Method, device and electronic device for determining duplicate points of interest on a map

Abstract

The present invention discloses a method, device and electronic device for determining duplicate points of interest on a map. The method comprises: obtaining a first attribute feature of a first map point of interest to be determined and a second attribute feature of a second map point of interest to be determined; and obtaining the similarity between the first attribute feature and the attribute features of the same type in the second attribute feature, and using the similarity as the cross feature between the first map point of interest and the second map point of interest; further, inputting the first attribute feature, the second attribute feature and the cross feature into a pre-trained machine learning model, determining whether the first map point of interest and the second map point of interest are the same through the machine learning model and outputting the determination result, so that the determination of duplicate points of interest is based on multiple dimensions such as the attribute features of the points of interest and the similarity between the attribute features in a globally optimal manner, thereby solving the technical problem of poor accuracy in the prior art of POI duplicate determination and improving the accuracy of duplicate determination.

Description

Technical Field

[0001] The present invention relates to the field of geographic information systems, and in particular to a method, a device and an electronic device for determining duplicate points of interest on a map. Background Art

[0002] In the field of geographic information systems, points of interest (POIs) are usually set to describe points on a map. On a map, each point that needs to be retrieved can be a POI, which is stored in a database. When a user needs to search, the corresponding POI is found based on the keywords entered by the user and displayed on the search interface.

[0003] There are many ways to express a POI. For example, "Tsinghua University", "North Gate of Tsinghua University", and "No. 30 Shuangqing Road, Haidian District, Beijing" may represent the same POI. Therefore, a process of POI duplicate detection is needed to determine whether two POIs in the database are the same point, so as to provide more accurate results for retrieval. Existing methods for POI duplicate detection are usually implemented through conditional judgment methods. For example, the name similarity, address similarity, coordinate distance and other attribute features corresponding to the two POIs are calculated respectively. If the name similarity meets condition 1, the address similarity meets condition 2, and the coordinate distance meets condition 3, the two POIs are considered to be the same or different POIs. This conditional judgment method is often only applicable to simple POI duplicate detection. For POIs containing multiple attribute features, the complexity of condition setting increases exponentially and its comprehensiveness becomes worse and worse, resulting in a greatly reduced accuracy of duplicate detection. Summary of the invention

[0004] The embodiments of the present invention provide a method, device and electronic device for determining duplicate points of interest on a map, which are used to solve the technical problem of poor accuracy in determining duplicate POIs in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a method for determining duplicate points of interest on a map, the method comprising:

[0006] Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point to be determined, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1;

[0007] Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map interest point and the second map interest point;

[0008] The first attribute feature, the second attribute feature and the cross feature are input into a pre-trained machine learning model, and the machine learning model is used to determine whether the first map point of interest and the second map point of interest are the same, and a determination result is output.

[0009] Optionally, before outputting the determination result, the method further includes:

[0010] comparing the first map point of interest and the second map point of interest with a preset whitelist and / or blacklist of map points of interest;

[0011] If the preset whitelist and / or blacklist of map points of interest contain a pair of map points of interest that are identical to the first map point of interest and the second map point of interest, obtaining a duplicate identifier of the pair of map points of interest in the preset whitelist and / or blacklist of map points of interest;

[0012] Modifying the determination result based on the duplicate identification;

[0013] The outputting the determination result includes: outputting the corrected determination result.

[0014] Optionally, the training method of the pre-trained machine learning model includes:

[0015] Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same;

[0016] For each reference instance, feature extraction and vectorization are performed on the attribute information of the two map interest points to obtain attribute features of each map interest point; similarity between the attribute features of the same type corresponding to the two map interest points is obtained, and the similarity is used as the cross feature between the two interest points;

[0017] The attribute features, the intersection features and the repeated identifiers corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

[0018] Optionally, the training samples include positive samples and negative samples in a preset ratio, wherein the two map interest points corresponding to the positive samples are the same, and the two map interest points corresponding to the negative samples are different.

[0019] Optionally, the attribute features include: the name, address, coordinates and / or category of the map point of interest;

[0020] The cross-features include: name similarity, address similarity, distance between coordinates and / or category similarity.

[0021] Optionally, the machine learning model includes:

[0022] Linear regression, decision trees, random forests, gradient boosted trees, or neural networks.

