Data processing system, method and electronic device

Abstract

The present application provides a data processing system, method, and electronic device. The data processing system includes: a server configured to: obtain first point of interest information and second point of interest information, wherein the first point of interest information includes: first description information and first adjacent point of interest information; and the second point of interest information includes: second description information and second adjacent point of interest information; input the first description information and the second description information into a first network model to obtain a first feature vector; input the first description information, the first adjacent point of interest information, the second description information, and the second adjacent point of interest information into a second network model to obtain a second feature vector; input the first feature vector and the second feature vector into a classification model to obtain a classification result; and a terminal configured to: send the first point of interest information and the second point of interest information to the server; receive the classification result sent by the server; and display the classification result. Embodiments of the present application can provide terminal users with accurate and complete point of interest information.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data processing system, method and electronic device. Background Art

[0002] Geospatial databases are central to many applications, such as mapping, navigation, and logistics. Building a geospatial database requires multiple providers to provide their own databases, each containing information about multiple points of interest (POIs). The geospatial database is then constructed by merging the information describing the same POI across multiple databases.

[0003] In addition, different providers may provide different POI information for the same POI, and therefore, it is impossible to provide complete and accurate POI information to the end user. Summary of the Invention

[0004] Various aspects of the present application provide a data processing system, method, and electronic device to provide complete and accurate point-of-interest information to end users.

[0005] A first aspect of an embodiment of the present application provides a data processing method, including: a server and a terminal, wherein:

[0006] The server is used to: obtain first point of interest information and second point of interest information, the first point of interest information and the second point of interest information are from different geospatial databases, the first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; input the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, the first feature vector is used to characterize the fusion of the first description information and the second description information; input the first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, the second feature vector is used to characterize the dependency relationship between the first point of interest and the second point of interest on the adjacent points of interest; input the first feature vector and the second feature vector into a classification model for classification to obtain a classification result, the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest;

[0007] The terminal is used to: obtain first point of interest information and second point of interest information; send the first point of interest information and the second point of interest information to the server so that the server determines a classification result based on the first point of interest information and the second point of interest information; receive the classification result sent by the server; and display the classification result.

[0008] A second aspect of an embodiment of the present application provides a data processing method, including: obtaining first point of interest information and second point of interest information, the first point of interest information and the second point of interest information coming from different geographic spatial databases, the first point of interest information including: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, the second point of interest information including: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; inputting the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, the first feature vector being used to characterize the fusion of the first description information and the second description information; inputting the first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, the second feature vector being used to characterize the dependency relationship between the first point of interest and the second point of interest on adjacent points of interest; inputting the first feature vector and the second feature vector into a classification model for classification to obtain a classification result, the classification result being used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest.

[0009] A third aspect of an embodiment of the present application provides a data processing method, including: obtaining first point of interest information and second point of interest information, the first point of interest information including: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information including: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; sending the first point of interest information and the second point of interest information to a server, so that the server determines a classification result based on the first point of interest information and the second point of interest information, the classification result being used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest; receiving the classification result sent by the server; and displaying the classification result.

[0010] A fourth aspect of an embodiment of the present application provides an electronic device, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the data processing method of the second aspect or the third aspect when executing the computer program.

[0011] The embodiment of the present application is applied to a geospatial entity recognition scenario, and the provided data processing system includes: a server and a terminal, the server is used to: obtain first point of interest information and second point of interest information, the first point of interest information and the second point of interest information come from different geospatial databases, the first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, the second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; input the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, the first feature vector is used to characterize the fusion of the first description information and the second description information; the first description information, The first adjacent point of interest information, the second description information, and the second adjacent point of interest information are input into the second network model for feature extraction to obtain a second feature vector, which is used to characterize the dependency relationship between the first point of interest and the second point of interest on the adjacent point of interest; the first feature vector and the second feature vector are input into the classification model for classification to obtain a classification result, which is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest; the terminal is used to: obtain the first point of interest information and the second point of interest information; send the first point of interest information and the second point of interest information to the server so that the server determines the classification result based on the first point of interest information and the second point of interest information; receive the classification result sent by the server; and display the classification result. The embodiment of the present application analyzes the description information of the first point of interest and the second point of interest and the adjacent points of interest by combining the first network model, the second network model, and the classification model, thereby accurately determining whether the first point of interest information and the second point of interest information describe the same point of interest, thereby providing the terminal user with accurate and complete point of interest information. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0013] Figure 1 A schematic structural diagram of a data processing system provided for an exemplary embodiment of the present application;

[0014] Figure 2 A flowchart of a data processing method provided for an exemplary embodiment of the present application;

[0015] Figure 3 A schematic diagram of point of interest information provided by an exemplary embodiment of the present application;

[0016] Figure 4 A flowchart of another data processing method provided by an exemplary embodiment of the present application;

[0017] Figure 5 A schematic diagram of another data processing method provided by an exemplary embodiment of the present application;

[0018] Figure 6 A flowchart of another data processing method provided by an exemplary embodiment of the present application;

[0019] Figure 7 A schematic structural diagram of an electronic device provided as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0020] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0021] In view of the problem that in the existing geospatial entity recognition scenario, the accuracy of determining whether different points of interest information describe the same point of interest is low, and thus it is impossible to provide complete and accurate points of interest information to the terminal user, the data processing system provided by the embodiment of the present application includes: a server and a terminal, the server is used to: obtain first point of interest information and second point of interest information, the first point of interest information and the second point of interest information come from different geospatial databases, the first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, the second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; the first description information and the second description information are input into the first network model for feature extraction to obtain a first feature vector, and the first feature vector is used to represent The invention relates to a method for integrating the first description information and the second description information; inputting the first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information into the second network model for feature extraction to obtain a second feature vector, the second feature vector being used to characterize the dependency relationship between the first point of interest and the second point of interest on the adjacent point of interest; inputting the first feature vector and the second feature vector into the classification model for classification to obtain a classification result, the classification result being used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest; the terminal is used to: obtain the first point of interest information and the second point of interest information; send the first point of interest information and the second point of interest information to the server so that the server determines the classification result based on the first point of interest information and the second point of interest information; receive the classification result sent by the server; and display the classification result. The embodiment of the present application analyzes the description information of the first point of interest and the second point of interest and the adjacent points of interest by combining the first network model, the second network model and the classification model, thereby being able to accurately determine whether the first point of interest information and the second point of interest information describe the same point of interest, thereby providing the terminal user with accurate and complete point of interest information.

