal components of each country, state and province of "San Jose" and the city name feature axis through the kernel network, so as to dynamically select the naming conventions that conform to the cultural habits of the target language during translation.In a specific example of the present application, the set of principal component coding vectors of contextual semantic features of address components to be strengthened and the set of principal component coding vectors of semantic features of address components to be strengthened are respectively subjected to principal component kernel association coding for each corresponding group of principal component coding vectors of contextual semantic features of address components to be strengthened and principal component coding vectors of semantic features of address components to be strengthened, so as to obtain a set of kernel association coding vectors between principal components of semantic features of address components to be strengthened; wherein, the association coding formula is:. ; in, represents the one-norm of a vector, and They represent trainable weighted hyperparameters, is the first one in the set of kernel association coding vectors between the principal components of the semantic features of the address component to be strengthened. The kernel correlation encoding vector between the principal components of the semantic features of the address components to be strengthened.

[0026] Then, the kernel association strength evaluation factor of the address component to be strengthened between any two kernel association coding vectors of the main components of the semantic features of the address component to be strengthened in the set of kernel association coding vectors of the main components of the semantic features of the address component to be strengthened is calculated to obtain the association topological matrix of the kernel association strength evaluation factor of the address component to be strengthened. Specifically, when dealing with address components with multi-level semantic nesting, for example, the main components of a component may simultaneously carry geographical hierarchical attributes, cultural metaphor characteristics and spatial orientation information, the traditional shallow association measurement based on Euclidean distance or cosine similarity is difficult to characterize the functional synergy relationship between these cross-dimensional features. As a task-oriented customized measurement function, the kernel association strength evaluation factor of the address component to be strengthened has the core purpose of converting the complex interaction pattern between high-dimensional kernel association coding vectors into quantifiable topological connection weights, so as to construct an adjacency relationship matrix that can reflect the semantic synergy effect for the graph neural network. Therefore, in the technical solution of the present application, the kernel association strength evaluation factor of the address component to be strengthened between any two kernel association coding vectors of the main components of the semantic features of the address component to be strengthened in the set of kernel association coding vectors of the main components of the semantic features of the address component to be strengthened is calculated to obtain the kernel association strength evaluation factor association topology matrix of the address component to be strengthened. That is, an interpretable semantic association strength map is constructed, and the directional reinforcement and suppression of feature interaction is achieved through a customized measurement function. Specifically, by calculating the kernel association strength evaluation factor of the address component to be strengthened, the complex interaction pattern between the kernel association coding vectors of the main components of the semantic features of the address component to be strengthened can be converted into quantifiable association weights, which not only reflect the collaborative contribution of the features to the translation task (such as the hierarchical constraint relationship between the "street type" and the "administrative division" features), but also dynamically adapt to the translation rules of different language pairs (such as Chinese to Japanese needs to strengthen the feature association of the Chinese character cultural circle, while Chinese to English needs to enhance the feature association of transliteration rules). The generated correlation topology matrix of the core correlation strength evaluation factors of the address components to be strengthened can accurately depict the influence of different semantic dimensions on translation decisions. For example, when translating streets with the same name, the correlation weight between the cultural memorial attribute characteristics of "Zhongshan Road" and the historical background characteristics of the superior administrative district will be strengthened to ensure that its commemorative meaning is retained when translated as "Zhongshan Road"; while when "Nanjing Road" is a commercial landmark, the correlation weight between its functional attribute characteristics and urban economic characteristics dominates, driving the use of the internationally accepted "Nanjing Road" rather than literal translation during translation. This dynamic quantification mechanism enables the system to adaptively handle complex scenarios, and achieves a precise balance between retaining the original cultural characteristics and conforming to the target language norms by adjusting the correlation strength of different semantic channels.In a specific example of the present application, the kernel association strength evaluation factor of the address component to be strengthened between any two kernel association coding vectors between the principal components of the semantic features of the address component to be strengthened in the set of kernel association coding vectors between the principal components of the semantic features of the address component to be strengthened is calculated using the following association strength calculation formula to obtain an association topology matrix of the kernel association strength evaluation factors of the address component to be strengthened; wherein the association strength calculation formula is:. ; ; in, and Respectively and The first one in the kernel correlation encoding vector between the principal components of the semantic features of the address components to be strengthened The eigenvalues ​​at the positions, is the number of eigenvalues ​​in the kernel association encoding vector between the principal components of the semantic features of the address component to be strengthened, for and The kernel correlation strength assessment factor of the address component to be strengthened between The topological matrix associated with the kernel association strength assessment factor of the address component to be strengthened.