[0023] In a second aspect, an embodiment of the present invention provides a device for determining duplicate points of interest on a map, the device comprising:

[0024] An acquiring unit, configured to acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point to be determined, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1;

[0025] a calculation unit, configured to obtain a similarity between the first attribute feature and an attribute feature of the same type in the second attribute feature, and use the similarity as a cross feature between the first map point of interest and the second map point of interest;

[0026] a judgment unit, configured to input the first attribute feature, the second attribute feature, and the cross feature into a pre-trained machine learning model, and determine whether the first map point of interest and the second map point of interest are the same through the machine learning model to obtain a determination result;

[0027] An output unit is used to output the determination result.

[0028] Optionally, the device further comprises:

[0029] a comparison unit, configured to compare the first map point of interest and the second map point of interest with a preset white list and / or black list of map points of interest before outputting a determination result; if the preset white list and / or black list of map points of interest contain a map point of interest pair identical to the first map point of interest and the second map point of interest, obtain a duplicate identifier of the map point of interest pair in the preset white list and / or black list of map points of interest;

[0030] The determination unit is further configured to modify the determination result based on the repetition mark;

[0031] The output unit is also used to output the corrected determination result.

[0032] Optionally, the device further comprises a training unit, wherein the training unit is used to:

[0033] Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same;

[0034] For each reference instance, feature extraction and vectorization are performed on the attribute information of the two map interest points to obtain attribute features of each map interest point; similarity between the attribute features of the same type corresponding to the two map interest points is obtained, and the similarity is used as the cross feature between the two interest points;

[0035] The attribute features, the intersection features and the repeated identifiers corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

[0036] Optionally, the training samples include positive samples and negative samples in a preset ratio, wherein the two map interest points corresponding to the positive samples are the same, and the two map interest points corresponding to the negative samples are different.

[0037] Optionally, the attribute features include: the name, address, coordinates and / or category of the map point of interest;

[0038] The cross-features include: name similarity, address similarity, distance between coordinates and / or category similarity.

[0039] Optionally, the machine learning model includes: linear regression, decision tree, random forest, gradient boosting tree or neural network.

[0040] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, wherein the one or more programs include instructions for performing the following operations:

[0041] Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point to be determined, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1;

[0042] Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map interest point and the second map interest point;

[0043] The first attribute feature, the second attribute feature and the cross feature are input into a pre-trained machine learning model, and the machine learning model is used to determine whether the first map point of interest and the second map point of interest are the same, and a determination result is output.

[0044] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0045] Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point to be determined, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1;

[0046] Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map interest point and the second map interest point;

[0047] The first attribute feature, the second attribute feature and the cross feature are input into a pre-trained machine learning model, and the machine learning model is used to determine whether the first map point of interest and the second map point of interest are the same, and a determination result is output.

[0048] The above one or more technical solutions in the embodiments of the present application have at least the following technical effects:

[0049] An embodiment of the present application provides a method for determining duplicate points of interest on a map, which includes obtaining a first attribute feature of a first map point of interest to be determined and a second attribute feature of a second map point of interest to be determined; and obtaining the similarity between the first attribute feature and the attribute features of the same type in the second attribute feature, and using the similarity as the cross feature between the first map point of interest and the second map point of interest; further, inputting the first attribute feature, the second attribute feature and the cross feature into a pre-trained machine learning model, and determining whether the first map point of interest and the second map point of interest are the same through the machine learning model and outputting a determination result, so that the determination of duplicate points of interest is based on multiple dimensional features of the points of interest and is determined in a globally optimal manner, and the accuracy of duplicate determination is greatly improved, thereby solving the technical problem of poor accuracy in POI duplicate determination in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flowchart of a method for determining duplicate points of interest on a map provided in an embodiment of the present application;

[0051] Figure 2 A block diagram of a device for determining duplicate points of interest on a map provided in an embodiment of the present application;

[0052] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0053] In the technical solution provided in the embodiment of the present application, a method for duplicate detection of map points of interest is provided, which obtains attribute features and cross-features of multiple dimensions of two points of interest, and detects duplicate points of interest through a pre-trained machine learning model, so that the duplicate detection of points of interest is performed in a globally optimal manner, so as to solve the technical problem of poor accuracy of POI duplicate detection in the prior art.

