POI classification and model training method and device, equipment and storage medium

Abstract

The present disclosure provides a POI classification and model training method and device, equipment and a storage medium, relates to the technical field of data processing, and in particular to the technical field of electronic maps, big data and the like. The POI classification method comprises: if a first POI does not satisfy a preset rule and a second POI associated with the first POI does not exist in target data, attribute data of the first POI is obtained; wherein the first POI is a POI to be classified in map data, and the target data is different from the map data; a POI classification model is obtained, the POI classification model being trained based on sample POIs in the target data; and the attribute data of the first POI is processed by using the POI classification model to determine a target category of the first POI. The present disclosure can improve the accuracy of POI classification.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, specifically to the fields of electronic maps, big data, etc., and particularly to a method, apparatus, device, and storage medium for POI classification and model training. Background Technology

[0002] In Geographic Information Systems (GIS), a Point of Interest (POI) is a meaningful point on a map, such as a house, shop, or bus stop. Each POI contains information such as its name, coordinates, and category. POI information allows users to obtain more information before traveling, enabling them to make better travel decisions.

[0003] In related technologies, POIs are usually classified according to the needs of map products. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, and storage medium for POI classification and model training.

[0005] According to one aspect of this disclosure, a POI classification method is provided, comprising: if a first POI does not meet a preset rule and there is no second POI associated with the first POI in the target data, obtaining attribute data of the first POI; wherein the first POI is a POI to be classified in map data, and the target data is different from the map data; obtaining a POI classification model, the POI classification model being trained based on sample POIs in the target data; and using the POI classification model to process the attribute data of the first POI to determine the target category of the first POI.

[0006] According to another aspect of this disclosure, a method for training a POI classification model is provided, comprising: acquiring training data, the training data including: attribute data of sample POIs and the true category of the sample POIs, wherein the sample POIs are POIs in target data that are associated with POIs in map data, and the true category is the category of the sample POIs in the target data, wherein the target data is different from the map data; processing the attribute data of the sample POIs using a POI classification model to determine the predicted category of the sample POIs; constructing a loss function based on the predicted category and the true category; and adjusting the model parameters of the POI classification model based on the loss function until a preset condition is met to obtain a final POI classification model, wherein the final POI classification model is used to determine the target category of the POI to be classified in the map data.

[0007] According to another aspect of this disclosure, a POI classification apparatus is provided, comprising: a first acquisition module, configured to acquire attribute data of the first POI if the first POI does not meet a preset rule and there is no second POI associated with the first POI in the target data; wherein the first POI is a POI to be classified in map data, and the target data is different from the map data; a second acquisition module, configured to acquire a POI classification model, the POI classification model being trained based on sample POIs in the target data; and a first determination module, configured to process the attribute data of the first POI using the POI classification model to determine the target category of the first POI.

[0008] According to another aspect of this disclosure, a training apparatus for a POI classification model is provided, comprising: an acquisition module for acquiring training data, the training data including: attribute data of sample POIs and the true category of the sample POIs, wherein the sample POIs are POIs in target data that are associated with POIs in map data, and the true category is the category of the sample POIs in the target data, wherein the target data is different from the map data; a prediction module for processing the attribute data of the sample POIs using the POI classification model to determine the predicted category of the sample POIs; a construction module for constructing a loss function based on the predicted category and the true category; and an adjustment module for adjusting the model parameters of the POI classification model based on the loss function until preset conditions are met to obtain a final POI classification model, wherein the final POI classification model is used to determine the target category of the POI to be classified in the map data.

[0009] According to another aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to said at least one processor; wherein the memory stores instructions executable by said at least one processor, said instructions being executed by said at least one processor to enable said at least one processor to perform the method as described in any of the foregoing aspects.

[0010] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method according to any of the preceding aspects.

[0011] According to another aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method according to any of the preceding aspects.

[0012] According to the technical solution disclosed herein, the accuracy of POI classification can be improved.