[0022] In this embodiment, the data processing system can implement the overall data processing method using a cloud computing system. Furthermore, the server executing the data processing method can be a cloud server, leveraging the advantages of cloud resources to run various neural network models. The data processing method can also be applied to server-side devices such as conventional servers or server arrays, relative to the cloud. Furthermore, the terminal executing the data processing method can be a mobile terminal or other intelligent terminal, without limitation.

[0023] Furthermore, the data processing method provided in the embodiments of the present application can be applied to comparing point of interest information in geospatial databases. For example, a first geospatial database contains point of interest information A1, and a second geospatial database contains point of interest information A2. The data processing method provided in the embodiments of the present application is used to determine whether point of interest information A1 and point of interest information B2 both describe point of interest A. For the aforementioned various application scenarios, the embodiments of the present application can quickly, efficiently, and accurately determine whether a pair of point of interest information in different geospatial data describes the same point of interest.

[0024] Furthermore, the data processing method of the embodiment of the present application provides support for building a complete and high-quality target geospatial database to improve the service quality of industries such as navigation, social networking, advertising, and logistics. For example, there are a first geospatial database and a second geospatial database, the first geospatial database includes point of interest information for each point of interest, and the second geospatial database also includes point of interest information for each point of interest. The data processing method of the embodiment of the present application can determine the point of interest information for the same point of interest A in the first geospatial database and the second geospatial database, and then can de-duplicate, merge, and other integration processes of the point of interest information describing the same point of interest A from different geospatial databases to serve as the point of interest information about the point of interest A in the target geospatial database.

[0025] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0026] Reference Figure 1 Schematic diagram of a data processing system provided by an exemplary embodiment of the present application. Figure 1 As shown, the terminal 11 obtains the first point of interest information and the second point of interest information, wherein the first point of interest information includes: first description information and first adjacent point of interest information. The second point of interest information includes: second description information and second adjacent point of interest information. The terminal 11 sends the first point of interest information and the second point of interest information to the server 12, and the server 12 inputs the first description information and the second description information into the first network model and outputs the first feature vector. The first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information are input into the second network model to obtain the second feature vector. The first feature vector and the second feature vector are input into the classification model to obtain the classification result, and then the server 12 sends the classification result to the terminal 11 for terminal display.

[0027] Figure 2 A flowchart of a data processing method provided by an exemplary embodiment of the present application. Figure 2 The data processing method shown specifically includes the following steps:

[0028] S201: Acquire first point of interest information and second point of interest information.

[0029] The first point of interest information and the second point of interest information are from different geospatial databases. The first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest. The second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest. In this embodiment of the present application, the first point of interest information and the second point of interest information obtained by the server may be sent by the terminal or obtained by the server from a database.

[0030] Reference Figure 3 , the first point of interest information is the information for the first point of interest in the first geospatial database. The first descriptive information is all descriptive information for the first point of interest. For example, the first descriptive information includes: the name, address (such as country, province, city, street, number, etc.), zip code and spatial geographic information (longitude, latitude) of the first point of interest. The first adjacent point of interest information includes: the descriptive information of the adjacent points of interest of the first point of interest. In addition, the first adjacent point of interest information may also include: the positional relationship between the adjacent point of interest and the first point of interest, for example, the adjacent point of interest is 50 meters due north of the first point of interest. Such as at least one of the name, address, zip code and spatial geographic information of the adjacent point of interest of the first point of interest. Among them, there can be multiple adjacent points of interest of the first point of interest, and then there are multiple corresponding first adjacent point of interest information. In addition, the description of the second point of interest information refers to the first point of interest and will not be repeated here.

[0031] S202: Input the first description information and the second description information into a first network model to perform feature extraction to obtain a first feature vector.

[0032] The first eigenvector is used to represent the fusion of the first description information and the second description information.

[0033] In an embodiment of the present application, the first network model can be a language model based on the Transformer (a model result) architecture, such as BERT (a pre-trained language representation model) or GPT-2 (a language model). Here, the first network model only uses the Transformer encoder to encode the first description information and the second description information to achieve the fusion of the first description information and the second description information, and obtain the first feature vector. In addition, the model parameters of the first network model are pre-trained.

[0034] S203: Input the first description information, the first adjacent interest point information, the second description information, and the second adjacent interest point information into a second network model for feature extraction to obtain a second feature vector.

[0035] The second eigenvector is used to characterize the dependency relationship between the first point of interest and the second point of interest on adjacent points of interest.

[0036] Specifically, the model parameters of the second network model are pre-trained. The second network model can determine the dependency relationship between the first point of interest and the second point of interest on the adjacent points of interest based on the input first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information.

[0037] In the embodiment of the present application, the introduction of the first adjacent point of interest information and the second adjacent point of interest information can further improve the accuracy of the comparison of point of interest information in different geographic spatial databases.

[0038] S204: Input the first eigenvector and the second eigenvector into a classification model for classification to obtain a classification result.

[0039] The classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest.

[0040] Specifically, if the classification result is 1, it indicates that the first point of interest information and the second point of interest information describe the same point of interest. If the classification result is 0, it indicates that the first point of interest information and the second point of interest information do not describe the same point of interest.

[0041] In addition, the classification model may be a binary linear classifier model, and the model parameters of the classification model are pre-trained.

[0042] During the training stage of an embodiment of the present application, the first network model, the second network model and the classification model can be regarded as an overall model. The input of the overall model is the sample first interest point information and the sample second interest point information, and the output is the predicted classification result. The loss value between the predicted classification result and the label classification result is determined, and the model parameters of the first network model, the second network model and the classification model are adjusted according to the loss value until the loss value is less than the threshold, thereby obtaining the trained first network model, the second network model and the classification model.