[0054] The main implementation principles, specific implementation methods and corresponding beneficial effects of the technical solutions of the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0055] Example

[0056] Please refer to Figure 1 , the present application embodiment provides a method for determining duplicate points of interest on a map, the method comprising:

[0057] S11: Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point to be determined;

[0058] S12: Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map point of interest and the second map point of interest;

[0059] S13: Input the first attribute feature, the second attribute feature and the cross feature into a pre-trained machine learning model, determine whether the first map point of interest and the second map point of interest are the same through the machine learning model, and output the determination result.

[0060] For the first and second map points of interest to be determined as duplicate, when executing S11, the first attribute feature of the first map point of interest and the second attribute feature of the second map point of interest can be obtained according to the attribute information of the map point of interest POI. The first attribute feature and the second attribute feature both include n types of attribute features, n≥1.

[0061] Among them, the map point of interest POI can be a road, a bridge, a building, a restaurant, a bus stop, etc. The attribute information of each POI stored in the database mainly includes: name, address, coordinates, etc. Further, in order to distinguish POI more effectively, the attribute information of POI can also include: number, alias, category, brand, etc. These attribute information can be obtained manually (through field investigation, manual contact confirmation, etc.) or crawled through the Internet. When obtaining attribute features based on the attribute information of POI, each attribute information can be directly vectorized to obtain the corresponding attribute features; or each attribute information can be first normalized such as feature extraction, information classification, etc., and then the normalized information is vectorized to obtain the corresponding attribute features. Among them, since the attribute information of POI is often not standardized or contains some noise, the attribute features obtained by normalizing first and then vectorizing are more accurate, which can improve the accuracy of duplicate judgment. For example, to normalize a name, you can first extract the name keywords and then vectorize them. Assuming that the name of the first point of interest is "KFC opposite Wanda Plaza", you can extract "Wanda Plaza" and "KFC" and vectorize the extracted text to obtain its name attribute features as its first attribute features. To normalize an address, you can extract features such as province, city, district, county, road, street, building, house number, etc. from the address, and then vectorize the extracted features to obtain its address attribute features as its first attribute features.

[0062] It should be noted that this embodiment does not limit the specific method of vectorization, and vectorization can be performed through one-hot vectorization training, word2vec vectorization training, LSTM vectorization training, etc.

[0063] For the first attribute feature and the second attribute feature obtained in S11, continue to execute S12 to obtain the cross-feature between the first map point of interest and the second map point of interest. Usually, there are multiple attribute features for the first map point of interest and the second map point of interest, and they include attribute features of the same type, for example, both include attribute features: name, address, category, etc. The similarity between two attribute features of the same type can often reflect the similarity between map points of interest POIs to a certain extent. For this reason, the similarity between the attribute features of the same type in the first attribute feature and the second attribute feature is obtained, and the similarity is used as the cross-feature between the first map point of interest and the second map point of interest, and the first attribute feature and the second attribute feature are used together to perform POI duplicate detection, and the accuracy of POI duplicate detection is improved through features in multiple dimensions.

[0064] The similarity between attribute features can be obtained by calculating their vector similarity, or by other similarity calculation methods, which is not specifically limited in this embodiment. For example, for the name, address, category and other attribute features contained in the first attribute feature and the second attribute feature, since these attribute features are text-like features represented by vectors, the similarity between the corresponding two vectors can be directly calculated, and the vector similarity can be used as the similarity between the attribute features. For attribute features such as coordinates and brands, their similarity can be calculated in a personalized way, such as: for coordinates, the distance between the two coordinates can be calculated, and the distance between the coordinates can be used as the similarity between the coordinate attribute features. For brands, the correlation between brands can be calculated to calculate their similarity. For example, although the characters of "Kentucky" and "KFC" are completely different, they are the same brand, and their correlation is 100%, and their correlation can be used as their similarity.

[0065] After obtaining the cross feature between the first map point of interest and the second map point of interest in S12, S13 is further executed to input the obtained first attribute feature, second attribute feature and cross feature into a pre-trained machine learning model, and the machine learning model is used to determine whether the first map point of interest and the second map point of interest are the same, and the determination result is output.

[0066] The pre-trained machine learning model is obtained by training in steps 1 to 3 as follows:

[0067] Step 1: Get a reference instance set.

[0068] The reference instance set includes multiple reference instances, that is, multiple POI data pairs. Each reference instance in the reference instance set includes two map points of interest POIs, attribute information of the two map points of interest POIs, and a repeated identifier indicating whether the two map points of interest POIs are the same.