[0027] Then, the kernel spatial distribution disorder optimization of the address component semantic features to be strengthened is performed on each kernel correlation coding vector between the principal components of the address component semantic features to be strengthened in the set of kernel correlation coding vectors between the principal components of the address component semantic features to be strengthened, so as to obtain the set of kernel correlation coding vectors between the principal components of the address component semantic features to be strengthened after optimization. In particular, the disorder of the spatial distribution of the kernel correlation coding vectors between the principal components of the address component semantic features to be strengthened under the action of the random potential field will lead to the distortion of the information transmission of the graph structure. For example, there may be a dimensional break between the principal component features of the address component (such as the administrative hierarchy axis and the cultural attribute axis) and the context association topology, which will block the semantic reasoning path across components. Traditional methods do not consider the compactness constraints of the feature principal components in the topological space, which leads to the attenuation or distortion of high-order correlation signals in the graph convolution process, affecting the semantic consistency of the translation decision. In a preferred example of the present application, the kernel correlation coding vectors between the main components of the semantic features of the address components to be strengthened are subjected to disorder optimization of the kernel spatial distribution of the address components to be strengthened in the set of kernel correlation coding vectors between the main components of the semantic features of the address components to be strengthened, so as to obtain the set of kernel correlation coding vectors between the main components of the semantic features of the address components to be strengthened after optimization. That is, in the technical scheme of the present application, the standardized transition of the topological space is realized by constructing a cross-section function matrix, and the disorder of the spatial distribution caused by the random potential field is eliminated. Specifically, the system projects the set of kernel correlation coding vectors between the main components of the semantic features of the discrete address components to be strengthened to the normalized topological form space through compactification vector generation and Gaussian correlation coefficient iterative optimization, so that the graph convolution network can perform multi-hop information propagation based on a stable spatial distribution. For example, in a cross-language address translation scenario, when processing address components with complex hierarchical relationships, the construction of the compactification matrix can ensure that the correlation topology between the main component axes of administrative levels such as "provincial-municipal-district" satisfies the spatial continuity, and avoids translation faults caused by dimensional dislocation.In this process, by adopting a dynamic space calibration mechanism based on the cross-section function matrix, first, the kernel association coding vector between the principal components of the semantic features of each address component to be strengthened is matrix-multiplied with the initial cross-section function matrix of the address component features to be strengthened, so as to reduce the dimensionality of the high-dimensional feature space and compactify it, and eliminate the interference signals of the redundant dimensions; then, the compacted cross-section matrix of the address component features to be strengthened is formed by two-dimensional splicing, and the Gaussian correlation coefficient between it and the topological matrix of the correlation strength evaluation factor of the kernel association strength of the address component to be strengthened is calculated. Through the iterative optimization process that constrains the Gaussian coefficient to approach zero, the system gradually corrects the parameters of the initial cross-section function matrix of the address component features to be strengthened, so that it becomes a standardized transition bridge connecting the association coding space of the principal components of the semantic features of the address components to be strengthened and the graph topological space; finally, the optimized initial cross-section function matrix of the address component features to be strengthened maps the kernel association coding vector between the principal components of the semantic features of the address components to be strengthened to the compact topological subspace, ensuring that the message passing of the graph convolutional network follows the normalized spatial distribution law. By eliminating the disorder of spatial distribution, the model can accurately capture cross-level and cross-dimensional semantic association patterns. For example, when processing multilingual mixed addresses, the implicit association between the principal component axis of cultural attributes and the language rule topology can be fully transmitted. This standardized spatial distribution mechanism enables high-order semantic signals to be losslessly aggregated through multi-layer graph convolution, ensuring that the translation results satisfy both local context constraints and global semantic consistency, fundamentally breaking through the topological distortion bottleneck of traditional methods in complex place name semantic reasoning.