[0043] An embodiment of the present application is applied to a geospatial entity recognition scenario, and the provided data processing method includes: obtaining first point of interest information and second point of interest information, the first point of interest information and the second point of interest information coming from different geospatial databases, the first point of interest information including: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, the second point of interest information including: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; inputting the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, the first feature vector being used to characterize the fusion of the first description information and the second description information; inputting the first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, the second feature vector being used to characterize the dependency relationship between the first point of interest and the second point of interest on adjacent points of interest; inputting the first feature vector and the second feature vector into a classification model for classification to obtain a classification result, the classification result being used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest. The embodiment of the present application combines the first network model, the second network model and the classification model to analyze the description information of the first point of interest and the second point of interest and the adjacent points of interest, thereby accurately determining whether the first point of interest information and the second point of interest information describe the same point of interest.

[0044] In the embodiment of the present application, another data processing method is provided, such as Figure 4 As shown, the data processing method specifically includes the following steps:

[0045] S401: Determine any point of interest information in a first geographic spatial database as first point of interest information.

[0046] The first geospatial database contains multiple points of interest information. For example, referring to Table 1, three points of interest information (P1, P2, and P3) in the first geospatial database are shown. Each point of interest information may include one or more adjacent points of interest information. For example, point of interest information P1 includes adjacent point of interest information X1, adjacent point of interest information X2, and adjacent point of interest information X3.

[0047] In the embodiment of the present application, any point of interest information may be determined as the first point of interest information, and the following processing may be performed: For example, in this step, the point of interest information P1 is determined as the first point of interest information.

[0048] Table 1

[0049]

[0050] S402: Determine second point of interest information in a second geographic spatial database based on the first point of interest information.

[0051] The similarity between the first point of interest information and the second point of interest information is greater than a preset threshold.

[0052] Referring to Table 2, three points of interest information (O1, O2, and O3) in the second geospatial database are exemplified. Determining the second point of interest information in the second geospatial database is determined based on the first point of interest information.

[0053] Table 2

[0054]

[0055] In one alternative, a distance is determined based on the latitude and longitude information of the first point of interest and the latitude and longitude information of each point of interest in the second geospatial database. If the distance is less than a distance threshold, then the similarity between the first point of interest and the point of interest is determined to be less than a similarity threshold. For example, referring to Tables 1 and 2, if the distances between the point of interest information O1 and the point of interest information O2 in Table 2 and the first point of interest information P1 are both less than the distance threshold, then the point of interest information O1 and the point of interest information O2 can be used as the second point of interest information and then compared with the first point of interest information P1 in turn.

[0056] Another optional method is to determine the second POI based on the similarity between the latitude and longitude information of the first POI and the names of the latitude and longitude information of each POI in the second geospatial database. For example, if the name of the first POI P1 is "Tavern", and the name of the POI O1 in the second geospatial database is similar to the name of the first POI P1 by more than a similarity threshold, POI O1 can be used as the second POI.

[0057] Another optional method is to combine the above distance and name to determine in the second geographic database the second point of interest information whose similarity with the first point of interest information is greater than the similarity threshold. The distance is less than or equal to the distance threshold, and the similarity corresponding to the point of interest information whose edit distance similarity of the name is less than or equal to the edit similarity threshold is greater than the similarity threshold. For example, the distance threshold is determined to be 2 kilometers, and the edit similarity threshold is 0.6. Among them, the edit distance similarity S1 = (1-edit distance / total edit distance). Among them, the edit distance is the edit distance from the second point of interest information to the first point of interest information. The total edit distance is the sum of the number of characters in the second point of interest information and the number of characters in the first point of interest. Rewrite to obtain the edit distance of the first point of interest information.

[0058] For example, in Table 1, the name corresponding to the first POI information P1 is "Tavern," and in Table 2, the name corresponding to the POI information O1 is "Good Tavern." The edit path from "Good Tavern" to "Tavern" is "Good Tavern" → "Good Tavern" → "Tavern," with an edit distance of 2 and a total edit distance of 6. The above calculation yields an edit distance similarity S1 of 2 / 3. Furthermore, if the distance between the first POI information P1 and the POI information O1 is less than 2 kilometers, it can be determined that the similarity between the first POI information P1 and the POI information O1 is greater than the similarity threshold, and the POI information O1 is then determined to be the second POI information.

[0059] In the embodiment of the present application, the method is not limited to the above method, and the second point of interest information can also be determined according to other methods.

[0060] In addition, in an embodiment of the present application, the second point of interest information is first determined based on the similarity, and there is no need to compare each point of interest information in the first geospatial database with each point of interest information in the second geospatial database, thereby improving the comparison efficiency of the point of interest information between the geospatial databases.

[0061] S403: The first attribute information is segmented according to the attribute type through the text processing layer of the semantic analysis sub-model to obtain a first segmented text.

[0062] The first description information includes: first attribute information, the second description information includes: second attribute information, and the first network model includes: semantic analysis sub-model. The first attribute information includes: the name, address, and zip code of the first point of interest. The second attribute information includes: the name, address, and zip code of the second point of interest. Figure 5 The speech analysis sub-model includes a text processing layer and an encoder, where the encoder can be one or more Transformer encoders.

[0063] Furthermore, the attribute type can be a name, address, or zip code, etc. The word segmentation processing method is: word segmentation text = [COL] attribute type 1 [VAL] attribute value 1 [COL] attribute type 2 [VAL] attribute value 2 ... [COL] attribute type k [VAL] attribute value k. Among them, [COL] is the tag of the attribute type, and [VAL] is the tag of the attribute value. For example, according to the word segmentation processing formula, the first attribute information of the first point of interest information P1 in Table 1 is word segmented, and the first word segmentation text obtained is: [COL] name [VAL] pub [COL] address [VAL] Heping Street.

[0064] S404: The second attribute information is segmented according to the attribute type through the text processing layer to obtain a second segmented text.

[0065] The second attribute information is segmented according to the above segmentation method. For example, the second attribute information of the second point of interest information O1 in Table 2 is segmented to obtain the following second segmented text: [COL] Name [VAL] Beautiful Tavern [COL] Address [VAL] No. 418 Heping [COL] Zip Code [VAL] 15212.

[0066] In the embodiment of the present application, word segmentation processing can make the first attribute information and the second attribute information have the same format, which can better combine the first attribute information and the second attribute information.

[0067] S405: Combine the first segmented text and the second segmented text through a text processing layer to obtain a combined text.

[0068] The combined processing method is: combined text = [CLS] first segmented text [SEP] second segmented text [SEP]. [CLS] and [SEP] are both tokens. [CLS] is the token that serializes the first segmented text and the second segmented text. [SEP] is the token that separates the first segmented text and the second segmented text.