[0069] When obtaining reference instances, two map points of interest POIs and their attribute information that have been manually confirmed as duplicate results can be obtained as reference instances, two map points of interest POIs with a master-substructure relationship (i.e., one POI contains another POI) and their attribute information can be obtained as reference instances, and the POI that reports an error during map retrieval, the correct POI corresponding to the POI that reports the error, and their corresponding attribute information can also be obtained as reference instances. The multiple reference instances obtained are packaged to form a reference instance set. When forming a reference instance set, positive examples (the two POIs in the reference instance are the same) and / or negative examples (the two POIs in the reference instance are different) can be obtained as elements in the reference instance set.

[0070] Step 2: Cross feature extraction.

[0071] For each reference instance in the reference instance set, feature extraction and vectorization are performed on the attribute information of the two map interest points in each reference instance to obtain the attribute features of each map interest point; then, the similarity between the attribute features of the same type corresponding to the two map interest points in each reference instance is obtained, and the similarity is used as the cross feature between the corresponding two interest points. For the specific cross feature extraction method, refer to S12.

[0072] Step 3: Sample training.

[0073] The attribute features, cross-features and repeated identifications corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model. Specifically, the attribute features and cross-features of the two map interest points of each reference instance are used as input data of the machine learning model, and the repeated identifications of each reference instance are used as output results of the machine learning model to train the machine learning model, and the training of the machine learning model is completed through multiple training samples. Among them, the machine learning model can adopt models such as linear regression, decision tree, random forest, gradient boosting tree or neural network, and the embodiments of the present application do not limit the type of machine learning model.

[0074] In the specific implementation process, the ratio of the number of positive samples (the two POIs in the reference instance are the same) and negative samples (the two POIs in the reference instance are different) will affect the accuracy of the final machine learning model results. If there are too many positive samples, during model training, because the model algorithm is to solve the global optimal solution, it may cause the trained model to tend to predict positive examples, and vice versa. For this reason, when performing sample training, a preset ratio of positive and negative samples is selected so that the positive and negative samples used for machine learning model training are as close as possible to the real-world sample distribution. Specifically, the preset ratio of positive and negative samples can be 1:1, that is, the ratio of the number of positive and negative samples is 1:1, so as to improve the accuracy of the machine learning model obtained by training.

[0075] The following is a specific example to illustrate the training of the machine learning model:

[0076] Assume that the positive example obtained is:

[0077]

[0078]

[0079] Negative examples are:

[0080]

[0081] Among them, the repeated mark 1 represents a positive example, the name and address are the attribute information of the POI, and the attribute information such as coordinates and telephone number is omitted here.

[0082] Feature extraction is performed on the above reference examples. The following table shows the feature extraction of the address of the positive example. The feature extraction of the address of the negative example is similar:

[0083]

[0084]

[0085] Next, cross-feature extraction is performed on the two POIs in each reference instance (including positive and negative examples). The cross-feature extraction of the positive example is shown in the following table, and the cross-feature extraction of the negative example is similar:

[0086]

[0087] Finally, the extracted attribute features and cross features are vectorized to obtain the final training samples for machine learning model training and a trained machine learning model. It should be noted that in this example, the attribute features are first cross-feature extracted and then vectorized. During the implementation process, the extracted attribute features can also be vectorized first, and then cross-feature extraction can be performed based on the vectorized attribute features to directly obtain vectorized cross features and avoid vectorization.

[0088] For the training of machine learning models, when new attribute features appear, we only need to add a new dimension of features to the original training data to form new training samples, and then input the new training samples into machine learning for training. The trained model can easily solve the problem of POI duplication with the newly added features. While ensuring the accuracy of the model judgment, it also shortens the time for later maintenance of the model.

[0089] After S13 uses the trained machine learning model to determine whether the first map point of interest and the second map point of interest are the same, the determination result can be directly output, or the following steps can be further performed to correct the determination result, and the corrected determination result can be output to further improve the accuracy of the determination result.

