CN117312478B - Address positioning method and device, electronic equipment and storage medium - Google Patents
Address positioning method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN117312478B CN117312478B CN202311618045.0A CN202311618045A CN117312478B CN 117312478 B CN117312478 B CN 117312478B CN 202311618045 A CN202311618045 A CN 202311618045A CN 117312478 B CN117312478 B CN 117312478B
- Authority
- CN
- China
- Prior art keywords
- address
- query
- information
- embedded representation
- representation vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 72
- 239000013598 vector Substances 0.000 claims abstract description 171
- 238000004458 analytical method Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 79
- 238000003062 neural network model Methods 0.000 claims description 71
- 238000013528 artificial neural network Methods 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 description 14
- 238000004891 communication Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000000605 extraction Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 239000013604 expression vector Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Remote Sensing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides an address positioning method, an address positioning device, electronic equipment and a storage medium, belonging to the technical field of positioning and searching, wherein the method comprises the following steps: determining a geographic grid range and an embedded representation vector of a query address based on the query address input by a user; determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information; and carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address. The invention can realize effective address positioning, improve the address inquiry matching efficiency and improve the address positioning precision.
Description
Technical Field
The present invention relates to the field of location search technologies, and in particular, to an address location method, an address location device, an electronic device, and a storage medium.
Background
With the rapid development of internet technology and intelligent positioning technology, in more and more internet service scenes (such as logistics transportation and intelligent navigation service), users are required to acquire corresponding actual geographic position information by inputting address text information. In order to solve the above-mentioned problems, an address resolution (Geocoding) algorithm is generally used at present, which resolves the actual geographic location corresponding to the text address from the Geocoding database according to the address information input by the user.
However, due to the limitations of the existing Geocoding technology, the number of points of interest (Point Of Interest, POIs) in the Geocoding database is limited, not all text addresses entered by the user can be covered, and moreover, the text addresses entered by the user may be under non-normative conditions, which are frequently encountered with text addresses that do not appear in the Geocoding database. For text addresses which do not appear in the geocode database, the nearby related POIs are generally found out one by one in the geocode database, and then a rough geographic position is output in a mode of interpolation and the like. If the query does not reach the nearby POI or the query is inaccurate, then the location cannot be or is invalid, and the efficiency of address query matching is also low.
Disclosure of Invention
The invention provides an address positioning method, an address positioning device, electronic equipment and a storage medium, which are used for solving the defects that the accuracy of geographic position information provided by a Geocoding technology in the prior art is low and the efficiency of address inquiry and matching is also low.
The invention provides an address positioning method, which comprises the following steps:
determining a geographic grid range and an embedded representation vector of a query address based on the query address input by a user;
determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information;
and carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address.
According to the address positioning method provided by the invention, the determining of the geographical grid range and the embedded representation vector of the query address based on the query address input by the user comprises the following steps:
inputting the query address input by the user into a positioning prediction neural network model to obtain a geographic grid range and an embedded representation vector of the query address output by the positioning prediction neural network model;
The positioning prediction neural network model is obtained by training according to the text address samples and the corresponding sample labels.
According to the address positioning method provided by the invention, the positioning prediction neural network model comprises a basic backbone sub-network, a first prediction sub-network and a second prediction sub-network; the step of inputting the query address input by the user to a positioning prediction neural network model to obtain the geographic grid range and the embedded representation vector of the query address output by the positioning prediction neural network model comprises the following steps:
inputting the query address input by the user to the basic backbone sub-network to obtain a text semantic vector of the query address output by the basic backbone sub-network; the basic backbone sub-network is constructed based on a transducer encoder;
inputting the text semantic vector of the query address to the first prediction sub-network to obtain a geographic grid range of the query address output by the first prediction sub-network; the first prediction sub-network is constructed based on a full-connection layer neural network;
inputting the text semantic vector of the query address to the second prediction sub-network to obtain an embedded representation vector of the query address output by the second prediction sub-network; the second prediction sub-network is constructed based on a sentence vector pooling layer neural network and a full-connection layer neural network.