[0069] In the embodiment of the present application, the first segmented text and the second segmented text are combined to achieve a combination of the first segmented text and the second segmented text. Furthermore, a cross-attribute comparison of the first attribute information and the second attribute information can be achieved, eliminating the need for manual one-to-one alignment based on attribute types, thereby improving the efficiency of the comparison of the first attribute information and the second attribute information.

[0070] S406 , encoding the combined text through the encoder of the semantic analysis sub-model to obtain a first feature sub-vector.

[0071] Multiple Transformer encoders can be configured, each employing an attention mechanism. The output of the previous encoder becomes the input to the next encoder, and the final encoder outputs a first feature subvector. Furthermore, the encoding parameters of the encoders are pre-trained. The encoders automatically align and compare values ​​of the same attribute type in the input text, and then output a first feature subvector representing the comparison between the first and second attribute information.

[0072] S407: Obtain a first eigenvector according to the first eigenvector.

[0073] In an optional manner, the first eigenvector may be used as the first eigenvector.

[0074] Another option is to refer to Figure 5S407 includes: inputting the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain a second feature sub-vector, where the second feature sub-vector is used to represent the distance between the first spatial geographic information and the second spatial geographic information; and splicing the first feature sub-vector and the second feature sub-vector to obtain a first feature vector.

[0075] The first description information further includes: first spatial geographic information; the second description information further includes: second spatial geographic information; and the first network model further includes: a distance determination sub-model.

[0076] Specifically, the first spatial geographic information and the second spatial geographic information are input into the distance determination sub-model for feature extraction to obtain a second characteristic sub-vector, including: using the distance determination sub-model to determine the distance between the first point of interest and the second point of interest based on the first spatial geographic information and the second spatial geographic information; using the distance determination sub-model to vectorize the distance to obtain the second characteristic sub-vector.

[0077] The Haversine (semi-formal) formula is used to calculate the distance between the first point of interest and the second point of interest. in, is the longitude of the first point of interest, is the longitude of the second point of interest, λ1 is the latitude of the first point of interest, λ2 is the latitude of the second point of interest, and r is the radius of the earth.

[0078] For example, the first spatial geographic information of the first point of interest P1 in Table 1 is: longitude -80.009, latitude 40.4535. The second spatial geographic information of the second point of interest O1 in Table 2 is: longitude -79.995, latitude 40.4535. The distance D between the first point of interest and the second point of interest can be determined using the Haversine formula.

[0079] In addition, vectorization processing is processed in embd (embedded) mode, as follows:

[0080]

[0081] In formula (1), embd is the second feature vector, α1 and β are the model parameters of the distance determination sub-model, which are pre-trained parameters. T refers to the transposition of α1. D is the distance between the first interest point and the second interest point obtained by the above calculation. max is a pre-set empirical distance, usually the maximum distance of the area where the first point of interest is located. For example, the comparison task is to compare the points of interest of city A in the first geospatial database with the points of interest of city A in the second geospatial database, then D maxIt can be the distance corresponding to the maximum span of city A.

[0082] In the embodiment of the present application, the distance determination sub-model can be used to combine the distance between the first point of interest and the second point of interest, thereby further improving the accuracy of the comparison between the first point of interest information and the second point of interest information.

[0083] S408 : Using a second network model to perform feature extraction on the first description information and the first adjacent interest point information to obtain a third feature sub-vector.

[0084] The third eigenvector is used to characterize the dependency relationship between the first adjacent point of interest information and the first point of interest.

[0085] Specifically, S408 includes: using the second network model to semantically represent the first description information to obtain a first representation vector; using the second network model to semantically represent each first adjacent point of interest information to obtain a second representation vector corresponding to each first adjacent point of interest information; using the second network model to determine the fusion feature of the second representation vector and the first representation vector to obtain a third feature sub-vector.

[0086] In an optional embodiment, the second network model extracts features from the first segmented text of the first attribute information and the segmented text of the first adjacent POI information (attribute information of the first adjacent POI) to obtain a third feature subvector. BERT may be used to semantically represent the first attribute information and the first adjacent POI, respectively, and the encoding parameters of the BERT may be pre-trained.

[0087] The first attribute information is semantically represented, and the first representation vector is h1. The first adjacent interest point information is semantically represented, and the second representation vector h corresponding to each of the first adjacent interest point information is obtained. 11 , the second characterization vector h 12 Until the second characterization vector h 1n . n represents the number of first adjacent interest point information and the number of second representation vectors.

[0088] Furthermore, the fusion feature is determined using the following formula:

[0089] Z=W*h (2)

[0090]

[0091]

[0092] n i =ReLU(∑ j∈N(i) α ij *Z j ) (5)

[0093] In the above formula (2), h is a linear transformation of h, where h refers to the first characterization vector or the second characterization vector, and W is a weight matrix that is pre-trained. In the embodiment of the present application, formula (2) is used to calculate the first characterization vector h1 to obtain Z1. Formula (2) is used to calculate each first adjacent interest point information h 11 、h 12 to h 1n Calculate them separately and get Z 11 , Z 12 to Z 1n .

[0094] The above formula (3) is a normalization process, where e ij is the normalized value associated with the first characterization vector. i Corresponding to the first point of interest information or the second point of interest information, for example, for the first point of interest information, i is 1. j Corresponding to the first adjacent interest point information, they can be Z 11 , Z 12 to Z 1n , then use formula (3) to calculate Z1 and Z 11 Calculate and get e 111 , for Z1 and Z 12 Calculate and get e 112 , for Z1 and Z 1n Calculate and get e 11n .

[0095] The above formula (4) is a logistic regression process, where α ij is the regression processing value. Wherein, N(i) refers to the adjacent interest point information group corresponding to the i-th interest point information. For example, when i is 1, N(1) refers to the adjacent interest point information group corresponding to the first interest point information. When i is 2, N(2) refers to the adjacent interest point information group corresponding to the second interest point information. ∑ k∈N(i) exp(e ik ) refers to the exp(e 111 )、exp(e 112 ) to exp(e 11n ). Formula 4 gives α 111 , α 112 to α 11n .