[0090] The correction step is specifically as follows: compare the first map point of interest and the second map point of interest with a preset map point of interest whitelist and / or blacklist; if the preset map point of interest whitelist and / or blacklist contains a map point of interest pair that is the same as the first map point of interest and the second map point of interest, obtain the duplicate identifier of the corresponding map point of interest pair in the preset map point of interest whitelist and / or blacklist; and correct the judgment result of the machine learning model based on the obtained duplicate identifier. Among them, the preset map point of interest whitelist and / or blacklist is a special case of POI, that is, the same map point of interest and / or different points of interest that do not meet the conventional duplicate judgment rules are collected in advance to form a map point of interest pair. The two map points of interest contained in the map point of interest pair in the whitelist may have very different names, but correspond to the same location. The corresponding duplicate identifier indicates that the two map points of interest are the same, which can be represented by "1"; the two map points of interest contained in the map point of interest pair in the blacklist may have very similar names, but correspond to different locations. The corresponding duplicate identifier indicates that the two map points of interest are different, which can be represented by "0". The first map point of interest is the same as one map point of interest in a map point of interest pair, and the second map point of interest is the same as another map point of interest in the same map point of interest pair, that is, the first map point of interest and the second map point of interest are the same as the map point of interest pair. If the duplicate mark of the map point of interest pair that is the same as the first map point of interest and the second map point of interest is "1", the determination result is corrected to be the same. If the duplicate mark of the map point of interest pair that is the same as the first map point of interest and the second map point of interest is "0", the determination result is corrected to be different. By adding a special case correction step after the machine learning model determines the duplicate POI, the inaccurate judgment of special cases by machine learning can be supplemented, thereby further improving the overall accuracy.

[0091] A method for determining duplicate points of interest on a map is provided in the above embodiment. The present application also provides a device for determining duplicate points of interest on a map. Figure 2 , the device comprises:

[0092] An acquiring unit 21 is used to acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point to be determined, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1;

[0093] A calculation unit 22, configured to obtain a similarity between the first attribute feature and an attribute feature of the same type in the second attribute feature, and use the similarity as a cross feature between the first map point of interest and the second map point of interest;

[0094] A judgment unit 23, configured to input the first attribute feature, the second attribute feature and the cross feature into a pre-trained machine learning model, and determine whether the first map point of interest and the second map point of interest are the same through the machine learning model to obtain a determination result;

[0095] The output unit 24 is used to output the determination result.

[0096] As an optional implementation, the device may further include a comparison unit 25, which is used to compare the first map point of interest and the second map point of interest with a preset map point of interest whitelist and / or blacklist before outputting the determination result; if the preset map point of interest whitelist and / or blacklist contains a map point of interest pair that is the same as the first map point of interest and the second map point of interest, obtain a duplicate identifier of the map point of interest pair in the preset map point of interest whitelist and / or blacklist. Further, the judgment unit 23 is also used to correct the determination result based on the duplicate identifier obtained by the comparison unit 25; the output unit 24 is also used to output the corrected determination result.

[0097] As an optional implementation, the device may further include a training unit 26, which is used to:

[0098] Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same;

[0099] For each reference instance, feature extraction and vectorization are performed on the attribute information of the two map interest points to obtain attribute features of each map interest point; similarity between the attribute features of the same type corresponding to the two map interest points is obtained, and the similarity is used as the cross feature between the two interest points;

[0100] The attribute features, the intersection features and the repeated identification corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

[0101] As an optional implementation, the training samples of the machine learning model include positive samples and negative samples in a preset ratio, wherein the two map interest points corresponding to the positive samples are the same, and the two map interest points corresponding to the negative samples are different. The machine learning model includes: linear regression, decision tree, random forest, gradient boosting tree or neural network.

[0102] As an optional implementation, the attribute features include: name, address, coordinates and / or category of the map interest point; the intersection features include: name similarity, address similarity, distance between coordinates and / or category similarity.

[0103] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0104] Figure 3 1 is a block diagram of an electronic device 800 for implementing duplicate detection of map points of interest according to an exemplary embodiment. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0105] Reference Figure 3 , the electronic device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / display (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0106] The processing component 802 generally controls the overall operation of the electronic device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0107] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage 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.

[0108] The power supply component 806 provides power to the various components of the electronic device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 800.

[0109] The multimedia component 808 includes a screen that provides a display interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0110] The audio component 810 is configured to present and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the electronic device 800 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for presenting audio signals.

[0111] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

[0112] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the electronic device 800. For example, the sensor assembly 814 can detect the open / closed state of the device 800, the relative positioning of components, such as the display and keypad of the electronic device 800, and the sensor assembly 814 can also detect the position change of the electronic device 800 or a component of the electronic device 800, the presence or absence of contact between the user and the electronic device 800, the orientation or acceleration / deceleration of the electronic device 800, and the temperature change of the electronic device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0113] The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 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.

