Poi processing method and apparatus, electronic device, and storage medium

Abstract

The present disclosure provides a POI processing method and device, electronic equipment and storage medium, relates to the field of data processing, specifically to the field of intelligent search, artificial intelligence and deep learning. The specific implementation scheme is: obtaining a key feature of a POI to be detected; comparing the key feature with features in a brand feature library to obtain a brand feature matched with the POI to be detected; and determining a brand word corresponding to the matched brand feature as a brand word corresponding to the POI to be detected. The present disclosure can improve the accuracy of POI brand word matching.

Description

Technical Field

[0001] This disclosure relates to the field of data processing, particularly to the fields of intelligent search, artificial intelligence and deep learning, and specifically to a POI processing method, apparatus, electronic device and storage medium. Background Technology

[0002] Location-based services (LBS) are a hot topic in mobile services.

[0003] Points of Interest (POIs) are core data for location-based services. POIs are widely used in applications. Users can find the POIs they need by searching for keywords within applications. Summary of the Invention

[0004] This disclosure provides a POI processing method, apparatus, electronic device, and storage medium.

[0005] According to one aspect of this disclosure, a POI processing method is provided, comprising:

[0006] Obtain the key features of the POI to be detected;

[0007] The key features are compared with the features in the brand feature library to obtain the brand features that match the POI to be detected;

[0008] The brand words corresponding to the matched brand features are identified as the brand words corresponding to the POI to be detected.

[0009] According to one aspect of this disclosure, a POI processing apparatus is provided, comprising:

[0010] The key feature acquisition module is used to acquire the key features of the POI to be detected.

[0011] The brand feature acquisition module is used to compare key features with features in the brand feature library to obtain brand features that match the POI to be detected.

[0012] The brand term identification module is used to identify the brand terms corresponding to the matched brand features as the brand terms corresponding to the POI to be detected.

[0013] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the POI processing method according to any embodiment of this disclosure.

[0017] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform the POI processing method described in any embodiment of this disclosure.

[0018] According to another aspect of this disclosure, a computer program object is provided, including a computer program that, when executed by a processor, implements the POI processing method described in any embodiment of this disclosure.

[0019] The embodiments disclosed herein can improve the accuracy of POI brand keyword matching.

[0020] 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

[0021] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0022] Figure 1 This is a flowchart of a POI processing method disclosed in an embodiment of this disclosure;

[0023] Figure 2 This is a flowchart of another POI processing method disclosed according to an embodiment of this disclosure;

[0024] Figure 3 This is a flowchart of another POI processing method disclosed according to an embodiment of this disclosure;

[0025] Figure 4 This is a scene diagram of another POI processing method disclosed in the embodiments of this disclosure;

[0026] Figure 5 This is a structural diagram of a POI processing apparatus disclosed in an embodiment of the present disclosure;

[0027] Figure 6 This is a block diagram of an electronic device used to implement the POI processing method of the embodiments of this disclosure. Detailed Implementation

[0028] 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.

[0029] Figure 1 This is a flowchart of a POI processing method disclosed in an embodiment of this disclosure. This embodiment can be applied to situations involving POI processing. The method of this embodiment can be executed by a POI processing device, which can be implemented in software and / or hardware and specifically configured in an electronic device with certain data processing capabilities. The electronic device can be a client device or a server device, such as a mobile phone, tablet computer, vehicle terminal, or desktop computer.

[0030] S101. Obtain the key features of the POI to be detected.

[0031] Key features can be features corresponding to brand-related information within the information of the POI to be detected. Key features describe the brand-related information within the POI to be detected. Key features are typically represented in vector or matrix form. Additionally, key features can also be represented using characters. The information of the POI to be detected can be information associated with the POI. Optionally, the information of the POI to be detected can include: the POI's identification information, location information, contact information, business information, and environmental images, etc. The brand-related information can be information about suspected brand terms within the POI to be detected. The identification information of the POI to be detected usually contains brand-related information. Specifically, brand-related information can be extracted from the identification information of the POI to be detected, and key features can be determined based on this brand-related information.

[0032] Optionally, key features can be extracted directly from the information of the POI to be detected using a pre-trained key feature extraction model. The key feature extraction model is used to extract the key features of the POI to be detected. The output of the key feature extraction model can be the key features of the POI to be detected. Optionally, the input to the key feature extraction model can be the information of the POI to be detected. Key feature extraction models can include: BERT (Bidirectional Encoder Representations from Transformers) model, Word2Vec (word vector) model, ALBERT (Lightweight Bidirectional Encoder Representations) model, or XLNet (out-of-order language) model, etc.

[0033] Optionally, a pre-trained key feature extraction model can be used to extract key features from partial information of the POI to be detected, thus obtaining the key features of the POI. The partial information of the POI to be detected may include a combination of at least one of the following: the POI's identification information and location information, contact information, business information, or environmental images. Optionally, the input to the key feature extraction model can be partial information of the POI to be detected.

[0034] For example, suppose the information of the POI to be detected includes: Branch 1 of X Brand Solar Water Heater in City Q, located at No. Q3, Q2 Street, Q1 District, City Q, business hours "8:00-17:00", and environmental images 1, 2, and 3. Partial information of the POI to be detected includes: Branch 1 of X Brand Solar Water Heater in City Q and No. Q3, Q2 Street, Q1 District, City Q. The key feature extraction model is the BERT model. Information or partial information of the POI to be detected can be directly input into the BERT model. The BERT model extracts key features to obtain the key features of the POI to be detected, such as the features corresponding to the X brand.

[0035] S102. Compare the key features with the features in the brand feature library to obtain the brand features that match the POI to be detected.