[0028] That is, in the calculation process of the above graph convolutional neural network model, the correlation topology matrix of the kernel correlation strength evaluation factor of the address component to be strengthened is As a topological form space, the kernel association encoding vector between the principal components of the semantic features of each address component to be strengthened is will act as spatial masters and slaves to obey the spatial distribution form, that is, , and considering the kernel correlation coding vector between the principal components of the semantic features of the address component to be strengthened A topological matrix associated with the strength assessment factor of the address component kernel to be strengthened The dimensional break between them requires the construction of a matrix of characteristic cross-section functions from the initial address component to be strengthened. To serve as the space standard transition specification field.

[0029] On the other hand, it is also necessary to correct the characteristic cross-section function matrix of the initial address component to be enhanced Due to the disorder of spatial distribution caused by random potential field, firstly, the kernel correlation coding vector between the principal components of the semantic features of each address component to be strengthened is The corresponding initial address component feature cross-section function matrix Perform matrix multiplication to obtain the compactified vector of the characteristic cross section of the address component to be enhanced , and then associate the kernel-related encoding vectors with the principal components of the semantic features of each address component to be strengthened The corresponding feature cross-section compactification vector of the address component to be enhanced Two-dimensional splicing to obtain the compact matrix of the characteristic cross section of the address component to be enhanced : ; ; in, is the kernel association encoding vector of the principal components of the semantic features of the address component to be strengthened. kernel correlation encoding vectors between principal components of semantic features of address components to be strengthened, is the characteristic cross-section function matrix of the initial address component to be strengthened, is matrix multiplication, is the first in the sequence of compactified vectors of the characteristic cross section of the address component to be enhanced. The feature cross-section compactification vector of the address component to be enhanced, For two-dimensional splicing processing, is the compactified matrix of the characteristic section of the address component to be strengthened.

[0030] In this way, the characteristic cross-section compactification matrix of the address component to be strengthened can be calculated A topological matrix associated with the strength assessment factor of the address component kernel to be strengthened and make the Gaussian correlation coefficient tend to zero to iterate the initial address component feature cross-section function matrix to be strengthened , in order to obtain the optimized initial address component characteristic cross-section function matrix : ; in, For positional subtraction, For the matrix norm, yes and The variance of the set of all matrix values ​​of , The natural constant The value of the natural exponential function with base .

[0031] Therefore, by optimizing the characteristic cross-section function matrix of the initial address component to be strengthened To optimize the kernel correlation encoding vector between the principal components of the semantic features of the address component to be strengthened: ; in, In order to optimize the characteristic cross-section function matrix of the initial address component to be strengthened, For the The optimized kernel association coding vector between the principal components of the semantic features of the address components to be strengthened corresponds to the kernel association coding vector between the principal components of the semantic features of the address components to be strengthened.

[0032] In the calculation process of the graph convolutional neural network model, it is possible to solve the problem of the kernel association encoding vector between each feature principal component and the kernel association strength evaluation factor association topology matrix of the address component to be strengthened under the representation of the random potential field. The spatial distribution disorder problem caused by the topological form space is solved, thereby improving the calculation results of the graph convolutional neural network model.