[0096] The above formula (5) is the attention weighted processing, and ReLU is the linear rectification function. The final n i is the fusion feature. n1 is the third feature sub-vector corresponding to the first interest point information. n2 is the fourth feature sub-vector corresponding to the second interest point information.

[0097] S409: Use the second network model to perform feature extraction on the second description information and the second adjacent interest point information to obtain a fourth feature sub-vector.

[0098] The fourth eigenvector is used to characterize the dependency between the second adjacent POI information and the second POI. Furthermore, the fourth eigenvector is also calculated using formulas (2) to (5), with specific reference to the method for determining the third eigenvector described above, which will not be repeated here.

[0099] S410 , performing feature concatenation on the third eigenvector and the fourth eigenvector to obtain a second eigenvector.

[0100] In the examples of this application, refer to Figure 5 , the semantic analysis sub-model, the second network model, and the classification model can be combined into an overall model framework to achieve comparison between the first point of interest information and the second point of interest information. Alternatively, the distance determination sub-model, the second network model, and the classification model can be combined into an overall model framework to achieve comparison between the first point of interest information and the second point of interest information. Alternatively, the semantic analysis sub-model, the distance determination sub-model, the second network model, and the classification model can be combined into an overall model framework to achieve comparison between the first point of interest information and the second point of interest information. Figure 5 Any one model and the combination of multiple models are within the scope of protection of this application.

[0101] S411: Input the first eigenvector and the second eigenvector into a classification model for classification to obtain a classification result.

[0102] The specific implementation process of this step is referred to S204 and will not be repeated here.

[0103] During model training, training samples and labeled data are first obtained. The training samples include information about multiple first-sample POIs from a first sample database and information about multiple second-sample POIs from a second sample database. The first-sample POIs include at least one of first-sample attribute information, first-sample geospatial information, and first-sample adjacent POIs. The first-sample attribute information includes at least one of name, address, and zip code. The second-sample POIs include at least one of second-sample attribute information, second-sample geospatial information, and second-sample adjacent POIs. The second-sample attribute information includes at least one of name, address, and zip code. Label data can be a markup by a crowdsourced user to determine whether the first-sample POI and the second-sample POI describe the same POI. If so, it is marked as 1, and if not, it is marked as 0. The training samples are then input into the constructed model framework, and a predicted classification result is output. The loss between the predicted classification result and the labeled data is calculated, and the parameters of each model are then adjusted.

[0104] Refer to Table 3, which shows the number of sample POI information in different cities for different sample databases.

[0105] Table 3

[0106] City The first sample database The second sample database The third sample database A 23985 31936 13699 B 11389 7546 3868 C 38286 18851 17204 D 9387 11579 6356

[0107] Refer to Table 4, which shows the labeling results of the POI information of the two sample databases.

[0108] Table 4

[0109]

[0110] During the training phase, part of the POI information can be input, such as inputting information other than the address, which can improve the robustness of the model.

[0111] Furthermore, the data in Table 2 above can be used to test the comparison accuracy of the data processing method of the present application and compare it with existing solutions. The experimental results in Table 5 show that the data processing method provided in the present embodiment has a higher comparison accuracy for point of interest information in different geospatial databases than the most accurate comparison method currently available, significantly improving accuracy.

[0112] Table 5

[0113]

[0114] Among them, DeepMatcher, SKyEx and Ditto in Table 5 are all models (methods) used in the prior art to compare POI information in different databases.

[0115] The embodiment of the present application is applied to the geospatial entity recognition scenario, and the provided data processing method includes analyzing the description information of the first point of interest and the second point of interest and the adjacent points of interest by combining the first network model, the second network model and the classification model, so as to accurately determine whether the first point of interest information and the second point of interest information describe the same point of interest.

[0116] Reference Figure 6 , is a flowchart of another data processing method provided in an embodiment of the present application, which is applied to a terminal. The data processing method includes:

[0117] S601: Acquire first point of interest information and second point of interest information.

[0118] The first point of interest information includes: first description information of the first point of interest and information about first neighboring points of interest of the first point of interest; the second point of interest information includes: second description information of the second point of interest and information about second neighboring points of interest of the second point of interest. Furthermore, the first point of interest information and the second point of interest information are derived from different geospatial databases. For details, please refer to the above embodiment and will not be elaborated here.

[0119] S602: Send first point of interest information and second point of interest information to a server.

[0120] Among them, after receiving the first point of interest information and the second point of interest information, the server determines a classification result based on the first point of interest information and the second point of interest information. The classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest. Specifically, the server can execute steps S201 to S204 to obtain the classification result. The details can be referred to the above embodiment and will not be repeated here.

[0121] S603: Receive the classification result sent by the server.

[0122] S604: Display the classification results.

[0123] Among them, displaying the classification results includes: if the first point of interest information and the second point of interest information describe the same point of interest, integrating the first point of interest information and the second point of interest information to obtain third point of interest information, the third point of interest information including: third attribute information, third spatial geographic information and third adjacent point of interest information; displaying the third attribute information, third spatial geographic information and third adjacent point of interest information in the form of a map; if the first point of interest information and the second point of interest information describe different points of interest, displaying the first point of interest information and the second point of interest information respectively in the form of a map.

[0124] Integrating the first and second POI information refers to deduplicating and combining the first and second POI information. For example, if the first attribute information is: Name: Pub; Address: Heping Street. The second attribute information is: Name: Beautiful Pub; Address: 418 Heping Street; Zip Code: 15212. The third attribute information obtained after integration is: Name: Beautiful Pub; Address: 418 Heping Street; Zip Code: 15212. The third spatial geographic information includes: longitude and latitude. The longitude can be the average of the longitude of the first spatial geographic information and the longitude of the second spatial geographic information; the latitude can be the average of the latitude of the first spatial geographic information and the latitude of the second spatial geographic information. The third adjacent POI information refers to the result of deduplicating and combining the first and second adjacent POI information. If the first adjacent POI information is "As Dessert Shop, Star Cafe, Beautiful Cinema," and the second adjacent POI information is "Star Cafe, Lovely Rice Noodle Shop," the third adjacent POI information is "As Dessert Shop, Star Cafe, Beautiful Cinema, Lovely Rice Noodle Shop."