[0114] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0115] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by a processor 820 of an electronic device 800 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0116] A non-transitory computer-readable storage medium, when the instructions in the storage medium are executed by a processor of a mobile terminal, enables the mobile terminal to perform a method for determining duplicate points of interest on a map, the method comprising:

[0117] Obtain a first attribute feature of a first map point of interest to be determined and a second attribute feature of a second map point of interest to be determined, wherein the first attribute feature and the second attribute feature both contain n types of attribute features, where n≥1; obtain the similarity between the first attribute feature and the attribute features of the same type in the second attribute feature, and use the similarity as a cross feature between the first map point of interest and the second map point of interest; input the first attribute feature, the second attribute feature and the cross feature into a pre-trained machine learning model, determine whether the first map point of interest and the second map point of interest are the same through the machine learning model, and output a determination result.

[0118] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0119] It should be understood that the present invention is not limited to the precise structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for determining duplicate points of interest on a map, characterized in that: The method comprises: Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1; wherein the attribute feature of the map interest point is obtained by normalizing and vectorizing the attribute information of the map interest point; the normalization process includes: feature extraction and information classification; Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map interest point and the second map interest point; Inputting the first attribute feature, the second attribute feature, and the cross feature into a pre-trained machine learning model, and determining whether the first map point of interest and the second map point of interest are the same by using the machine learning model to obtain a determination result; Comparing the first map point of interest and the second map point of interest with a preset whitelist and / or blacklist of map points of interest; the whitelist contains pre-collected pairs of identical map points of interest that do not conform to conventional duplicate determination rules, and the blacklist contains pre-collected pairs of map points of interest formed by different points of interest that do not conform to conventional duplicate determination rules; If the preset whitelist and / or blacklist of map points of interest contain a pair of map points of interest that are the same as the first map point of interest and the second map point of interest, obtaining a duplicate identifier of the pair of map points of interest in the preset whitelist and / or blacklist of map points of interest; and correcting the determination result based on the duplicate identifier; outputting the determination result; The training method of the pre-trained machine learning model includes: Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same; For each reference instance, normalizing and vectorizing the attribute information of the two map interest points to obtain attribute features of each map interest point; Obtaining the similarity between the attribute features of the same type corresponding to the two map interest points, and using the similarity as the cross feature between the two interest points; The attribute features, the intersection features and the repeated identifiers corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

2. The method according to claim 1, characterized in that The training samples include positive samples and negative samples in a preset ratio, wherein the two map interest points corresponding to the positive samples are the same, and the two map interest points corresponding to the negative samples are different.

3. The method according to any one of claims 1 to 2, characterized in that: The attribute features include: the name, address, coordinates and / or category of the map point of interest; The cross-features include: name similarity, address similarity, distance between coordinates and / or category similarity.

4. The method according to any one of claims 1 to 2, characterized in that: The machine learning model includes: Linear regression, decision trees, random forests, gradient boosted trees, or neural networks.

5. A device for determining duplicate points of interest on a map, characterized in that: The device comprises: An acquisition unit is used to acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1; wherein the attribute feature of the map interest point is obtained by normalizing and vectorizing the attribute information of the map interest point; the normalization process includes: feature extraction and information classification; a calculation unit, configured to obtain a similarity between the first attribute feature and an attribute feature of the same type in the second attribute feature, and use the similarity as a cross feature between the first map point of interest and the second map point of interest; a judgment unit, configured to input the first attribute feature, the second attribute feature, and the cross feature into a pre-trained machine learning model, and determine whether the first map point of interest and the second map point of interest are the same through the machine learning model to obtain a judgment result; a comparison unit, configured to compare the first map point of interest and the second map point of interest with a preset white list and / or black list of map points of interest; the white list contains pre-collected identical map point of interest pairs that do not conform to conventional duplicate determination rules, and the black list contains pre-collected map point of interest pairs formed by different points of interest that do not conform to conventional duplicate determination rules; if the preset white list and / or black list of map points of interest contains the same map point of interest pair as the first map point of interest and the second map point of interest, obtaining a duplicate identifier of the map point of interest pair in the preset white list and / or black list of map points of interest; The determination unit is further configured to modify the determination result based on the repetition mark; An output unit, used for outputting the determination result; The training unit is used to: Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same; For each reference instance, normalize and vectorize the attribute information of the two map interest points to obtain the attribute features of each map interest point; obtain the similarity between the attribute features of the same type corresponding to the two map interest points, and use the similarity as the cross feature between the two interest points; The attribute features, the intersection features and the repeated identification corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

6. The device according to claim 5, characterized in that The training samples include positive samples and negative samples in a preset ratio, wherein the two map interest points corresponding to the positive samples are the same, and the two map interest points corresponding to the negative samples are different.