According to the address locating method provided by the invention, the matching analysis is carried out on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and the geographic position information of the query address is determined, and the method comprises the following steps:
calculating the distance between the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information by using a similarity matching method so as to determine similarity information of the embedded representation vector of the query address relative to the embedded representation vector of each piece of candidate address information;
determining target candidate address information corresponding to the maximum value in the similarity information from the candidate address information;
and obtaining the geographic position information of the query address according to the target candidate address information.
According to the address locating method provided by the invention, before the query address input by the user is input to the location prediction neural network model, the method further comprises the following steps:
taking the text address sample and the corresponding sample label thereof as a group of training samples, and obtaining a plurality of groups of training samples;
and training the positioning prediction neural network model by utilizing the plurality of groups of training samples.
According to the address positioning method provided by the invention, the training of the positioning prediction neural network model by utilizing the plurality of groups of training samples comprises the following steps:
for any group of training samples, inputting the training samples into a positioning prediction neural network model, and outputting the prediction probability corresponding to the training samples;
calculating a loss value according to the prediction probability corresponding to the training sample and the sample label in the training sample by using a preset loss function;
based on the loss value, adjusting model parameters of the positioning prediction neural network model until the model training times reach preset times;
and taking the model parameters obtained when the model training times reach the preset times as the model parameters of the trained positioning prediction neural network model.
According to the address positioning method provided by the invention, before the geographic grid range of the query address and the embedded representation vector are determined based on the query address input by the user, the method further comprises the following steps:
carrying out geographic grid division on the space region range, numbering each divided geographic grid, and determining the grid number of each geographic grid;
Carrying out association processing on all address text data and corresponding geographic coordinate information of the space region range and grid numbers of each geographic grid, and acquiring association information after the association processing;
constructing a geographic grid space index rule for all address text data and corresponding geographic coordinate information based on the association information, and determining embedded representation vectors of all address text data;
and storing the geographic grid space index rule, the embedded representation vector of all the address text data, all the address text data and the corresponding geographic coordinate information into a space database to obtain the address coordinate library.
The invention also provides an address positioning device, which comprises:
the processing module is used for determining the geographic grid range and the embedded representation vector of the query address based on the query address input by the user;
the matching module is used for determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information;
And the positioning module is used for carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the address location method as described in any of the above when executing the program.
The invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements an address location method as described in any of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements the address location method as described in any of the above.
According to the address positioning method, the device, the electronic equipment and the storage medium, the address coordinate library is constructed in advance by utilizing all address text data in the space area range and corresponding geographic coordinate information, and the semantic feature embedded representation of the address text is utilized to improve the address text query matching efficiency, so that after a query address input by a user is received, a plurality of candidate address information matched with the geographic grid range and the embedded representation vector of each candidate address information can be quickly matched from the address coordinate library by determining the geographic grid range and the embedded representation vector of the query address, further, the embedded representation vector of the query address and the embedded representation vector of each candidate address information are subjected to matching analysis, the geographic position information of the query address is obtained according to the successfully matched candidate address information, the effective address positioning is realized, the address query matching efficiency is improved, and the address positioning precision is improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of an address locating method according to the present invention;
FIG. 2 is a second flowchart of an address location method according to the present invention;
FIG. 3 is a schematic diagram of an address location apparatus according to the present invention;
fig. 4 is a schematic diagram of the physical structure of the electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the invention, it should be noted that, unless explicitly stated and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
The address location method, apparatus, electronic device, and storage medium of the present invention are described below with reference to fig. 1 to 4.
Fig. 1 is a schematic flow chart of an address positioning method according to the present invention, as shown in fig. 1, including: step 110, step 120 and step 130.
Step 110, determining a geographic grid range and an embedded representation vector of the query address based on the query address input by the user;
step 120, determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information;
And 130, carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address.
Specifically, the query address described in the embodiment of the present invention refers to text information that is input by a user on a front-end man-machine interaction interface and is used for describing a geographic location to be queried, and may specifically be a place name or detailed address information that includes regional levels of province, city, district, street/town, etc.
The embedded representation vector described in the embodiment of the invention refers to a vector representation obtained by extracting text features of a query address input by a user through a text feature extraction algorithm, and can be used for representing related semantic information of the text of the query address.
The geographical grid range described in the embodiment of the invention refers to a geographical grid range to which a query address input by a user belongs, which is obtained by predicting and query matching the query address by using a geographical position prediction algorithm or model.