[0013] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0014] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0015] Figure 1 This is a schematic diagram based on the first embodiment of the present disclosure;

[0016] Figure 2 This is a schematic diagram illustrating an application scenario provided according to embodiments of this disclosure;

[0017] Figure 3 This is a schematic diagram of the overall architecture of the POI classification method according to an embodiment of the present disclosure;

[0018] Figure 4 This is a schematic diagram according to the second embodiment of the present disclosure;

[0019] Figure 5 This is a schematic diagram according to the third embodiment of the present disclosure;

[0020] Figure 6 This is a schematic diagram according to the fourth embodiment of the present disclosure;

[0021] Figure 7 This is a schematic diagram according to the fifth embodiment of the present disclosure;

[0022] Figure 8 This is a schematic diagram of an electronic device used to implement the POI classification method or POI classification model training method of the embodiments of this disclosure. Detailed Implementation

[0023] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0024] In related technologies, POIs are usually classified according to the needs of map products. For example, the primary classification is mainly divided into: food, hotels, shopping, and lifestyle services.

[0025] However, the accuracy of the current map classification system is insufficient. For example, for a wedding photography studio A, the current map classification system will classify A as "photo studio". If a user searches for "wedding photography" on the map, because A is classified as "photo studio", the search results may not include A, resulting in problems with the accuracy and recall of search results.

[0026] To improve the accuracy of POI classification, this disclosure provides the following embodiments.

[0027] Figure 1 Based on the schematic diagram of the first embodiment of this disclosure, this embodiment provides a POI classification method, which includes:

[0028] 101. If the first POI does not meet the preset rules and there is no second POI associated with the first POI in the target data, obtain the attribute data of the first POI; wherein, the first POI is a POI to be classified in the map data, and the target data is different from the map data.

[0029] 102. Obtain the POI classification model, which is trained based on the sample POIs in the target data.

[0030] 103. Using the POI classification model, process the attribute data of the first POI to determine the target category of the first POI.

[0031] The target data is different from the map data and can also be called third-party data.

[0032] The target data can specifically be data from a vertical industry, such as data from real estate platforms, wedding photography platforms, or electronics platforms.

[0033] Unlike industries with broad and comprehensive content, vertical industries focus on a specific area, such as IT, entertainment, sports, and real estate. Because they focus on a particular area, the data they provide will be accurate, comprehensive, and rich.

[0034] Taking data from one or more vertical industries as an example, the data from that vertical industry can include Points of Interest (POIs) and their categories. For instance, in wedding photography platform data, a photography studio A might be categorized as "wedding photography" instead of "photo studio" as it is typically found in map data; similarly, in real estate platform data, a residential community X might be categorized as "real estate" instead of "community" as it is typically found in map data.

[0035] To distinguish them, the POI to be classified in the map data can be called the first POI, and the POI associated with the first POI in the target data can be called the second POI.

[0036] The first and second POIs that are related point of interest (POI) refer to the same object. For example, both the first and second POIs refer to a photography studio A or a residential community X.

[0037] The attribute data of the first POI may include the basic data and / or deep data of the POI. The basic data may include, for example, name and coordinates, while the deep data may include, for example, details and comments.

[0038] The input to the POI classification model is the attribute data of the first POI, and the output is the target category of the first POI.

[0039] The POI classification model is pre-trained, and the samples used to train the model include POIs in the target data, which can be called sample POIs.

[0040] In this embodiment, since the sample POI belongs to the target data, the sample POI can be used to train a POI classification model that conforms to the target data classification system. Then, when the first POI is classified using the POI classification model, the target category that conforms to the target data classification system can be obtained, thereby improving the accuracy of POI classification.

[0041] To better understand this disclosure, the application scenarios of the embodiments of this disclosure will be described.

[0042] Figure 2 This is a schematic diagram illustrating an application scenario provided according to embodiments of this disclosure. For example... Figure 2 As shown, map data collectors can use user terminals to collect map data; there can be one or more user terminals. Figure 2 The user terminals in the map are denoted as 201a, 201b to 201n. Map data includes collected data of Points of Interest (POIs), such as names and coordinates. The user terminals send the collected POI data to server 202. Server 202 can obtain relevant data about the POI based on the collected data. This relevant data may include the collected data as well as other data obtained from other sources. The server can classify the POIs based on their relevant data to determine their category. User terminals 201a, 201b to 201n include, for example, personal computers (PCs), mobile devices (such as mobile phones and tablets), laptops, and wearable devices (such as smartwatches). Server 202 can be a local server or a cloud server, and there can be one or more servers. Communication between the user terminals and the server can be via wired and / or wireless networks.