7. The device according to any one of claims 5 to 6, characterized in that: The attribute features include: the name, address, coordinates and / or category of the map point of interest; The cross-features include: name similarity, address similarity, distance between coordinates and / or category similarity.

8. The device according to any one of claims 5 to 6, characterized in that: The machine learning model includes: Linear regression, decision trees, random forests, gradient boosted trees, or neural networks.

9. An electronic device, characterized in that: The system comprises a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, wherein the one or more programs include instructions for performing the following operations: Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1; wherein the attribute feature of the map interest point is obtained by normalizing and vectorizing the attribute information of the map interest point; the normalization process includes: feature extraction and information classification; Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map interest point and the second map interest point; Inputting the first attribute feature, the second attribute feature, and the cross feature into a pre-trained machine learning model, and determining whether the first map point of interest and the second map point of interest are the same by using the machine learning model to obtain a determination result; Comparing the first map point of interest and the second map point of interest with a preset whitelist and / or blacklist of map points of interest; the whitelist contains pre-collected pairs of identical map points of interest that do not conform to conventional duplicate determination rules, and the blacklist contains pre-collected pairs of map points of interest formed by different points of interest that do not conform to conventional duplicate determination rules; If the preset whitelist and / or blacklist of map points of interest contain a pair of map points of interest that are the same as the first map point of interest and the second map point of interest, obtaining a duplicate identifier of the pair of map points of interest in the preset whitelist and / or blacklist of map points of interest; and correcting the determination result based on the duplicate identifier; outputting the determination result; The training method of the pre-trained machine learning model includes: Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same; For each reference instance, normalize and vectorize the attribute information of the two map interest points to obtain the attribute features of each map interest point; obtain the similarity between the attribute features of the same type corresponding to the two map interest points, and use the similarity as the cross feature between the two interest points; The attribute features, the intersection features and the repeated identifiers corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the following steps are implemented: Acquire a first attribute feature of a first map interest point to be determined and a second attribute feature of a second map interest point, wherein the first attribute feature and the second attribute feature both include n types of attribute features, where n≥1; wherein the attribute feature of the map interest point is obtained by normalizing and vectorizing the attribute information of the map interest point; the normalization process includes: feature extraction and information classification; Acquire the similarity between the first attribute feature and the attribute feature of the same type in the second attribute feature, and use the similarity as the cross feature between the first map interest point and the second map interest point; Inputting the first attribute feature, the second attribute feature, and the cross feature into a pre-trained machine learning model, and determining whether the first map point of interest and the second map point of interest are the same by using the machine learning model to obtain a determination result; Comparing the first map point of interest and the second map point of interest with a preset whitelist and / or blacklist of map points of interest; the whitelist contains pre-collected pairs of identical map points of interest that do not conform to conventional duplicate determination rules, and the blacklist contains pre-collected pairs of map points of interest formed by different points of interest that do not conform to conventional duplicate determination rules; If the preset whitelist and / or blacklist of map points of interest contain a pair of map points of interest that are the same as the first map point of interest and the second map point of interest, obtaining a duplicate identifier of the pair of map points of interest in the preset whitelist and / or blacklist of map points of interest; and correcting the determination result based on the duplicate identifier; outputting the determination result; The training method of the pre-trained machine learning model includes: Acquire a reference instance set, each reference instance in the reference instance set including two map interest points, attribute information of the two map interest points, and a repeated identifier indicating whether the two map interest points are the same; For each reference instance, normalize and vectorize the attribute information of the two map interest points to obtain the attribute features of each map interest point; obtain the similarity between the attribute features of the same type corresponding to the two map interest points, and use the similarity as the cross feature between the two interest points; The attribute features, the intersection features and the repeated identifiers corresponding to the two map interest points of each reference instance are used as training samples to train the machine learning model.

Images