[0036] Brand features are used to describe brand terms. Brand features can typically be represented in vector or matrix form. Additionally, brand features can also be represented using characters. Brand features can be features that match the key features of the POI to be detected. The brand feature library contains a large number of features corresponding to brand terms. Accordingly, the brand feature library includes a large number of features. Optionally, features in the brand feature library can be obtained by extracting features from pre-collected brand terms using a pre-trained brand feature extraction model. The brand feature extraction model is used to extract features from brand terms. The input to the brand feature extraction model can be brand terms, and the output of the brand feature extraction model is the features of those brand terms. For example, the brand feature extraction model can be: BERT model, Word2Vec model, ALBERT model, or XLNet model. Brand terms can be obtained by web scraping. Optionally, brand terms can be extracted from the brand website's logo image, brand website name, or brand company name; brand terms can also be extracted from the website's store sign images and registration information images. The brand website name can be the brand flagship store or brand self-operated store, etc. The brand company name can be the company name corresponding to the brand as stated on the brand website. After the brand feature extraction model extracts features, the obtained brand word features can be added to the brand feature library.

[0037] Brand features that match the POI to be detected refer to features in the brand feature library that are the same as or similar to the key features of the POI to be detected. The key features are compared with features in the brand feature library to identify the features in the library that are the same as or most similar to the key features.

[0038] Optionally, the key features can be compared with features in the brand feature library, and the feature with the most identical characters in the key features can be used as the brand feature that matches the POI to be detected.

[0039] Optionally, key features and features from a brand feature library can be input into a pre-trained brand feature comparison model. The brand feature comparison model compares the features in the brand feature library with the key features and outputs the feature closest to the key feature as the brand feature matching the POI to be detected. Specifically, the brand feature comparison model is used to identify the text feature in the brand feature library that is closest to the key feature. The input to the brand feature comparison model is the key feature and features from the brand feature library, and the output is the brand feature closest to the key feature. For example, the brand feature comparison model can include: DSSM (Deep Structured Semantic Models), ESIM (Enhanced Sequential Inference Model), ABCNN (Attention-Based Convolutional Neural Network), BiMPM (Bilateral Multi-Perspective Matching), DIIN (Densely Interactive Inference Network), or DRCN (Dual Correlation Reduction Network), etc.

[0040] S103. The brand words corresponding to the matched brand features are identified as the brand words corresponding to the POI to be detected.

[0041] When storing the features corresponding to brand words in the brand feature database, the correspondence between these features and brand words can be recorded. Based on this pre-established correspondence, the brand words corresponding to the matching brand features can be determined. These brand words are then identified as the brand words corresponding to the POI to be detected.

[0042] After identifying the brand keywords corresponding to the POI to be detected, these brand keywords can be added as attribute information of the POI. This attribute information allows for searching, querying, and displaying the POI. Furthermore, the brand keywords can also be displayed as attribute information within the POI's overall information.

[0043] In existing technologies, the method for detecting brand words matching a POI involves comparing the text string of the POI with the text string of the brand word, and identifying the common part of the text strings as the brand word that matches the POI. However, this method of detecting the common part of the text strings is inefficient, and the common part may contain other invalid information besides the brand word, resulting in poor detection accuracy.

[0044] According to the technical solution of this disclosure, by acquiring the key features of the POI to be detected and comparing the key features with features in the brand feature library, brand features matching the POI to be detected are obtained. By obtaining the brand feature library in advance, the efficiency of screening brand features matching the POI to be detected is improved. In the process of determining the brand features matching the POI to be detected, the key features of the POI to be detected are compared with the features in the brand feature library, which reduces the interference of invalid information and improves the accuracy of brand feature screening. At the same time, by comparing features, the accuracy of brand feature screening is further improved compared with the comparison of text strings. The brand words corresponding to the matched brand features are determined as the brand words corresponding to the POI to be detected, which improves the accuracy of the brand words matched by the POI to be detected.

[0045] Figure 2 This is a flowchart of another POI processing method disclosed in this disclosure, which is further optimized and extended based on the above technical solution and can be combined with the above optional implementation methods. Obtaining key features of the POI to be detected includes: obtaining the identification information of the POI to be detected; performing component detection on the identification information of the POI to be detected and extracting the main component information; and performing feature extraction on the main component information to obtain key features.

[0046] S201. Obtain the identification information of the POI to be detected.

[0047] Identification information is used to identify a Point of Interest (POI). Typically, identification information includes information related to the brand name. Optionally, identification information may include: the POI's name and / or POI's aliases, etc. The POI's name can be the full name of the thing the POI represents. Alternatively, the POI's name can be a certified name for the thing the POI represents. The POI's alias can be any name other than the POI's name. Optionally, the POI's alias may include: an abbreviation of the POI's name, a shortened version of the POI's name, common alternatives to the POI's name, popular online terms related to the POI's name, a combination of the aforementioned information with other information identifying the POI, or a combination of the POI's name with other information, etc. The POI's name is the registered name of a company. The POI's alias can be an abbreviation of the registered name. For example, the POI's name could be XX Limited Liability Company. The POI's aliases could include: XX, YY, QT, XX Company Store 1, or QT Branch, etc.

[0048] Specifically, the name and / or alias of the POI to be detected can be obtained from the information of the POI to be detected to determine the identification information of the POI to be detected.

[0049] S202. Perform component detection on the identification information of the POI to be detected and extract the main component information.