[0043] When classifying Points of Interest (POIs) in map data, you can first classify them according to the classification system of the map product. The resulting categories can be called the original categories. For example, for a certain community X, its original category is "community", and for a certain photography studio A, its original category is "photo studio".

[0044] Correspondingly, the more accurate category obtained after processing can be called the target category.

[0045] The original categories described above are not precise enough. To obtain more accurate target categories, this embodiment will introduce target data that differs from the map data. Target data can also be referred to as third-party data.

[0046] The target data can specifically be data from a vertical industry. For example... Figure 3 As shown, the data for vertical industries includes: real estate platform data, wedding photography platform data, etc.

[0047] The POIs to be classified in the map data can be called the first POI. For the first POI, such as... Figure 3 As shown, there are three ways to obtain the target category of the first POI.

[0048] First, it can be based on preset rules. These rules can be pre-set, recording the mapping relationship between the original category and the target category. Based on this mapping relationship, the original category is mapped to the target category. For example, if the rule records the mapping relationship between "residential community" and "real estate," then if the original category of a POI is "residential community," then based on this mapping relationship, the target category of that POI is set to "real estate."

[0049] Second, it can be determined based on the categories in the target data.

[0050] In this method, we can first identify the second POI in the target data that is associated with the first POI. The first POI and the second POI that are associated mean that the first POI and the second POI point to the same POI, for example, both pointing to the same cell X, or both pointing to the same photography studio A. Figure 3 In Chinese, this is referred to as "POI hook".

[0051] After identifying the second POI in the target data, its category can be obtained from the target data, and this category can be used as the target category for the first POI. Since the target data can also be referred to as third-party data, Figure 3 The term "third-party classification equalization" is used in this context.

[0052] For example, if a wedding photography platform classifies a certain photography studio A as "wedding photography," then the target category for that studio A in the map data will also be determined as "wedding photography," instead of the original category "photo studio."

[0053] Third, it can be determined based on the POI classification model.

[0054] For POIs that can be linked, that is, POIs that exist in both map data and vertical industry data, the second method mentioned above can be used to determine the target category.

[0055] However, because there are many POIs in the map data, but few in the vertical industry data, there will be many POIs in the map data that cannot be linked. Unlinkable POIs refer to POIs that exist in the map data but do not exist in the vertical industry data.

[0056] For POIs that cannot be attached, a POI classification model can be used to determine the target category of the POI.

[0057] like Figure 3 As shown, the input of the POI classification model (which can be simply referred to as the classification model) is the attribute data of the POI. The classification model processes the attribute data of the POI and the output is the target category of the POI.

[0058] Attribute data can include basic POI data (such as name, coordinates, etc.), as well as POI detail data, comment data, etc.

[0059] The classification model is trained based on samples, which can come from the target data, user click data, and / or manually labeled data.

[0060] Samples from target data can be called sample POIs. Sample POIs refer to POIs in target data that are related to POIs in map data.

[0061] For a sample POI, the target data can record the attribute data and its category of the sample POI. Then, the category can be used as the true category, and the <attribute data, true category> of the sample POI can be combined into a set of training data.

[0062] User click data can come from click logs. For example, after displaying search results based on search terms, the click log records the search results clicked by the user, thus generalizing the search terms. For instance, if the search term is "4S" and the user clicks on search results for "car dealership" and "4S store," then "car dealership" and "4S store" become the generalized search terms. Based on the generalized search terms, new category labels can be generated, such as using "car dealership" and "4S store" as new category labels.

[0063] Manually labeled data refers to the collection of Points of Interest (POIs) by humans, and the manual labeling of those POIs with category labels.