The address coordinate library described in the embodiment of the invention is an address database which is constructed in advance based on all address text data of a space region range and corresponding geographic coordinate information, and the database stores all address data and embedded representation vector data of text semantic features corresponding to all address data.
In the embodiment of the present invention, in step 110, text feature extraction is performed on a query address input by a user by receiving query address information input by a front end of the user, and geographic location prediction and embedded vector representation are performed according to the extracted text feature, so as to obtain a geographic grid range and embedded representation vector of the query address.
Further, in the embodiment of the present invention, in step 120, according to the obtained geographical grid range, from the address coordinate library that is built in advance, a plurality of candidate address information that matches the geographical grid range may be found, and an embedded representation vector of each candidate address information may be determined.
Further, in the embodiment of the present invention, in step 130, after determining a plurality of candidate address information and corresponding embedded representation vectors that match the geographic grid range, matching analysis is performed on the embedded representation vector of the query address and the embedded representation vector of each candidate address information, and the matching may be performed on the candidate address information corresponding to the matched embedded representation vector by calculating the similarity between the embedded representation vector of the query address and the embedded representation vector of each candidate address information, so as to reversely deduce the candidate address information corresponding to the embedded representation vector of the query address, thereby finally using the candidate address information as the geographic location information of the query address, and thus completing rapid and accurate positioning of the query address input by the user.
According to the address positioning method, the address coordinate library is constructed in advance by utilizing all address text data in the space area range and corresponding geographic coordinate information, and the semantic feature embedded representation of the address text is utilized to improve the address text query matching efficiency, so that after a query address input by a user is received, a plurality of candidate address information matched with the geographic grid range and the embedded representation vector of each candidate address information can be quickly matched from the address coordinate library by determining the geographic grid range and the embedded representation vector of the query address, further, the embedded representation vector of the query address and the embedded representation vector of each candidate address information are subjected to matching analysis, the geographic position information of the query address is obtained according to the successfully matched candidate address information, the effective address positioning is realized, the address query matching efficiency is improved, and the address positioning accuracy is improved.
Based on the foregoing embodiment, as an alternative embodiment, determining the geographic grid range and the embedded representation vector of the query address based on the query address input by the user includes:
inputting the query address input by the user into the positioning prediction neural network model to obtain the geographic grid range and the embedded representation vector of the query address output by the positioning prediction neural network model;
The positioning prediction neural network model is obtained by training according to the text address samples and the corresponding sample labels.
Specifically, the positioning prediction neural network model described in the embodiment of the invention is obtained by training a deep neural network model according to a text address sample and a sample label corresponding to the text address sample, different text address information is identified based on strong learning capacity and characteristic extraction capacity of the deep neural network, association relations between text characteristics of the learned address and information of a real geographic grid range of the text characteristics are obtained, and the text characteristics of the learned address are converted into regular relations between embedded representation vectors of text semantic characteristics of the text characteristics, so that the geographic grid range corresponding to the text address and the corresponding embedded representation vectors are output simultaneously.
It should be noted that, for the embedded representation vector, a fixed neural network structure may be used in the positioning prediction neural network model to perform a vector embedded representation operation on the text address, so as to extract the embedded representation vector of the text address.
The deep neural network may be formed by a transducer neural network model including a multi-head attention mechanism, a pooling layer neural network, a front connection layer neural network, or the like, or may be other neural networks for accurately identifying and outputting a geographical grid range corresponding to a text address and embedding a representation vector, which is not particularly limited in the present invention.
The sample labels described in the embodiments of the present invention are predetermined according to the text address samples and are in one-to-one correspondence with the text address samples. That is, each text address sample in the training samples is preset to carry a corresponding sample label. Wherein the sample labels include labels required for training the model to predict the geographic grid range, and labels required for training the model to generate the embedded representation vector.
Further, in the embodiment of the invention, the query address input by the user is input to the trained positioning prediction neural network model, the deep neural network is utilized to identify and process the query address input by the user, and the geographic grid range and the embedded representation vector of the query address are obtained at the same time so as to prepare for subsequent address query matching.