[0050] Components are used to segment identification information, describing the content that distinguishes the segmented information from other parts. For example, identification information can be segmented according to attribute type to obtain at least one component. For instance, POI identification information may include: information related to brand information, location information, time information, functional information, or branch information, etc. Core component information describes the information among the components that best represents the identification information. For instance, core component information may be the component information related to brand information within the identification information.

[0051] Specifically, keyword extraction algorithms or models can be used to detect the components of the identifier information of the POI to be detected, and then the identifier information of the POI to be detected can be split according to different components. After splitting, the information that best represents the identifier information is obtained as the backbone component information. For example, keyword extraction algorithms can include Text Rank algorithms, etc. Keyword extraction models can include ERNIE (Enhanced Representation through Knowledge Integration) models, etc.

[0052] For example, the Text Rank algorithm or ERNIE model can be used to perform component detection on the identification information of the POI to be detected, and the identification information of the POI to be detected can be split according to different components (e.g., information related to brand information, location information, time information, functional information or branch information, etc.). After splitting, the information that best represents the identification information (e.g., information related to brand information) is obtained as the backbone component information.

[0053] S203. Extract features from the main component information to obtain key features.

[0054] Specifically, the core component information can be input into the key feature extraction model, which will then extract the key features.

[0055] For example, assuming the key feature extraction model is the BERT model, the backbone component information can be input into the BERT model, and the BERT model will extract the key features to obtain the feature vectors corresponding to the key features.

[0056] S204. Compare the key features with the features in the brand feature library to obtain the brand features that match the POI to be detected.

[0057] S205. The brand words corresponding to the matched brand features are identified as the brand words corresponding to the POI to be detected.

[0058] In an optional embodiment of this disclosure, before obtaining the key features of the POI to be detected, the method further includes: obtaining online POIs; determining brand words based on the identification information of each online POI; extracting features from the brand words and adding them to the brand feature library, wherein the feature extraction method of the brand words is the same as the feature extraction method of the main component information.

[0059] Step A involves obtaining the online Points of Interest (POIs).

[0060] "Online POIs" refers to POIs that have already been launched and put into use. All online POIs have been verified by this platform or a third-party platform. The identification information of online POIs refers to the identification information of POIs that is authentic and reliable, meaning that the identification information of the POI can be considered correct.

[0061] Specifically, you can obtain listed Points of Interest (POIs) through this platform or third-party platforms. The acquisition of listed POIs is authorized and confirmed by this platform or third-party platforms, and complies with legal regulations and does not violate public order and good morals.

[0062] Step B: Determine the brand keywords based on the identification information of each online POI.

[0063] Optionally, the common parts of the identification information of a large number of online POIs can be identified as brand keywords.

[0064] Optionally, keyword extraction algorithms or models can be used to detect the components of the identification information of each online POI, and the identification information of each online POI can be split according to different components. The information that best represents the identification information can be obtained after splitting and used as brand words.

[0065] Optionally, the identifier information of the online POIs can be split into a large amount of text information. A brand term extraction algorithm can then be used to extract a preset number of text messages with high word frequencies. These preset number of text messages with high word frequencies are then identified as brand terms. The brand term extraction algorithm is used to extract the text messages with high word frequencies from the identifier information of the online POIs as brand terms. For example, the brand term extraction algorithm may include a Term Frequency-Inverse Document Frequency (TF-IDF) extraction algorithm or a Latent Dirichlet Allocation (LDA) extraction algorithm. The preset number of outputs can be set and adjusted based on the experience of technical personnel.

[0066] Step C: Extract features from brand words and add them to the brand feature library. The feature extraction method for brand words is the same as that for the main component information.

[0067] The feature extraction method for brand words is the same as that for the main component information, which avoids interference from feature comparison caused by different feature extraction methods and further improves the accuracy of brand features obtained through feature comparison.

[0068] Specifically, brand words can be feature extracted using the same method as that used for extracting core component information, and the extracted brand features can be added to the brand feature library. These feature extraction methods can include BERT, Word2Vec, ALBERT, or XLNet models.

[0069] By acquiring online Points of Interest (POIs), determining brand terms based on their identifiers, extracting features from these brand terms, and adding them to a brand feature library, the process of determining brand terms based on the identifiers of online POIs—that is, utilizing more authentic and reliable identifiers—improves the authenticity and reliability of brand terms. By employing the same feature extraction method as for the core component information, the influence of different feature extraction methods on the acquired key features and features in the brand feature library is avoided. This improves the accuracy of brand features determined by comparing key features with features in the brand feature library, thereby enhancing the accuracy of brand term matching for the detected POIs.

[0070] In an optional embodiment of this disclosure, the brand term is determined based on the identification information of each online POI, specifically as follows: clustering the online POIs to obtain candidate classes; determining the public information corresponding to the candidate class from the identification information of the POIs in the same candidate class; performing brand verification on the public information; performing image clustering according to the signs of the POIs in the target class that have passed brand verification; and detecting whether the public information of the target class is a brand term based on the image clustering results.

[0071] Step B1: Cluster the online POIs to obtain candidate classes.

[0072] The alternative classes can be those obtained by clustering already online POIs, including at least one already online POI.

[0073] Specifically, text clustering algorithms can be used to cluster a large number of online POIs, resulting in high similarity among online POIs in the same candidate cluster and low similarity among online POIs in different candidate clusters, thus generating multiple candidate clusters. Optionally, text clustering algorithms can include: K-means, BIRCH (Balanced Iterative Reduction and Clustering Using Hierarchies), GMM (Gaussian mixture model), or GAAC (Group-average Agglomerative Clustering), etc.

[0074] Step B2: Determine the public information corresponding to the candidate class from the identification information of POIs in the same candidate class.