[0064] In conjunction with the above application scenarios, this disclosure also provides a training method for a POI classification model.

[0065] Figure 4 Based on the schematic diagram of the second embodiment of this disclosure, this embodiment provides a training method for a POI classification model, the method comprising:

[0066] 401. Obtain training data, wherein the training data includes: attribute data of sample POIs and the true category of the sample POIs, wherein the sample POIs are POIs in target data that are associated with POIs in map data, and the true category is the category of the sample POIs in the target data, wherein the target data is different from the map data.

[0067] 402. Using a POI classification model, the attribute data of the sample POI is processed to determine the predicted category of the sample POI.

[0068] 403. Construct a loss function based on the predicted category and the true category.

[0069] 404. Adjust the model parameters of the POI classification model based on the loss function until the preset conditions are met.

[0070] Taking data from a vertical industry as an example, since there are more POIs in the map data than in the vertical industry data, there will be many POIs in the map data that cannot be attached. Unattached POIs refer to POIs that exist in the map data but do not exist in the vertical industry data.

[0071] For POIs that cannot be attached, a POI classification model can be used to determine the target category of these POIs that cannot be attached (i.e., POIs to be classified).

[0072] In the data of vertical industries, there are also POIs that can be linked, that is, POIs that are related to POIs in map data. These POIs that can be linked can be used as sample POIs to train a POI classification model.

[0073] Understandably, the sample may also include click data and labeled data.

[0074] When training a POI classification model using sample POIs, the attribute data of the sample POIs can be used as the model input, and the model output is the predicted category.

[0075] The category of the POI in the target data is taken as the true category. A loss function is constructed based on the predicted category and the true category. The specific form of the loss function can be set, such as the cross-entropy function.

[0076] After constructing the loss function, the model parameters can be adjusted using the loss function until the preset conditions are met, resulting in the final POI classification model. The final POI classification model can be used in the inference stage to determine the target category of the POI to be classified in the map data.

[0077] The adjusted parameters can be calculated as: Adjusted parameters = Unadjusted parameters - Learning rate * Gradient value, where the gradient value is obtained by differentiating the loss function with respect to the model parameters, and the learning rate is a preset hyperparameter. The initial values ​​for each parameter can be set; for example, they can be randomized to obtain random initial values.

[0078] Preset conditions include, for example, reaching a preset number of iterations or meeting convergence conditions (such as the absolute value of the difference between the loss function of the previous round and the loss function of the current round being less than a preset value). Models that meet the preset conditions can be used as the final POI classification models.

[0079] In this embodiment, a POI classification model is trained based on sample POIs. Since the sample POIs are POIs in the target data, which are different from the map data, a model that conforms to the classification system of the target data can be obtained. When this model is used to classify the POIs to be classified in the map data, information from the target data that is different from the map data can be introduced to improve the accuracy of classifying POIs in the map data.

[0080] Furthermore, the POI classification model may include a feature extraction network and a classification network. The process of using the POI classification model to process the attribute data of the sample POI to determine the predicted category of the sample POI may include:

[0081] The feature extraction network is used to process the attribute data of the sample POI to obtain the feature vector of the sample POI;

[0082] The classification network is used to process the feature vector of the sample POI to determine the predicted category of the sample POI.

[0083] The feature extraction network can be the backbone network of a deep neural network, such as a convolutional neural network (CNN) or a recurrent neural network (RNN).

[0084] Classified networks include, for example, fully connected (FC) networks.

[0085] Feature extraction networks can convert attribute data into feature vectors, and classification networks can obtain predicted categories based on these feature vectors.

[0086] When adjusting model parameters based on the loss function, specifically the model parameters of the feature extraction network and the classification network can be adjusted.

[0087] In this embodiment, by setting the classification model to include a feature extraction network and a classification network, suitable feature extraction networks and classification networks can be trained, thereby extracting suitable feature vectors and improving the accuracy of the target category.

[0088] In conjunction with the above application scenarios, this disclosure also provides a POI classification method.