According to the method provided by the embodiment of the invention, the deep neural network model with double-head output is adopted, the geographical grid range of the query address is calculated, the query address embedded representation vector for subsequent address matching calculation is calculated, the query address is only required to be processed once through the model, the problems of complicated processing steps and time consumption are greatly reduced, and the effects of text address query matching and positioning can be effectively improved.
Based on the foregoing embodiments, as an alternative embodiment, the positioning prediction neural network model includes a base backbone subnetwork, a first prediction subnetwork, and a second prediction subnetwork; inputting the query address input by the user into the positioning prediction neural network model to obtain the geographic grid range and the embedded representation vector of the query address output by the positioning prediction neural network model, wherein the method comprises the following steps:
inputting the query address input by the user into the basic backbone sub-network to obtain a text semantic vector of the query address output by the basic backbone sub-network; the basic backbone sub-network is constructed based on a transducer encoder;
inputting the text semantic vector of the query address into a first prediction sub-network to obtain a geographic grid range of the query address output by the first prediction sub-network; the first prediction sub-network is constructed based on a full-connection layer neural network;
inputting the text semantic vector of the query address into a second prediction sub-network to obtain an embedded representation vector of the query address output by the second prediction sub-network; the second prediction sub-network is constructed based on a sentence vector pooling layer neural network and a full-connection layer neural network.
Specifically, the basic backbone sub-network described in the embodiment of the present invention is used to extract text features of a query address input by a user, and output text semantic vectors of the query address, which may also be described as a backbone network. It is constructed based on a transducer encoder, which can be constructed specifically using the base version Roberta-wwm pre-trained model.
The base Roberta-wwm model is formed by stacking 12 transducer encoder modules, and each transducer encoder comprises 12 multi-head self-attention modules, 1 feedforward neural network and 1 residual error connection. The multi-headed self-attention mechanism divides the input sequence into a plurality of subspaces and performs self-attention computation for each subspace separately. In this way, the model may focus on information from different subspaces simultaneously, thereby more fully understanding the input sequence. In terms of training strategies, roberta-wwm removes the sentence-making prediction task proposed in the language model BERT and employs a dynamic whole word mask pre-training mechanism to update model parameters.
The first prediction subnetwork described in the embodiment of the present invention is used for predicting, according to the text semantic vector of the query address, the geographic location information, such as the geographic grid number, to which the query address text belongs, and may also be described as a geographic grid prediction subnetwork. The method is constructed based on a full-connection layer neural network, and specifically can be constructed by adopting two layers of full-connection neural networks, wherein the two full-connection layers are connected through a Tanh activation function.
The second prediction sub-network described in the embodiment of the present invention is configured to further perform feature mapping on a text semantic vector according to the text semantic vector of a query address, so as to obtain an embedded representation vector of the query address, which is used for matching a subsequent embedded representation vector, and may also be described as a query matching sub-network. The method is constructed based on a sentence vector pooling layer neural network and a full-connection layer neural network, and can be specifically composed of 1 sentence vector pooling layer and 1 full-connection layer.
Alternatively, in an embodiment of the present invention, when a user inputs a piece of inquiry address text information, two special identifiers [ CLS ] and [ SEP ] may be added first and last, respectively. Wherein [ CLS ] is located at the beginning of the sentence for aggregating text semantic vector representations of the whole text and as input to the first prediction sub-network, and [ SEP ] identifier is used for identifying the end of the sentence. And then, splitting the input query address into a plurality of independent characters by using a word segmentation device, and inputting the characters into a base backbone subnetwork Roberta-wwm model for feature extraction to obtain text semantic vectors of the query address.
In the embodiment of the present invention, geographical grid division is used for the spatial region range to which the address library belongs in advance, and each grid is represented by a unique number.
Further, in the embodiment of the present invention, the text semantic vector of the [ CLS ] identifier of the query address may be input to the first prediction sub-network, processed by the two-layer full-connection layer neural network in the first prediction sub-network, and converted into the probability that the query address belongs to each geographic grid number after the second full-connection layer by using a softmax function, and the result with the highest probability is output, thereby obtaining the geographic grid range of the query address, that is, the geographic grid number to which the query address belongs.