[0075] Public information can be the same or similar information among the identification information of various POIs in the same candidate class. Specifically, the identification information of at least two POIs in the same candidate class can be compared to determine the same or similar information among the at least two identification information, which is then identified as the public information corresponding to the candidate class.

[0076] For example, the identification information of a POI in the same candidate class may include three pieces of identification information: H (J Street Branch), H Limited Liability Company, and H Fried Chicken. By comparing the three pieces of identification information in the same candidate class, the common information among the three pieces of identification information is determined to be H. Therefore, H is the common information corresponding to this candidate class.

[0077] Step B3: Validate the brand information.

[0078] Brand verification can be a method of verifying whether public information is a brand term. Optionally, this can be done by searching the public information to find the brand's official website, or by using an authoritative brand verification platform.

[0079] Optionally, you can verify the public information by searching for the brand's official website. If the brand's official website can be found through the public information search, the verification is successful; otherwise, the verification fails.

[0080] Alternatively, public information can be verified through an authoritative brand verification platform. If the brand verification platform approves the public information, the verification is successful; otherwise, the verification fails.

[0081] Step B4: In the target class, perform image clustering according to the signs of the POIs of the target class.

[0082] The target category can be the category of a POI that has passed brand verification. The POI sign can be an image of a sign containing textual content such as the brand name or brand business information corresponding to the POI. POI signs can be obtained from official websites retrieved through brand verification, from map data, or through online searches.

[0083] Specifically, within the target class that has passed brand verification, image clustering algorithms can be used to cluster the signs of at least two POIs in the target class, yielding image clustering results. Optionally, image clustering algorithms can include: K-means algorithm, GAAC (Group-average Agglomerative Clustering), or DBSCAN (Density-Based Spatial Clustering of Applications with Noise), etc.

[0084] Step B5: Based on the image clustering results, detect whether the public information of the target class is a brand term.

[0085] Image clustering results may include at least one of the following: at least one cluster obtained from image clustering, and the POIs included in each image cluster. Furthermore, the image clustering results may also include the number of POIs included in each image cluster and the number of image clusters. The POIs included in the target cluster are the data to be image clustered. The image cluster can be the various categories obtained after image clustering of the signs of the POIs in the target cluster. Specifically, each POI in the target cluster provides only one sign. The number of signs of the POIs included in each image cluster is the same as the number of POIs included in each image cluster.

[0086] Specifically, based on the image clustering results and the number of POIs contained in each image cluster, the percentage of POIs in the most frequent image cluster relative to the total number of POIs in the target cluster can be calculated. If this percentage is greater than a preset percentage threshold, the public information of the target cluster is determined to be a brand term; otherwise, it is not. Alternatively, the number of image clusters can be used to determine if it exceeds a preset number threshold. If the number of image clusters is less than or equal to the preset number threshold, the public information of the target cluster is determined to be a brand term; otherwise, it is not.

[0087] For example, assuming there are 2500 online POIs, a preset percentage threshold of 80%, and a preset quantity threshold of 3, a text clustering algorithm can be used to cluster the 2500 online POIs, generating multiple candidate classes, such as two. The first candidate class can contain 1500 online POIs, and the second candidate class can contain 1000 online POIs. The identifier information of the POIs in the first candidate class is compared, and the same information is identified as the public information corresponding to that candidate class, such as "M Brand Store". The identifier information of the POIs in the second candidate class is also compared, and the same information is identified as the public information corresponding to that candidate class, such as "M Brand". Brand verification can be performed by querying the brand's official website. If the brand's official website can be found through "M Brand", the brand verification for the second candidate class corresponding to "M Brand" passes; if the brand's official website cannot be found through "M Brand Store", the brand verification for the first candidate class corresponding to "M Brand Store" fails. The second candidate class after brand verification becomes the target class. An image clustering algorithm is used to cluster the signs of the 1000 POIs in the target class, resulting in three image clusters: the first image cluster contains 850 signs, the second contains 80 signs, and the third contains 70 signs. Optionally, the percentage of the number of POIs in the largest image cluster (the first image cluster, i.e., 850) out of the total number of POIs in the target class (i.e., 1000) can be calculated, which is 85%. If the percentage of the number of POIs in the largest image cluster (i.e., 85%) is greater than a preset percentage threshold (i.e., 80%), the public information of the target class (i.e., brand M) can be identified as the brand term. Alternatively, if the number of image clusters obtained (i.e., 3) equals the preset number threshold (i.e., 3), the public information of the target class (i.e., brand M) is identified as the brand term.

[0088] By clustering the online Points of Interest (POIs), candidate classes are obtained. Within each candidate class, the public information corresponding to each candidate class is determined from its identifier information. Brand verification is performed on the public information corresponding to each candidate class, and target classes that pass brand verification are selected, achieving the first verification of the public information. Further image clustering is performed on the signs of the target class's POIs, and based on the image clustering results, the public information corresponding to the target class is verified again to determine if it is a brand term. The verified public information is then identified as the brand term. Through these two verifications, the obtained brand terms are ensured to pass both brand verification and image clustering result verification, improving the accuracy of the brand terms.

[0089] In an optional embodiment of this disclosure, the identification information includes: the name and alias of the POI to be detected; the key features include: the key features of the name and the key features of the alias; the key features are compared with features in the brand feature library to obtain brand features that match the POI to be detected, specifically: the key features are compared with features in the brand feature library to obtain the comparison results of each key feature; based on the comparison results of each key feature, the brand features that match the POI to be detected are obtained.

[0090] In step S2041, the key features are compared with the features in the brand feature library to obtain the comparison results of each key feature.