[0089] Figure 5 Based on a schematic diagram of the third embodiment of this disclosure, this embodiment provides a POI classification method, which includes:

[0090] 501. Traverse the first POI, which is the POI to be classified in the map data.

[0091] There can be one or more first POIs, and each first POI can be traversed sequentially to determine the target category of each POI.

[0092] 502. Determine whether the first POI meets the preset rules. If yes, execute 503; otherwise, execute 504.

[0093] Among these, the original category of the first POI can be obtained. The original category refers to the category determined according to the classification system of the map product.

[0094] The preset rules can record the mapping relationship between the original category and the target category. If the mapping relationship recorded in the preset rules does not contain the original category of the first POI, then it is determined that the first POI does not meet the preset rules. Conversely, if the mapping relationship recorded in the preset rules contains the original category of the first POI, then it is determined that the first POI meets the preset rules.

[0095] 503. Based on the preset rules, determine the target category of the first POI.

[0096] Among them, the original category of the first POI can be mapped to the corresponding target category based on the mapping relationship recorded in the preset rules.

[0097] For example, if the mapping relationship records "residential community" as "real estate", then if the original category of the first POI is "residential community", then the target category of the first POI is determined to be "real estate".

[0098] In this embodiment, the mapping relationship between the original category and the target category is recorded in the rules, and the target category of the first POI can be obtained efficiently based on this mapping relationship.

[0099] 504. Determine whether there is a second POI associated with the first POI in the target data. If yes, proceed to 505; otherwise, proceed to 506.

[0100] Specifically, the similarity between the identification information of the first POI and the identification information of each POI in the target data can be calculated. If there is a POI in the target data with a similarity greater than a preset value, then it is determined that there is a second POI in the target data, and the POI with a similarity greater than the preset value is taken as the second POI.

[0101] For example, if the target data includes POI-1 and POI-2, and the similarity between the identification information of POI-1 and the identification information of the first POI is greater than a preset value, then POI-1 will be used as the second POI.

[0102] Furthermore, if there are multiple POIs in the target data with a similarity greater than a preset value, the POI with the highest similarity can be used as the second POI.

[0103] The aforementioned identification information may include: name and / or coordinates.

[0104] Taking names as an example, names can be converted into vectors (e.g., using the word2vec algorithm), and the similarity between two vectors (e.g., cosine similarity) can be calculated as the similarity between the two identifiers mentioned above.

[0105] 505. Determine the target category of the first POI based on the category of the second POI in the target data.

[0106] The category of the second POI in the target data can be used as the target category of the first POI.

[0107] For example, if a wedding photography platform classifies a certain photography studio A as "wedding photography" in its data, then the target category for that photography studio A in the map data will also be determined as "wedding photography" in the map data.

[0108] It is understandable that the category of the second POI in the target data can also be processed in a certain way, such as unifying the category name, and the category after unifying the category name can be used as the target category of the first POI.

[0109] In this embodiment, since the second POI is associated with the first POI and the category in the target data is more accurate, the accuracy of the category of the first POI can be improved by using the category of the second POI in the target data as the target category of the first POI.

[0110] 506. Obtain the attribute data of the first POI and the pre-trained POI classification model.

[0111] The attribute data can include basic POI data (such as name, coordinates, etc.), as well as POI detail data, comment data, etc.

[0112] The POI classification model is trained based on sample POIs in the target data. The training process of the POI classification model can be found in the above-mentioned embodiments, and will not be repeated here.

[0113] 507. Using the POI classification model, the attribute data of the first POI is processed to determine the target category of the first POI.

[0114] The input to the POI classification model is attribute data, and the output is the target category.

[0115] Furthermore, the POI classification model can include: a feature extraction network and a classification network;

[0116] The step of using a pre-trained POI classification model to process the attribute data of the first POI to determine the target category of the first POI includes:

[0117] The feature extraction network is used to process the attribute data of the first POI to obtain the feature vector of the first POI.

[0118] The classification network is used to process the feature vector of the first POI to determine the target category of the first POI.

[0119] In this embodiment, by setting the classification model to include a feature extraction network and a classification network, suitable feature extraction networks and classification networks can be trained, thereby extracting suitable feature vectors and improving the accuracy of the target category.