[0125] Furthermore, the third attribute information, the third spatial geographic information and the third adjacent point of interest information are displayed in the form of a map, including the third adjacent point of interest information in combination with the name, address, zip code of the third attribute information, the longitude and latitude in the third spatial address information, and the azimuth distance between the third adjacent point of interest information and the point of interest corresponding to the third attribute information, etc.

[0126] In an optional embodiment, if the first point of interest information and the second point of interest information describe different points of interest, the first point of interest information and the second point of interest information are displayed separately.

[0127] In an embodiment of the present application, first point of interest information and second point of interest information can be obtained from different geographic spatial databases, and the first point of interest information and the second point of interest information can be sent to a server. Based on the classification results given by the server, more complete and accurate point of interest information can be provided to the terminal user.

[0128] In an embodiment of the present application, in addition to providing a data processing method, a data processing device (not shown) is also provided. The data processing device includes:

[0129] an acquisition module, configured to acquire first point of interest information and second point of interest information, the first point of interest information and the second point of interest information being from different geospatial databases, the first point of interest information including: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information including: second description information of the second point of interest and second adjacent point of interest information of the second point of interest;

[0130] A first feature extraction module is used to input the first description information and the second description information into the first network model for feature extraction to obtain a first feature vector, where the first feature vector is used to represent the fusion of the first description information and the second description information;

[0131] a second feature extraction module, configured to input the first description information, the first adjacent point of interest information, the second description information, and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, wherein the second feature vector is used to characterize a dependency relationship between the first point of interest and the second point of interest on the adjacent point of interest;

[0132] The classification module is used to input the first feature vector and the second feature vector into the classification model for classification to obtain a classification result, which is used to indicate whether the first point of interest information and the second point of interest describe the same point of interest.

[0133] In an optional embodiment, the first description information includes: first attribute information, the second description information includes: second attribute information, and the first network model includes: a semantic analysis sub-model; the first feature extraction module is specifically used to: perform word segmentation processing on the first attribute information according to the attribute type through the text processing layer of the semantic analysis sub-model to obtain a first word segmentation text; perform word segmentation processing on the second attribute information according to the attribute type through the text processing layer to obtain a second word segmentation text; combine the first word segmentation text and the second word segmentation text through the text processing layer to obtain a combined text; encode the combined text through the encoder of the semantic analysis sub-model to obtain a first feature sub-vector; and obtain a first feature vector based on the first feature sub-vector.

[0134] In an optional embodiment, when the first feature extraction module is used to obtain the first feature vector based on the first feature sub-vector, it is specifically the same as: inputting the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain the second feature sub-vector, and the second feature sub-vector is used to represent the distance between the first spatial geographic information and the second spatial geographic information; splicing the first feature sub-vector and the second feature sub-vector to obtain the first feature vector.

[0135] In an optional embodiment, when the first feature extraction module is used to input the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain the second feature sub-vector, it is specifically the same as: based on the first spatial geographic information and the second spatial geographic information, using the distance determination sub-model to determine the distance between the first point of interest and the second point of interest; using the distance determination sub-model to vectorize the distance to obtain the second feature sub-vector.

[0136] In an optional embodiment, the second feature extraction module is specifically used to: use the second network model to perform feature extraction on the first description information and the first adjacent interest point information to obtain a third feature sub-vector, and the third feature sub-vector is used to characterize the dependency relationship between the first adjacent interest point information and the first interest point; use the second network model to perform feature extraction on the second description information and the second adjacent interest point information to obtain a fourth feature sub-vector, and the fourth feature sub-vector is used to characterize the dependency relationship between the second adjacent interest point information and the second interest point; perform feature splicing on the third feature sub-vector and the fourth feature sub-vector to obtain a second feature vector.

[0137] In an optional embodiment, when the second feature extraction module is used to perform feature extraction on the first description information and the first adjacent interest point information using the second network model to obtain the third feature sub-vector, it is specifically used to: use the second network model to semantically represent the first description information to obtain a first representation vector; use the second network model to semantically represent each first adjacent interest point information separately to obtain a second representation vector corresponding to each first adjacent interest point information; use the second network model to determine the fusion feature of the second representation vector and the first representation vector to obtain a third feature sub-vector.

[0138] In an optional embodiment, the acquisition module is specifically used to: determine any point of interest information in the first geospatial database as first point of interest information; determine second point of interest information in the second geospatial database based on the first point of interest information, and the similarity between the first point of interest information and the second point of interest information is greater than a preset threshold.

[0139] The embodiment of the present application is applied to the geospatial entity recognition scenario, and the provided data processing device includes analyzing the description information of the first point of interest and the second point of interest and the adjacent points of interest by combining the first network model, the second network model and the classification model, so as to accurately determine whether the first point of interest information and the second point of interest information describe the same point of interest.

[0140] In an optional embodiment, another data processing device (not shown) is further provided, the data processing device comprising:

[0141] The acquisition module is used to acquire the first point of interest information and the second point of interest information.

[0142] The first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest; the second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest.

[0143] The sending module is used to send the first point of interest information and the second point of interest information to the server.

[0144] After receiving the first point of interest information and the second point of interest information, the server determines a classification result based on the first point of interest information and the second point of interest information, and the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest.

[0145] The receiving module is used to receive the classification results sent by the server.

[0146] Display module, used to display classification results.

[0147] Among them, the display module is specifically used to: if the first point of interest information and the second point of interest information describe the same point of interest, then integrate the first point of interest information and the second point of interest information to obtain the third point of interest information, the third point of interest information includes: third attribute information, third spatial geographic information and third adjacent point of interest information; display the third attribute information, third spatial geographic information and third adjacent point of interest information in the form of a map; if the first point of interest information and the second point of interest information describe different points of interest, then display the first point of interest information and the second point of interest information respectively in the form of a map.

[0148] The data processing device provided in the embodiment of the present application can obtain the first point of interest information and the second point of interest information from different geographic spatial databases, send the first point of interest information and the second point of interest information to the server, and provide the terminal user with more complete and accurate point of interest information based on the classification results given by the server.

[0149] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or in parallel. They are only used to distinguish between different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence, nor do they limit "first" and "second" to be different types.

[0150] Figure 7 This is a schematic diagram of the structure of an electronic device provided by an exemplary embodiment of the present application. The electronic device is used to run the upper body voice response method. Figure 7 As shown, the electronic device includes a memory 74 and a processor 75 .