[0091] Specifically, the key features of the name corresponding to the POI to be detected and the key features of the alias corresponding to the POI to be detected can be compared with the features in the brand feature library, respectively, to obtain the comparison results of the key features of the name and the comparison results of the key features of the alias.

[0092] S2042. Based on the comparison results of each key feature, obtain the brand features that match the POI to be detected.

[0093] Specifically, the comparison results of the key features of the name and the key features of the alias can be compared again to determine the comparison result of the key features of the name and the key features of the alias. Based on the comparison result of the second comparison, the brand features that match the POI to be detected can be determined.

[0094] In practice, the key features of the name are compared with the features in the brand feature library, and the key features of at least one alias are compared with the features in the brand feature library. The feature in the brand feature library that is most similar to the POI to be detected is selected as the brand feature that matches the POI to be detected.

[0095] For example, the identification information may include: the name U and alias V of the POI to be detected, and the key features include: key features of name U and key features of alias V. The key features of name U can be compared with features in the brand feature library to obtain the comparison results of the key features of name U. For example, the similarity between brand feature O and the key features of name U is 1. Similarly, the key features of alias V can be compared with features in the brand feature library to obtain the comparison results of the key features of alias V. For example, the similarity between brand feature P and the key features of alias V is 2. The comparison results of the key features of name U and alias V can be compared again, i.e., the highest similarity is found between the key features of brand feature P and the key features of alias V. That is, brand feature P is determined to be the brand feature matching the POI to be detected.

[0096] By specifying the identification information as the name and alias of the POI to be detected, and specifying the key features as the key features of the name and the key features of the alias, each key feature is compared with the features in the brand feature library to obtain the comparison results of each key feature. Based on the comparison results of each key feature, the brand features that match the POI to be detected are obtained. This realizes a secondary comparison of the comparison results corresponding to the key features, further improving the accuracy of the matched brand features and ensuring the accuracy of brand word matching.

[0097] According to the technical solution of this disclosure, the identification information of the POI to be detected is obtained; the component detection of the identification information of the POI to be detected is performed to obtain the main component information of different components, and the key features are obtained by feature extraction of the main component information. The information of the extracted key features is further refined into the identification information of the POI to be detected, so as to select the information required for the extraction of key features more accurately; by performing component detection on the identification information of the POI to be detected, the identification information of the POI to be detected is split by different components, which ensures that the main component information contains only information of a single component and avoids the interference of invalid information on the extraction of main component information; by extracting features from the main component information, the key features are obtained, which improves the accuracy of key feature selection, thereby improving the accuracy of brand features determined by key features and improving the accuracy of brand words matched with the POI to be detected.

[0098] Figure 3 This is a flowchart of another POI processing method disclosed in this embodiment, which is further optimized and extended based on the above technical solution and can be combined with the above optional implementation methods. The method compares key features with features in a brand feature library to obtain brand features matching the POI to be detected. Specifically, this involves: calculating the similarity between the key features and features in the brand feature library; and determining the brand features matching the POI to be detected based on the similarity between the key features and features in the brand feature library.

[0099] S301. Obtain the key features of the POI to be detected.

[0100] S302. Calculate the similarity between key features and features in the brand feature library.

[0101] Specifically, the similarity between key features and various brand features in the brand feature library can be calculated using similarity calculation methods. Optional similarity calculation methods may include: Euclidean distance, cosine distance, Pearson correlation coefficient, or Jaccard similarity coefficient, etc.

[0102] S303. Based on the similarity between the key features and the features in the brand feature library, determine the brand features that match the POI to be detected.

[0103] Optionally, based on the similarity between the calculated key features and the features in the brand feature library, the most similar features in the brand feature library can be determined as the brand features that match the POI to be detected.

[0104] Optionally, the most similar features in the brand feature library can be obtained by comparing the calculated key features with features in the brand feature library. Then, the similarity of the features in the brand feature library with a preset similarity threshold can be compared to determine the brand features that match the POI to be detected. The preset similarity threshold can be set and adjusted based on the experience of technical personnel.

[0105] For example, assuming the similarity calculation method is cosine distance, the preset similarity threshold can be a preset cosine distance threshold. The cosine distance between the key feature and each feature in the brand feature library can be calculated. Based on the calculated cosine distances, the feature with the minimum cosine distance is selected as the most similar feature. Optionally, the feature in the brand feature library with the minimum cosine distance can be selected as the brand feature matching the POI to be detected. Alternatively, it can be determined whether the minimum cosine distance is less than the preset cosine distance threshold. If the minimum cosine distance is less than the preset cosine distance threshold, then the feature in the brand feature library corresponding to the minimum cosine distance is determined to be a brand feature matching the POI to be detected; if the minimum cosine distance is greater than or equal to the preset cosine distance threshold, then the feature in the brand feature library corresponding to the minimum cosine distance is determined to be a feature not matching the POI to be detected.

[0106] For example, assuming the similarity calculation method is cosine distance, and the preset similarity threshold is the preset cosine distance threshold, the key features include: key features of the name, key features of the alias (1), and key features of the alias (2). The cosine distances between the key features of the name, the key features of the alias (1), and the key features of the alias (2) and each feature in the brand feature library can be calculated respectively, to obtain the cosine distances between the key features of the name and each feature in the brand feature library, the cosine distances between the key features of the alias (1) and each feature in the brand feature library, and the cosine distances between the key features of the alias (2) and each feature in the brand feature library. The minimum value of the cosine distance of the key features of the name, the minimum value of the cosine distance of the key features of the alias (1), and the minimum value of the cosine distance of the key features of the alias (2) are compared to obtain the feature in the brand feature library corresponding to the minimum value of the three cosine distances. It can then be determined whether the minimum cosine distance is less than a preset cosine distance threshold. If the minimum cosine distance is less than the preset cosine distance threshold, then the feature in the brand feature library corresponding to the minimum cosine distance is determined to be a brand feature that matches the POI to be detected. If the minimum cosine distance is greater than or equal to the preset cosine distance threshold, then the feature in the brand feature library corresponding to the minimum cosine distance is determined to be a feature that does not match the POI to be detected.