[0120] In this embodiment, when the first POI meets the preset rules, the target category of the first POI is determined based on the preset rules. When the first POI does not meet the preset rules but a second POI associated with the first POI exists in the target data, the target category of the first POI is determined based on the category of the second POI in the target data. When the first POI does not meet the preset rules and no second POI associated with the first POI exists in the target data, a POI classification model is used to determine the target category of the first POI. This allows for the selection of an appropriate method for determining the target category in different situations, thereby obtaining a more accurate target category. Furthermore, by introducing target data, the precise, comprehensive, and rich information of the target data can be utilized to improve the classification effect for POIs to be classified in the map data.

[0121] Figure 6 This is a schematic diagram based on the fourth embodiment of the present disclosure. This embodiment provides a POI classification device 600, which includes: a first acquisition module 601, a second acquisition module 602, and a first determination module 603.

[0122] The first acquisition module 601 is used to acquire the attribute data of the first POI if the first POI does not meet the preset rules and there is no second POI associated with the first POI in the target data; wherein, the first POI is a POI to be classified in the map data, and the target data is different from the map data; the second acquisition module 602 is used to acquire a POI classification model, which is trained based on sample POIs in the target data; the first determination module 603 is used to process the attribute data of the first POI using the POI classification model to determine the target category of the first POI.

[0123] In this embodiment, since the sample POI belongs to the target data, the sample POI can be used to train a POI classification model that conforms to the target data classification system. Then, when the first POI is classified using the POI classification model, the target category that conforms to the target data classification system can be obtained, thereby improving the accuracy of POI classification.

[0124] In some embodiments, the POI classification model includes: a feature extraction network and a classification network;

[0125] The first determining module 603 is further configured to: use the feature extraction network to process the attribute data of the first POI to obtain the feature vector of the first POI; and use the classification network to process the feature vector of the first POI to determine the target category of the first POI.

[0126] In this embodiment, by setting the classification model to include a feature extraction network and a classification network, suitable feature extraction networks and classification networks can be trained, thereby extracting suitable feature vectors and improving the accuracy of the target category.

[0127] In some embodiments, the device 600 further includes a second determining module, configured to determine the category of the first POI based on the category of the second POI in the target data if the first POI does not meet a preset rule and there is a second POI associated with the first POI in the target data.

[0128] In this embodiment, when the first POI does not meet the preset rules but there is a second POI associated with the first POI in the target data, the target category of the first POI is determined based on the category of the second POI in the target data. The appropriate method for determining the target category can be selected in different situations, thereby obtaining a more accurate target category.

[0129] In some embodiments, the second determining module is further configured to:

[0130] The category of the second POI in the target data is taken as the category of the first POI.

[0131] In this embodiment, since the second POI is associated with the first POI and the category in the target data is more accurate, the accuracy of the category of the first POI can be improved by using the category of the second POI in the target data as the target category of the first POI.

[0132] In some embodiments, the device 600 further includes a third determining module, configured to determine the target category of the first POI based on the preset rules if the first POI satisfies the preset rules.

[0133] In this embodiment, when the first POI meets the preset rules, the target category of the first POI is determined based on the preset rules. The appropriate method for determining the target category can be selected under different circumstances, thereby obtaining a more accurate target category.

[0134] In some embodiments, the preset rules record the mapping relationship between the original category and the target category; the third determining module is further configured to: if the mapping relationship includes the original category of the first POI, determine that the first POI satisfies the preset rules, and map the original category of the first POI to the corresponding target category based on the mapping relationship.

[0135] In this embodiment, the mapping relationship between the original category and the target category is recorded in the rules, and the target category of the first POI can be obtained efficiently based on this mapping relationship.

[0136] Figure 7 This is a schematic diagram based on the fifth embodiment of the present disclosure. This embodiment provides a training device for a POI classification model. The device 700 includes: an acquisition module 701, a prediction module 702, a construction module 703, and an adjustment module 704.