[0033] Furthermore, the set of kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened after optimization and the association topology matrix of the kernel association strength evaluation factors of the address components to be strengthened are input into the graph convolutional neural network model to obtain the semantic strengthening coding matrix of the address components to be strengthened. Specifically, although the kernel association coding can capture the direct association between the principal components, it cannot explicitly express the cross-level and cross-dimensional composite semantic influences. For example, the administrative division attributes of the address components may indirectly affect the road naming rules through multi-level node transmission, while cultural characteristics may form implicit associations with language habits through long-range connections in the graph structure. This complex interaction pattern is difficult to capture through shallow networks and needs to rely on the deep aggregation mechanism of graph convolution. Specifically, the graph convolutional neural network regards the kernel association coding vector between the principal components of the semantic features of each address component to be strengthened as a graph node, and aggregates the semantic features of adjacent nodes layer by layer based on the connection weights defined by the association topology matrix of the kernel association strength evaluation factors of the address components to be strengthened. For example, when dealing with addresses containing multiple administrative levels, the features of "provincial nodes" can influence the translation decisions of "street-level nodes" through multi-hop propagation of graph convolution, ensuring that the translation results not only meet local context constraints but also inherit the characteristics of the global administrative system. At the same time, the attention gating mechanism in the graph structure can dynamically adjust the contribution of nodes at different levels. For example, when dealing with culturally sensitive place names, it can enhance the propagation weight of historical evolution nodes and suppress the interference of ordinary descriptive nodes. Taking the address components with multiple administrative nesting relationships as an example, low-level nodes (such as road names) receive jurisdictional constraint signals from high-level nodes (such as provincial administrative regions) through multi-hop propagation, while transmitting geographical specific features (such as regional naming conventions) in reverse, forming a two-way semantic correction. Each layer of convolution operation is essentially a joint embedding process of feature space and topological space, and the implicit association rules across components are gradually refined through nonlinear transformation. The generated semantic enhancement coding matrix of the address components to be strengthened contains multi-granularity semantic associations of address components from local to global. This global semantic reasoning mechanism based on graph structure effectively solves the problem of over-reliance on local co-occurrence patterns in traditional methods, and realizes the coordinated optimization of semantic fidelity and cross-language adaptability in place name translation. In a specific example of the present application, the set of kernel association coding vectors between the principal components of the semantic features of the optimized address components to be strengthened and the association topology matrix of the kernel association strength evaluation factors of the address components to be strengthened are input into the graph convolutional neural network model using the following graph convolution formula to obtain the semantic enhancement coding matrix of the address components to be strengthened; wherein, the graph convolution formula is: ; in, For the The kernel correlation coding vector between the principal components of the semantic features of the address components to be strengthened corresponds to the optimized kernel correlation coding vector between the principal components of the semantic features of the address components to be strengthened, represents the graph convolutional neural network model, The semantic enhancement coding matrix of the address component to be enhanced.

[0034] In particular, the S5 obtains the target language address text based on the semantic reinforcement coding features of the address components to be strengthened. That is, in the technical solution of the present application, the semantic reinforcement coding matrix of the address components to be strengthened is input into the pre-trained neural machine translation model to obtain the target language address text. Specifically, when the traditional translation model directly processes the original address text, it is limited by the discreteness of the surface symbols and the heterogeneity of the language structure, and it is difficult to synchronously coordinate the complex mapping relationship between geographical hierarchical constraints, cultural metaphor features and multilingual grammatical rules. Therefore, in the technical solution of the present application, the semantic reinforcement coding matrix of the address components to be strengthened is input into the pre-trained neural machine translation model to realize the accurate mapping of structured semantic features to the target language sequence. This deep semantic-driven generation mechanism realizes the lossless conversion from structured semantic representation to multilingual sequence generation, ensuring that the translation result meets the requirements of both geographical accuracy and language naturalness.

[0035] In summary, the real-time multilingual translation method of place names and addresses based on a neural network model according to the embodiment of the present application is explained, which first extracts the topological relationship of the address components through structured analysis, and then uses a neural network model based on deep learning to deeply model the contextual dependencies across components, so as to mine the deep semantic representation of the address text through semantic association learning across components, and based on this, realize the disambiguation of place names based on dynamic semantic reasoning. In this way, the problem of semantic deviation in complex scenarios such as the same name in different places and the same place with different names is solved, and the accuracy of distinguishing the ambiguity of geographic entities is significantly improved, so that the translation results can dynamically adapt to the expression habits of place names in different languages, effectively eliminate translation ambiguity caused by the limitations of the static knowledge base or imperfect rules, and achieve a more accurate multilingual address conversion effect.

[0036] Furthermore, a real-time multi-language translation system for place names and addresses based on a neural network model is also provided.

[0037] Figure 5 FIG. 1 is a block diagram of a real-time multilingual translation system for place names and addresses based on a neural network model according to an embodiment of the present application. Figure 5As shown, according to the embodiment of the present application, the real-time place name and address multilingual translation system 300 based on the neural network model includes: an address text acquisition module 310, which is used to acquire the address text input by the user; an address parsing module 320, which is used to perform structured parsing of the address text using the address parsing model to obtain a set of address components; a semantic encoding module 330, which is used to perform semantic encoding on the address components to be strengthened and the context information of the address components to be strengthened in the set of address components, respectively, to obtain the semantic embedding encoding features of the address components to be strengthened and the semantic encoding features of the context information of the address components to be strengthened; a semantic enhancement module 340, which is used to perform semantic enhancement on the semantic embedding encoding features of the address components to be strengthened based on the semantic encoding features of the context information of the address components to be strengthened, to obtain the semantic enhancement encoding features of the address components to be strengthened; a target language address text generation module 350, which is used to obtain the target language address text based on the semantic enhancement encoding features of the address components to be strengthened.