[0107] S304. The brand words corresponding to the matched brand features are identified as the brand words corresponding to the POI to be detected.

[0108] In an optional embodiment of this disclosure, after determining the brand words corresponding to the matched brand features as the brand words corresponding to the POI to be detected, the POI processing method further includes: obtaining the brand words corresponding to the POI to be detected and adding tag information to the POI to be detected; receiving search keywords; querying the tag information corresponding to the search keywords and determining the POI with the corresponding tag information as the search result of the search keywords.

[0109] Step i involves obtaining the brand keywords corresponding to the POI to be detected and adding tag information to the POI to be detected.

[0110] Tag information can be POI matching information. By searching for tag information, you can find POIs that contain that tag.

[0111] Specifically, the brand words corresponding to the POI to be detected can be obtained, the brand words can be used as tag information, and tag information can be added to the POI to be detected.

[0112] Step ii: Receive search keywords.

[0113] Search keywords can be keywords from the information entered by the user when performing a POI search.

[0114] Specifically, it can receive search keywords entered by the user.

[0115] Step iii: Query the tag information corresponding to the search keywords, and determine the POIs of the corresponding tag information as the search results for the search keywords.

[0116] The tag information corresponding to the search keyword can include: tag information that is the same as the search keyword or tag information that contains the search keyword. Search results for the search keyword can be search results that include the tag information corresponding to the search keyword.

[0117] Specifically, you can query the tag information corresponding to the search keywords and identify the POIs with the corresponding tag information as the search results for the search keywords.

[0118] Optionally, when retrieving search results, you can prioritize retrieving search results with corresponding tag information for the search keywords; or you can perform a full search of search results, but prioritize displaying search results with corresponding tag information for the search keywords when displaying the search results.

[0119] For example, the brand term corresponding to the POI to be detected is "Brand W". "Brand W" is added as a tag to the POI. In a POI search scenario, the user-input search keyword is received, and the tag information related to the search keyword, i.e., "Brand W", is queried. The POI with the tag "Brand W" is identified as the search result for the search keyword. When retrieving search results, there can be 10 results, of which 5 results have tag information, i.e., the search results for the search keyword. Optionally, the 5 search results for the search keyword can be retrieved first, followed by 5 search results without tag information. On the search results display page, the 5 search results for the search keyword with tag information are displayed before the 5 search results without tag information. The display order of the 5 search results for the search keyword can be based on distance or other methods, which are not limited here. Optionally, all 10 search results can be retrieved simultaneously, but on the search results display page, the 5 search results for the search keyword with tag information are displayed before the 5 search results without tag information.

[0120] By acquiring brand keywords corresponding to the POIs to be detected, adding tag information to the POIs, receiving search keywords, querying the tag information corresponding to the search keywords, and identifying the POIs with the corresponding tag information as the search results for the search keywords, precise POI searching using tag information is achieved. By processing the POIs, the accuracy of the brand keywords corresponding to the POIs to be detected is improved. Adding tag information to the POIs using brand keywords ensures the accuracy of the tag information for the POIs to be detected. By receiving search keywords, querying the tag information corresponding to the search keywords, and identifying the POIs with the corresponding tag information as the search results for the search keywords, the accuracy of the search is improved based on more accurate tag information.

[0121] According to the technical solution disclosed herein, by calculating the similarity between key features and features in the brand feature library, the brand features that match the POI to be detected are determined based on the similarity between the key features and features in the brand feature library, thereby improving the efficiency of determining brand features and also improving the accuracy of the determined brand features.

[0122] Figure 4 This is a scene diagram illustrating another POI processing method disclosed in this embodiment. The POI processing method may include:

[0123] S401. Perform component detection on the name of the POI to be detected to obtain the main component information of the POI to be detected.

[0124] like Figure 4As shown, the name of the POI to be detected can be input into the keyword extraction model (e.g., the ERNIE model). The keyword extraction model performs component detection on the name of the POI to be detected and extracts the main component information of the name of the POI to be detected.

[0125] S402. Obtain the key features of the name of the POI to be detected.

[0126] like Figure 4 As shown, the core component information of the name of the POI to be detected can be input into the key feature extraction model (the same as the brand feature extraction model) to obtain the key features of the name of the POI to be detected. Features in the brand feature library are also obtained through the brand feature extraction model (e.g., the BERT model).

[0127] S403. Compare the key features of the name of the POI to be detected with the features in the brand feature library to obtain the brand features that match the POI to be detected.

[0128] like Figure 4 As shown, the key features of the POI's name can be compared with features in the brand feature library (e.g., similarity calculation) to determine the brand features that match the POI to be detected.

[0129] S404. The brand words corresponding to the matched brand features are identified as the brand words corresponding to the POI to be detected.

[0130] like Figure 4 As shown, brand words that correspond to the matched brand features are identified as the brand words corresponding to the POI to be detected.

[0131] S405. Post-process the brand keywords corresponding to the POI to be detected.