[0137] The acquisition module 701 is used to acquire training data, which includes: attribute data of sample POIs and the true category of the sample POIs. The sample POIs are POIs in target data that are associated with POIs in map data, and the true category is the category of the sample POI in the target data, which is different from the map data. The prediction module 702 is used to process the attribute data of the sample POIs using a POI classification model to determine the predicted category of the sample POIs. The construction module 703 is used to construct a loss function based on the predicted category and the true category. The adjustment module 704 is used to adjust the model parameters of the POI classification model based on the loss function until a preset condition is met to obtain the final POI classification model. The final POI classification model is used to determine the target category of the POI to be classified in the map data.

[0138] In this embodiment, a POI classification model is trained based on sample POIs. Since the sample POIs are POIs in the target data, which are different from the map data, a model that conforms to the classification system of the target data can be obtained. When this model is used to classify the POIs to be classified in the map data, information from the target data that is different from the map data can be introduced to improve the accuracy of classifying POIs in the map data.

[0139] In some embodiments, the POI classification model includes: a feature extraction network and a classification network;

[0140] The prediction module 703 is further used for:

[0141] The feature extraction network is used to process the attribute data of the sample POI to obtain the feature vector of the sample POI;

[0142] The classification network is used to process the feature vector of the sample POI to determine the predicted category of the sample POI.

[0143] In this embodiment, by setting the classification model to include a feature extraction network and a classification network, suitable feature extraction networks and classification networks can be trained, thereby extracting suitable feature vectors and improving the accuracy of the target category.

[0144] It is understood that the same or similar content in different embodiments of this disclosure can be referred to each other.

[0145] It is understood that the terms "first" and "second" in the embodiments of this disclosure are only used for distinction and do not indicate the degree of importance or the order of events.

[0146] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0147] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0148] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement embodiments of the present disclosure is shown. Electronic device 800 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 800 may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0149] like Figure 8 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the electronic device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0150] Multiple components in electronic device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of displays, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows electronic device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0151] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as POI classification methods or POI classification model training methods. For example, in some embodiments, the POI classification method or POI classification model training method can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the POI classification method or POI classification model training method described above can be performed. Alternatively, in other embodiments, the computing unit 801 may be configured by any other suitable means (e.g., by means of firmware) to perform a POI classification method or a training method for a POI classification model.

[0152] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0153] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to the processor or controller of a general-purpose computer, special-purpose computer, or other programmable load balancing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0154] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0155] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0156] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0157] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service ecosystem, addressing the shortcomings of traditional physical hosts and VPS (Virtual Private Server, or simply "VPS") services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0158] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0159] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for classifying Points of Interest (POIs), comprising: Iterate through the first POI, which is the POI to be classified in the map data; If the first POI satisfies the preset rules, the original category of the first POI is mapped to the corresponding target category based on the mapping relationship recorded in the preset rules. The original category is obtained after classification according to the classification system of the map product. If the first POI does not meet the preset rules, and there is a second POI associated with the first POI in the target data, the category of the second POI in the target data is taken as the target category of the first POI; the target data is different from the map data and is third-party data from a vertical industry. If the first POI does not meet the preset rules, and there is no second POI associated with the first POI in the target data, obtain the attribute data of the first POI; A POI classification model is obtained, which is trained based on sample POIs in the target data and user click data; the sample POIs are correlated with POIs in the map data; and the user click data is used as a new category label. The attribute data of the first POI is processed using the POI classification model to determine the target category of the first POI.

2. The method according to claim 1, wherein, The POI classification model includes: a feature extraction network and a classification network; A pre-trained POI classification model is used to process the attribute data of the first POI to determine the target category of the first POI, including: The feature extraction network is used to process the attribute data of the first POI to obtain the feature vector of the first POI. The classification network is used to process the feature vector of the first POI to determine the target category of the first POI.