[0038] As described above, the real-time place name and address multi-language translation system 300 based on a neural network model according to an embodiment of the present application can be implemented in various wireless terminals, such as a server with a real-time place name and address multi-language translation algorithm based on a neural network model. In a possible implementation, the real-time place name and address multi-language translation system 300 based on a neural network model according to an embodiment of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the real-time place name and address multi-language translation system 300 based on a neural network model can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the real-time place name and address multi-language translation system 300 based on a neural network model can also be one of the many hardware modules of the wireless terminal.

[0039] Alternatively, in another example, the real-time place name and address multilingual translation system 300 based on the neural network model and the wireless terminal may also be separate devices, and the real-time place name and address multilingual translation system 300 based on the neural network model may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0040] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A real-time multi-language translation method for place names and addresses based on a neural network model, characterized in that: include: Get the address text entered by the user; Use the address parsing model to perform structured parsing on the address text to obtain a set of address components; Selecting an address component to be enhanced from a set of address components and using other address components in the set of address components as context information of the address component to be enhanced; Performing semantic embedding coding on the address component to be strengthened to obtain the semantic embedding coding features of the address component to be strengthened, and semantically enhancing the semantic embedding coding features of the address component to be strengthened based on the context information of the address component to be strengthened to obtain the semantic enhancement coding features of the address component to be strengthened; Based on the semantic enhancement encoding features of the address components to be enhanced, the target language address text is obtained.

2. The real-time multi-language translation method of place names and addresses based on a neural network model according to claim 1, characterized in that: The address component to be strengthened is semantically embedded and encoded to obtain the semantic embedded encoding features of the address component to be strengthened, and the semantic embedded encoding features of the address component to be strengthened are semantically enhanced based on the context information of the address component to be strengthened to obtain the semantic enhanced encoding features of the address component to be strengthened, including: Use the embedding matrix of the address component to be strengthened to perform semantic embedding coding on the address component to be strengthened, so as to obtain the semantic embedding coding vector of the address component to be strengthened and use it as the semantic embedding coding feature of the address component to be strengthened; Based on the context information of the address component to be strengthened, the semantic embedding coding features of the address component to be strengthened are semantically enhanced to obtain the semantic enhancement coding features of the address component to be strengthened.

3. The real-time multi-language translation method of place names and addresses based on a neural network model according to claim 2 is characterized in that: Based on the context information of the address component to be strengthened, the semantic embedding coding features of the address component to be strengthened are semantically enhanced to obtain the semantic enhancement coding features of the address component to be strengthened, including: Semantically encoding the context information of the address component to be strengthened to obtain a semantic encoding vector of the context information of the address component to be strengthened and use it as a semantic encoding feature of the context information of the address component to be strengthened; Based on the semantic coding features of the context information of the address component to be strengthened, the semantic embedding coding features of the address component to be strengthened are semantically enhanced to obtain the semantic enhancement coding features of the address component to be strengthened.

4. The method for real-time multilingual translation of place names and addresses based on a neural network model according to claim 3, characterized in that: The context information of the address component to be strengthened is semantically encoded to obtain a semantic encoding vector of the context information of the address component to be strengthened, including: Performing semantic embedding coding on each address component in the context information of the address component to be strengthened, so as to obtain a context sequence of semantic embedding coding vectors of the place name component; The context sequence of the semantic embedding coding vector of the place name component is subjected to context semantic encoding based on the bidirectional LSTM model to obtain the semantic coding vector of the context information of the address component to be strengthened.

5. The method for real-time multilingual translation of place names and addresses based on a neural network model according to claim 4, characterized in that: Based on the semantic coding feature of the context information of the address component to be strengthened, semantic enhancement is performed on the semantic embedding coding feature of the address component to be strengthened to obtain the semantic enhancement coding feature of the address component to be strengthened, including: Performing kernel association information encoding based on principal component analysis on the semantic encoding vector of the context information of the address component to be strengthened and the semantic embedding encoding vector of the address component to be strengthened, so as to obtain a set of kernel association encoding vectors between principal components of the semantic features of the address component to be strengthened; A kernel association strength evaluation factor is constructed between every two kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened in the set of kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened, and based on the kernel association strength evaluation factor, the set of kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened is subjected to graph-structured association coding to obtain the semantic strengthening coding matrix of the address components to be strengthened and used as the semantic strengthening coding feature of the address components to be strengthened.