[0132] like Figure 4 As shown, post-processing is used to delete brand terms that do not meet the requirements. Optionally, post-processing may include: blacklist filtering of brand terms, tag filtering of brand terms, and sensitive industry filtering of brand terms. Blacklist filtering can be used to remove brands on a blacklist. Brands on the blacklist may include brands with poor quality or poor service. Sensitive industry filtering is used to remove brands belonging to sensitive industries. Tag filtering is used to remove brands that have already been tagged.

[0133] Optionally, the obtained brand terms can be pre-processed before storing the features corresponding to the brand terms in the brand feature database. Pre-processing is also used to delete non-compliant brand terms. Optionally, pre-processing may include: blacklist filtering of brand terms, tag filtering of brand terms, and sensitive industry filtering of brand terms. Non-compliant brand terms can be deleted at either the pre-processing or post-processing stage.

[0134] Specifically, text filtering algorithms can be used to post-process brand keywords. For example, text filtering algorithms may include: DFA (Deterministic Finite Automaton) algorithm, etc.

[0135] S406. Add the brand words that have passed the post-processing to the POI to be detected to obtain the brand POI.

[0136] like Figure 4 As shown, the brand words that pass the post-processing are identified as the brand words corresponding to the POI to be detected, and the brand words are added to the POI, thus transforming the POI from a POI to a brand POI. Here, a brand POI is a POI containing brand words.

[0137] According to the technical solution of this disclosure, the main component information of the name of the POI to be detected is determined by performing component detection on the name of the POI to be detected. Feature extraction is then performed on the main component information to obtain the key features of the name of the POI to be detected. These key features are compared with features in a brand feature library to determine the brand features that match the POI to be detected. The brand words corresponding to the matching brand features are then identified as the brand words corresponding to the POI to be detected, ensuring the accuracy of brand feature acquisition and thus the accuracy of the brand words corresponding to the POI to be detected. Post-processing is then performed on the brand words corresponding to the POI to be detected to filter out incompatible brand words. The post-processed brand words are then added to the POI to be detected, resulting in a brand POI. This process achieves the filtering of incompatible brand words and adds compatible brand words to the POI to be detected, facilitating POI search and improving the accuracy of POI search.

[0138] According to embodiments of this disclosure, Figure 5 This is a structural diagram of the POI processing device according to an embodiment of this disclosure. This embodiment is applicable to situations where a POI processing method is running. The device is implemented in software and / or hardware and is specifically configured in an electronic device with a certain data processing capability.

[0139] like Figure 5 The POI processing device 500 shown includes: a key feature acquisition module 501, a brand feature acquisition module 502, and a brand keyword determination module 503; wherein,

[0140] The key feature acquisition module 501 is used to acquire the key features of the POI to be detected.

[0141] The brand feature acquisition module 502 is used to compare key features with features in the brand feature library to obtain brand features that match the POI to be detected.

[0142] The brand term determination module 503 is used to determine the brand term corresponding to the matched brand features as the brand term corresponding to the POI to be detected.

[0143] According to the technical solution of this disclosure, by acquiring the key features of the POI to be detected and comparing the key features with features in the brand feature library, brand features matching the POI to be detected are obtained. By obtaining the brand feature library in advance, the efficiency of screening brand features matching the POI to be detected is improved. In the process of determining the brand features matching the POI to be detected, the key features of the POI to be detected are compared with the features in the brand feature library, which reduces the interference of invalid information and improves the accuracy of brand feature screening. At the same time, by comparing features, the accuracy of brand feature screening is further improved compared with the comparison of text strings. The brand words corresponding to the matched brand features are determined as the brand words corresponding to the POI to be detected, which improves the accuracy of the brand words matched by the POI to be detected.

[0144] In an optional embodiment of this disclosure, the key feature acquisition module 501 includes: an identification information acquisition unit for acquiring identification information of the POI to be detected; a backbone information acquisition unit for performing component detection on the identification information of the POI to be detected and extracting backbone component information; and a key feature extraction unit for extracting features from the backbone component information to obtain key features.

[0145] In an optional embodiment of this disclosure, the key feature acquisition module 501 further includes: an online POI acquisition unit for acquiring online POIs; a brand word determination unit for determining brand words based on the identification information of each online POI; and a brand word extraction unit for extracting features from the brand words and adding them to the brand feature library, wherein the feature extraction method for the brand words is the same as the feature extraction method for the main component information.

[0146] In an optional embodiment of this disclosure, the brand term determination unit includes: a candidate class acquisition subunit, used to cluster online POIs to obtain candidate classes; a public information determination subunit, used to determine the public information corresponding to the candidate class from the identification information of POIs in the same candidate class; a brand verification subunit, used to perform brand verification on the public information; a signboard image clustering subunit, used to perform image clustering on the signs of POIs in the target class that has passed brand verification; and a brand term detection subunit, used to detect whether the public information of the target class is a brand term based on the image clustering results.

[0147] In an optional embodiment of this disclosure, the identification information includes: the name and alias of the POI to be detected; the key features include: the key features of the name and the key features of the alias; the brand feature acquisition module 502 includes: a key feature comparison unit, used to compare the key features with features in the brand feature library to obtain the comparison results of each key feature; and a text feature acquisition unit, used to obtain the brand features that match the POI to be detected based on the comparison results of each key feature.

[0148] In an optional embodiment of this disclosure, the brand feature acquisition module 502 includes: a similarity calculation unit for calculating the similarity between key features and features in a brand feature library; and a text feature determination unit for determining brand features that match the POI to be detected based on the similarity between the key features and features in the brand feature library.