3. A training method for a POI classification model, comprising: Acquire training data, which includes: attribute data of sample POIs and the true category of the sample POIs. The sample POIs are POIs in target data that are related to POIs in map data. The true category is the category of the sample POI in the target data. The target data is different from the map data and is third-party data from a vertical industry. The training data also includes: user click data, which serves as a new category label. A POI classification model is used to process the attribute data of the sample POIs in order to determine the predicted category of the sample POIs. Based on the predicted category and the true category, a loss function is constructed; The model parameters of the POI classification model are adjusted based on the loss function until the preset conditions are met to obtain the final POI classification model. The final POI classification model is used to determine the target category of the POI to be classified in the map data when the first POI does not meet the preset rules and there is no second POI associated with the first POI in the target data. The first POI is a POI to be classified in the map data. When a preset rule is met, the original category of the first POI is mapped to the corresponding target category based on the mapping relationship recorded in the preset rule. The original category is obtained after classification according to the classification system of the map product. Alternatively, if the preset rule is not met, and there is a second POI associated with the first POI in the target data, the category of the second POI in the target data is taken as the target category of the first POI. The target data is different from the map data and is third-party data from a vertical industry.

4. The method according to claim 3, wherein, The POI classification model includes: a feature extraction network and a classification network; The step of using a POI classification model to process the attribute data of the sample POIs to determine the predicted category of the sample POIs includes: The feature extraction network is used to process the attribute data of the sample POI to obtain the feature vector of the sample POI; The classification network is used to process the feature vector of the sample POI to determine the predicted category of the sample POI.

5. A POI classification device, comprising: The first acquisition module is used to acquire the attribute data of the first POI if the first POI does not meet the preset rules and there is no second POI associated with the first POI in the target data; wherein the first POI is a POI to be classified in the map data, and the target data is different from the map data and is third-party data of a vertical industry. The second acquisition module is used to acquire a POI classification model, which is trained based on sample POIs in the target data; the sample POIs are correlated with the POIs in the map data; The first determining module is used to process the attribute data of the first POI using the POI classification model to determine the target category of the first POI. The second determining module is used to determine the category of the second POI in the target data as the target category of the first POI if the first POI does not meet the preset rules and there is a second POI associated with the first POI in the target data. The third determining module is used to map the original category of the first POI to the corresponding target category based on the mapping relationship recorded in the preset rules if the first POI meets the preset rules. The original category is obtained after classification according to the classification system of the map product.

6. The apparatus according to claim 5, wherein, The POI classification model includes: a feature extraction network and a classification network; The first determining module is further configured to: The feature extraction network is used to process the attribute data of the first POI to obtain the feature vector of the first POI. The classification network is used to process the feature vector of the first POI to determine the target category of the first POI.

7. A training device for a POI classification model, comprising: The acquisition module is used to acquire training data, which includes: attribute data of sample POIs and the true category of the sample POIs. The sample POIs are POIs in target data that are related to POIs in map data. The true category is the category of the sample POI in the target data. The target data is different from the map data and is third-party data from a vertical industry. The training data also includes: user click data, which serves as a new category label. The prediction module is used to process the attribute data of the sample POI using a POI classification model to determine the predicted category of the sample POI. The construction module is used to construct a loss function based on the predicted category and the true category; The adjustment module is used to adjust the model parameters of the POI classification model based on the loss function until the preset conditions are met to obtain the final POI classification model. The final POI classification model is used to determine the target category of the POI to be classified in the map data when the first POI does not meet the preset rules and there is no second POI associated with the first POI in the target data. The first POI is a POI to be classified in the map data. When a preset rule is met, the original category of the first POI is mapped to the corresponding target category based on the mapping relationship recorded in the preset rule. The original category is obtained after classification according to the classification system of the map product. Alternatively, if the preset rule is not met, and there is a second POI associated with the first POI in the target data, the category of the second POI in the target data is taken as the target category of the first POI. The target data is different from the map data and is third-party data from a vertical industry.

8. The apparatus according to claim 7, wherein, The POI classification model includes: a feature extraction network and a classification network; The prediction module is further used for: The feature extraction network is used to process the attribute data of the sample POI to obtain the feature vector of the sample POI; The classification network is used to process the feature vector of the sample POI to determine the predicted category of the sample POI.

9. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-4.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-4.

11. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-4.

Images