[0151] The memory 74 is used to store computer programs and can be configured to store various other data to support operations on the electronic device. The memory 74 can be an object storage service (OSS).

[0152] The memory 74 can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0153] The processor 75 is coupled to the memory 74 and is used to execute the computer program in the memory 74 to: obtain first point of interest information and second point of interest information, the first point of interest information and the second point of interest information are from different geographic spatial databases, the first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; input the first description information and the second description information into the first network model for feature extraction to obtain a first feature vector, the first feature vector is used to characterize the fusion of the first description information and the second description information; input the first description information, the first adjacent point of interest information, the second description information and the second adjacent point of interest information into the second network model for feature extraction to obtain a second feature vector, the second feature vector is used to characterize the dependency relationship between the first point of interest and the second point of interest on the adjacent points of interest; input the first feature vector and the second feature vector into the classification model for classification to obtain a classification result, the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest.

[0154] Further optionally, when the processor 75 inputs the first description information and the second description information into the first network model for feature extraction to obtain the first feature vector, it is specifically used to: perform word segmentation processing on the first attribute information according to the attribute type through the text processing layer of the semantic analysis sub-model to obtain the first word segmentation text; perform word segmentation processing on the second attribute information according to the attribute type through the text processing layer to obtain the second word segmentation text; combine the first word segmentation text and the second word segmentation text through the text processing layer to obtain the combined text; encode the combined text through the encoder of the semantic analysis sub-model to obtain the first feature sub-vector; and obtain the first feature vector based on the first feature sub-vector.

[0155] Further optionally, when the processor 75 obtains the first feature vector based on the first feature sub-vector, it is specifically used to: input the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain the second feature sub-vector, and the second feature sub-vector is used to represent the distance between the first spatial geographic information and the second spatial geographic information; and splice the first feature sub-vector and the second feature sub-vector to obtain the first feature vector.

[0156] Further optionally, when the processor 75 inputs the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain the second characteristic sub-vector, it is specifically used to: determine the distance between the first point of interest and the second point of interest based on the first spatial geographic information and the second spatial geographic information using the distance determination sub-model; and vectorize the distance using the distance determination sub-model to obtain the second characteristic sub-vector.

[0157] In an optional embodiment, when the processor 75 inputs the first eigenvector and the second eigenvector into the classification model for classification and obtains the classification result, it is specifically used to use the second network model to perform feature extraction on the first description information and the first adjacent interest point information to obtain a third eigenvector, and the third eigenvector is used to characterize the dependency relationship between the first adjacent interest point information and the first interest point; use the second network model to perform feature extraction on the second description information and the second adjacent interest point information to obtain a fourth eigenvector, and the fourth eigenvector is used to characterize the dependency relationship between the second adjacent interest point information and the second interest point; and perform feature splicing on the third eigenvector and the fourth eigenvector to obtain a second eigenvector.

[0158] In an optional embodiment, the processor 75 uses the second network model to perform feature extraction on the first description information and the first adjacent interest point information to obtain the third feature sub-vector, which is specifically used to use the second network model to semantically represent the first description information to obtain the first representation vector; use the second network model to semantically represent each first adjacent interest point information to obtain the second representation vector corresponding to each first adjacent interest point information; use the second network model to determine the fusion feature of the second representation vector and the first representation vector to obtain the third feature sub-vector.

[0159] In an optional embodiment, when the processor 75 is used to obtain the first point of interest information and the second point of interest information, it is specifically used to: determine any point of interest information in the first geospatial database as the first point of interest information; determine the second point of interest information in the second geospatial database based on the first point of interest information, and the similarity between the first point of interest information and the second point of interest information is greater than a preset threshold.

[0160] The processor 75 is coupled to the memory 74 and is used to execute the computer program in the memory 74 to: obtain first point of interest information and second point of interest information; wherein the first point of interest information includes: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information includes: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; send the first point of interest information and the second point of interest information to the server; wherein, after receiving the first point of interest information and the second point of interest information, the server determines a classification result based on the first point of interest information and the second point of interest information, and the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest; receive the classification result sent by the server; and display the classification result.

[0161] In an optional embodiment, the processor 75 is specifically used to display the classification results: if the first point of interest information and the second point of interest information describe the same point of interest, then integrate the first point of interest information and the second point of interest information to obtain third point of interest information, the third point of interest information including: third attribute information, third spatial geographic information and third adjacent point of interest information; display the third attribute information, third spatial geographic information and third adjacent point of interest information in the form of a map; if the first point of interest information and the second point of interest information describe different points of interest, then display the first point of interest information and the second point of interest information respectively in the form of a map.

[0162] Furthermore, if Figure 7 The electronic device also includes: a firewall 71, a load balancer 72, a communication component 76, a power supply component 78 and other components. Figure 7 Only some components are shown schematically, which does not mean that the electronic device only includes Figure 7 Components shown.

[0163] The electronic device provided by the embodiment of the present application, in the geospatial entity recognition scenario, provides a data processing device, which includes analyzing the description information of the first point of interest and the second point of interest and the adjacent points of interest by combining the first network model, the second network model and the classification model, and then accurately determining whether the first point of interest information and the second point of interest information describe the same point of interest, and can obtain the first point of interest information and the second point of interest information from different geospatial databases, send the first point of interest information and the second point of interest information to the server, and provide the terminal user with more complete and accurate point of interest information based on the classification results given by the server.

[0164] Accordingly, the embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to implement Figure 2 、 Figure 4 or Figure 6 Steps in the method shown.

[0165] Accordingly, the embodiment of the present application further provides a computer program product, including a computer program / instruction, which, when executed by a processor, causes the processor to implement Figure 2 、 Figure 4 or Figure 6 Steps in the method shown.

[0166] above Figure 7 The communication component is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related text from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0167] above Figure 7 A power supply component provides power to various components of the device in which the power supply component is located. The power supply component may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device in which the power supply component is located.