[0149] In an optional embodiment of this disclosure, the POI processing device further includes: a tag information adding module, used to obtain the brand words corresponding to the POI to be detected and add tag information to the POI to be detected; a search keyword receiving module, used to receive search keywords; and a search result determining module, used to query the tag information corresponding to the search keywords and determine the POI with the corresponding tag information as the search result of the search keywords.

[0150] The above-described POI processing apparatus can execute the POI processing method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the POI processing method.

[0151] 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.

[0152] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program object.

[0153] Figure 6 A schematic area diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, 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.

[0154] like Figure 6As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded from storage unit 608 into random access memory (RAM) 603. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0155] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0156] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 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 601 performs the various methods and processes described above, such as the POI processing method. For example, in some embodiments, the POI processing method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the POI processing method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the POI processing method by any other suitable means (e.g., by means of firmware).

[0157] 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 objects (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.

[0158] 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 a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or area diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0159] 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.

[0160] 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).

[0161] 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.

[0162] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0163] It should be understood that the various forms of processes shown above can be used to rearrange, 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.

[0164] 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 POI processing method, comprising: Obtain the key features of the POI to be detected; The key features are compared with features in the brand feature library to obtain brand features that match the POI to be detected; The brand words corresponding to the matched brand features are determined as the brand words corresponding to the POI to be detected; Prior to obtaining the key features of the POI to be detected, the process also includes: Get the POIs that are already online; Cluster the already launched Points of Interest (POIs) to obtain candidate classes; In the identification information of POIs of the same candidate class, determine the public information corresponding to the candidate class; Brand verification is performed on the aforementioned public information; In the target class where the brand verification is passed, image clustering is performed according to the signs of the target class's POI; Based on the image clustering results, detect whether the public information of the target class is a brand word; The brand words are feature extracted and added to the brand feature library.

2. The method according to claim 1, wherein, The key features for obtaining the POI to be detected include: Obtain the identification information of the POI to be detected; The identification information of the POI to be detected is subjected to component detection, and the main component information is extracted; Feature extraction is performed on the core component information to obtain key features.

3. The method according to claim 2, wherein the feature extraction method of the brand words is the same as the feature extraction method of the core component information.

4. The method according to claim 2, wherein, The identification information includes: the name and alias of the POI to be detected; the key features include: the key features of the name and the key features of the alias; The step of comparing the key features with features in the brand feature library to obtain brand features that match the POI to be detected includes: The key features are compared with features in the brand feature library to obtain the comparison results of each key feature; Based on the comparison results of each key feature, the brand features that match the POI to be detected are obtained.

5. The method according to claim 1, wherein, The step of comparing the key features with features in the brand feature library to obtain brand features that match the POI to be detected includes: Calculate the similarity between the key features and features in the brand feature library; Based on the similarity between the key features and the features in the brand feature library, the brand features that match the POI to be detected are determined.

6. The method according to claim 1, further comprising: Obtain the brand keywords corresponding to the POI to be detected, and add tag information to the POI to be detected; Receive search keywords; Query the tag information corresponding to the search keyword, and determine the POI of the corresponding tag information as the search result of the search keyword.

7. A POI processing apparatus, comprising: The key feature acquisition module is used to acquire the key features of the POI to be detected. The brand feature acquisition module is used to compare the key features with features in the brand feature library to obtain brand features that match the POI to be detected. The brand term determination module is used to determine the brand term corresponding to the matched brand feature as the brand term corresponding to the POI to be detected; The key feature acquisition module further includes: The POI acquisition unit is now online and is used to acquire POIs that are already online. The candidate class is a sub-unit used to cluster the online POIs to obtain candidate classes. The public information determination subunit is used to determine the public information corresponding to the candidate class from the identification information of the POIs in the same candidate class; The brand verification subunit is used to verify the brand of the public information. The signboard image clustering subunit is used to cluster signs according to the POI of the target class in the target class that has passed brand verification; The brand word detection subunit is used to detect whether the public information of the target class is a brand word based on the image clustering results. The brand term extraction unit is used to extract features from the brand terms and add them to the brand feature library.

8. The apparatus according to claim 7, wherein, The key feature acquisition module includes: The identification information acquisition unit is used to acquire the identification information of the POI to be detected; The backbone information acquisition unit is used to perform component detection on the identification information of the POI to be detected and extract backbone component information; The key feature extraction unit is used to extract features from the backbone component information to obtain key features.

9. The apparatus according to claim 8, wherein the feature extraction method of the brand term is the same as the feature extraction method of the main component information.

10. The apparatus according to claim 8, wherein, The identification information includes: the name and alias of the POI to be detected; the key features include: the key features of the name and the key features of the alias; The brand feature acquisition module includes: A key feature comparison unit is used to compare the key features with features in the brand feature library to obtain the comparison results of each key feature; The text feature acquisition unit is used to obtain the brand features that match the POI to be detected based on the comparison results of each of the key features.

11. The apparatus according to claim 7, wherein, The brand feature acquisition module includes: A similarity calculation unit is used to calculate the similarity between the key features and features in the brand feature library; The text feature determination unit is used to determine the brand features that match the POI to be detected based on the similarity between the key features and the features in the brand feature library.

12. The apparatus according to claim 7, further comprising: The tag information adding module is used to obtain the brand words corresponding to the POI to be detected and add tag information to the POI to be detected. The search keyword receiving module is used to receive search keywords; The search result determination module is used to query the tag information corresponding to the search keyword and determine the POI of the corresponding tag information as the search result of the search keyword.

13. 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 POI processing method according to any one of claims 1-6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the POI processing method according to any one of claims 1-6.

15. A computer program object comprising a computer program that, when executed by a processor, implements the POI processing method according to any one of claims 1-6.

Images