6. The method for real-time multilingual translation of place names and addresses based on a neural network model according to claim 5, characterized in that: The kernel association information encoding based on principal component analysis is performed on the semantic encoding vector of the context information of the address component to be strengthened and the semantic embedding encoding vector of the address component to be strengthened, so as to obtain a set of kernel association encoding vectors between the principal components of the semantic features of the address component to be strengthened, including: Performing feature principal component analysis on the semantic encoding vector of the context information of the address component to be strengthened and the semantic embedding encoding vector of the address component to be strengthened, so as to obtain a set of principal component encoding vectors of the context semantic features of the address component to be strengthened and a set of principal component encoding vectors of the semantic features of the address component to be strengthened; The principal component coding vectors of the contextual semantic features of the address components to be strengthened and the principal component coding vectors of the semantic features of the address components to be strengthened are respectively subjected to principal component kernel association coding, so as to obtain a set of kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened.

7. The method for real-time multilingual translation of place names and addresses based on a neural network model according to claim 6, characterized in that: A kernel association strength evaluation factor between each two kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened is constructed, and based on the kernel association strength evaluation factor, the kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened are subjected to graph-structure-based association coding to obtain a semantic strengthening coding matrix of the address components to be strengthened, including: Calculate the address component kernel association strength assessment factor to be strengthened between any two address component kernel association coding vectors to be strengthened in the set of address component semantic feature principal component kernel association coding vectors to be strengthened, so as to obtain the address component kernel association strength assessment factor association topology matrix; Based on the association topology matrix of the kernel association strength evaluation factors of the address components to be strengthened, the set of kernel association coding vectors between the principal components of the semantic features of the address components to be strengthened is subjected to graph-structured association coding to obtain the semantic strengthening coding matrix of the address components to be strengthened.

8. The method for real-time multilingual translation of place names and addresses based on a neural network model according to claim 7, characterized in that: Based on the association topology matrix of the kernel association strength evaluation factor of the address component to be strengthened, the set of kernel association coding vectors between the principal components of the semantic features of the address component to be strengthened is subjected to graph-structured association coding to obtain the semantic strengthening coding matrix of the address component to be strengthened, including: Performing disordered optimization of the address component kernel space distribution for each of the address component semantic feature principal component kernel association coding vectors to be strengthened in the set of address component semantic feature principal component kernel association coding vectors to be strengthened, so as to obtain a set of optimized address component semantic feature principal component kernel association coding vectors to be strengthened; The set of kernel association coding vectors between the principal components of the optimized semantic features of the address components to be strengthened and the association topology matrix of the kernel association strength evaluation factors of the address components to be strengthened are input into the graph convolutional neural network model to obtain the semantic enhancement coding matrix of the address components to be strengthened.

9. The method for real-time multi-language translation of place names and addresses based on a neural network model according to claim 8, characterized in that: Based on the semantic enhancement encoding features of the address components to be enhanced, the target language address text is obtained, including: The semantic enhancement encoding matrix of the address components to be enhanced is input into the pre-trained neural machine translation model to obtain the target language address text.

10. A real-time multilingual translation system of place names and addresses based on a neural network model, characterized in that: include: An address text acquisition module is used to acquire the address text input by the user; An address parsing module, used for performing structured parsing on an address text using an address parsing model to obtain a set of address components; A semantic encoding module, used for semantically encoding the address components to be strengthened and the context information of the address components to be strengthened in the set of address components respectively to obtain semantic embedding encoding features of the address components to be strengthened and semantic encoding features of the context information of the address components to be strengthened; A semantic enhancement module, for performing semantic enhancement on the semantic embedding coding features of the address component to be enhanced based on the semantic coding features of the context information of the address component to be enhanced to obtain the semantic enhancement coding features of the address component to be enhanced; The target language address text generation module is used to obtain the target language address text based on the semantic enhancement coding features of the address components to be enhanced.

Images