[0168] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0169] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0170] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0171] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0172] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0173] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0174] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store text. The text can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store text that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0175] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0176] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A data processing system, characterized in that: include: Servers and terminals, including: The server is configured to obtain first point of interest information and second point of interest information, the first point of interest information and the second point of interest information being from different geospatial databases, the first point of interest information including first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information including second description information of the second point of interest and second adjacent point of interest information of the second point of interest; Inputting the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, where the first feature vector is used to represent the fusion of the first description information and the second description information; Inputting the first description information, the first adjacent point of interest information, the second description information, and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, where the second feature vector is used to characterize the dependency relationship between the first point of interest and the second point of interest on adjacent points of interest; Inputting the first feature vector and the second feature vector into a classification model for classification to obtain a classification result, wherein the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest; The terminal is used to: obtain the first point of interest information and the second point of interest information; sending the first point of interest information and the second point of interest information to a server, so that the server determines the classification result according to the first point of interest information and the second point of interest information; receiving the classification result sent by the server; The classification result is displayed.

2. A data processing method, characterized in that: Applied to a server, the data processing method includes: Obtaining first point of interest information and second point of interest information, where the first point of interest information and the second point of interest information are from different geospatial databases, the first point of interest information including: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information including: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; Inputting the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, where the first feature vector is used to represent the fusion of the first description information and the second description information; Inputting the first description information, the first adjacent point of interest information, the second description information, and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, where the second feature vector is used to characterize the dependency relationship between the first point of interest and the second point of interest on adjacent points of interest; The first feature vector and the second feature vector are input into a classification model for classification to obtain a classification result, where the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest.

3. The data processing method according to claim 2, characterized in that: The first description information includes first attribute information, the second description information includes second attribute information, and the first network model includes a semantic analysis sub-model; the first description information and the second description information are input into the first network model for feature extraction to obtain a first feature vector, including: The first attribute information is processed by the text processing layer of the semantic analysis sub-model according to the attribute type to obtain a first segmented text; Performing word segmentation processing on the second attribute information by the text processing layer according to the attribute type to obtain a second word segmentation text; Combining the first segmented text and the second segmented text through the text processing layer to obtain a combined text; Encoding the combined text through the encoder of the semantic analysis sub-model to obtain a first feature sub-vector; The first eigenvector is obtained according to the first eigenvector.

4. The data processing method according to claim 3, wherein: The first description information further includes: first spatial geographic information; the second description information further includes: second spatial geographic information; the first network model further includes: a distance determination submodel; and obtaining the first feature vector based on the first feature subvector includes: Inputting the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain a second feature sub-vector, where the second feature sub-vector is used to represent the distance between the first spatial geographic information and the second spatial geographic information; The first eigenvector and the second eigenvector are concatenated to obtain the first eigenvector.

5. The data processing method according to claim 4, characterized in that: The step of inputting the first spatial geographic information and the second spatial geographic information into the distance determination sub-model for feature extraction to obtain a second feature sub-vector includes: Determining the distance between the first point of interest and the second point of interest using the distance determination sub-model according to the first spatial geographic information and the second spatial geographic information; The distance is vectorized using a distance determination sub-model to obtain the second characteristic sub-vector.

6. The data processing method according to any one of claims 2 to 5, characterized in that: The step of inputting the first feature vector and the second feature vector into a classification model for classification to obtain a classification result includes: Using the second network model to perform feature extraction on the first description information and the first adjacent point of interest information to obtain a third feature subvector, wherein the third feature subvector is used to characterize the dependency relationship between the first adjacent point of interest information and the first point of interest; performing feature extraction on the second description information and the second adjacent point of interest information using the second network model to obtain a fourth feature subvector, wherein the fourth feature subvector is used to characterize a dependency relationship between the second adjacent point of interest information and the second point of interest; Feature concatenation is performed on the third eigenvector and the fourth eigenvector to obtain the second eigenvector.

7. The data processing method according to claim 6, characterized in that: The using the second network model to extract features from the first description information and the first adjacent interest point information to obtain a third feature sub-vector includes: Using the second network model to semantically represent the first description information to obtain a first representation vector; Using the second network model to perform semantic representation on each piece of the first adjacent point of interest information, to obtain a second representation vector corresponding to each piece of the first adjacent point of interest information; The second network model is used to determine the fusion features of the second characterization vector and the first characterization vector to obtain the third feature sub-vector.

8. The data processing method according to any one of claims 2 to 5, characterized in that: The acquiring of the first point of interest information and the second point of interest information includes: Determining any point of interest information in a first geospatial database as the first point of interest information; The second point of interest information is determined in a second geospatial database based on the first point of interest information, and a similarity between the first point of interest information and the second point of interest information is greater than a preset threshold.

9. A data processing method, characterized in that: Applied to a terminal, the data processing method includes: Obtaining first point of interest information and second point of interest information, where the first point of interest information and the second point of interest information are from different geospatial databases, the first point of interest information including: first description information of the first point of interest and first adjacent point of interest information of the first point of interest, and the second point of interest information including: second description information of the second point of interest and second adjacent point of interest information of the second point of interest; Sending the first point of interest information and the second point of interest information to a server, so that the server determines a classification result based on the first point of interest information and the second point of interest information, wherein the classification result is used to indicate whether the first point of interest information and the second point of interest information describe the same point of interest; wherein the server obtains the classification result by inputting the first description information and the second description information into a first network model for feature extraction to obtain a first feature vector, inputting the first description information, the first adjacent point of interest information, the second description information, and the second adjacent point of interest information into a second network model for feature extraction to obtain a second feature vector, and inputting the first feature vector and the second feature vector into a classification model for classification; wherein the first feature vector is used to represent the fusion of the first description information and the second description information; and the second feature vector is used to represent the dependency relationship between the first point of interest and the second point of interest on adjacent points of interest; receiving the classification result sent by the server; The classification result is displayed.

10. The data processing method according to claim 9, characterized in that: The displaying of the classification result includes: If the first point of interest information and the second point of interest information describe the same point of interest, integrating the first point of interest information and the second point of interest information to obtain third point of interest information, the third point of interest information including: third attribute information, third spatial geographic information, and third adjacent point of interest information; Displaying the third attribute information, the third spatial geographic information, and the third adjacent point of interest information in the form of a map; If the first point of interest information and the second point of interest information describe different points of interest, the first point of interest information and the second point of interest information are displayed separately in the form of maps.

11. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the data processing method according to any one of claims 2 to 10 when executing the computer program.

Images