Further, in the embodiment of the invention, text semantic vectors of the query address are simultaneously input to a sentence vector pooling layer and a full-connection layer in the second prediction sub-network, the semantic vectors of each character in the query address output by the basic backbone sub-network are pooled into vector representations with fixed dimensions through the sentence vector pooling layer, and then the full-connection layer is used for further projecting the features to obtain the embedded representation vector of the query address finally used for subsequent similarity matching calculation.
According to the method, a deep neural network model with double-head output is constructed by adopting a basic backbone sub-network, a first prediction sub-network and a second prediction sub-network, so that the geographic grid number of a query address and a semantic feature embedded representation vector for address similarity calculation can be calculated simultaneously, a candidate address set is screened out from a global address coordinate library by utilizing a predicted geographic grid, similarity calculation with all addresses is avoided, and query matching efficiency is improved; and similarity calculation is carried out by utilizing the embedded expression vector of the query address, so that the effect of query matching is improved.
Based on the foregoing embodiment, as an optional embodiment, performing matching analysis on the embedded representation vector of the query address and the embedded representation vector of each candidate address information, to determine geographic location information of the query address includes:
Calculating the distance between the embedded representation vector of the query address and the embedded representation vector of each candidate address information by using a similarity matching method to determine similarity information of the embedded representation vector of the query address relative to the embedded representation vector of each candidate address information;
determining target candidate address information corresponding to the maximum value in the similarity information from the candidate address information;
and obtaining the geographic position information of the query address according to the target candidate address information.
Specifically, the target candidate address information described in the embodiment of the present invention refers to candidate address information with the greatest similarity determined by performing similarity matching calculation between an embedded representation vector of a query address and an embedded representation vector of each candidate address information.
In the embodiment of the invention, the similarity sorting can be performed according to the calculated Euclidean distance by calculating the Euclidean distance between the embedded representation vector of the query address and the embedded representation vector of each candidate address information. The smaller the Euclidean distance is, the larger the similarity is, and the smaller the Euclidean distance is. Therefore, the similarity information of the embedded representation vector of the query address relative to the embedded representation vector of each candidate address information can be determined, and the candidate address information can be further sequenced according to the similarity order.
Further, in the embodiment of the invention, the target candidate address information corresponding to the maximum value in the similarity information is determined from the candidate address information, and the target candidate address information is used as the geographic position information of the query address, so that the query address is positioned.
According to the method provided by the embodiment of the invention, the candidate address set is screened out from the global address coordinate library, and the similarity calculation is carried out by utilizing the embedded representation vector of the query address and the embedded representation vector of the candidate address, so that the complexity of similarity calculation with all global addresses is avoided, the efficiency of matching the query address is improved, and the effect of address positioning is improved.
Based on the foregoing embodiment, as an alternative embodiment, before inputting the query address input by the user into the location prediction neural network model, the method further includes:
taking the text address sample and the corresponding sample label thereof as a group of training samples, and obtaining a plurality of groups of training samples;
and training the positioning prediction neural network model by utilizing a plurality of groups of training samples.
Specifically, in the embodiment of the present invention, before the query address input by the user is input to the location prediction neural network model, the location prediction neural network model is further trained to obtain a trained location prediction neural network model.
In the embodiment of the invention, the training set data is utilized to train the positioning prediction neural network model, and the specific training process is as follows:
and taking the text address samples and the corresponding sample labels thereof as a group of training samples, and acquiring a plurality of groups of training samples aiming at different text address samples.
In the embodiment of the invention, the text address samples are in one-to-one correspondence with the sample tags carried by the text address samples.
Then, after a plurality of groups of training samples are obtained, the plurality of groups of training samples are sequentially input into the positioning prediction neural network model, and the positioning prediction neural network model is trained by utilizing the plurality of groups of training samples, namely:
the text address samples in each group of training samples and the corresponding sample labels are simultaneously input into a positioning prediction neural network model, model parameters in the positioning prediction neural network model are adjusted by calculating loss function values according to the prediction results output by each time in the positioning prediction neural network model, and under the condition that preset training termination conditions are met, the whole training process of the positioning prediction neural network model is finally completed, so that a trained positioning prediction neural network model is obtained.
According to the method provided by the embodiment of the invention, the text address sample and the corresponding sample label are used as a group of training samples, and the positioning prediction neural network model is trained by utilizing a plurality of groups of training samples, so that the model precision of the trained positioning prediction neural network model is improved.
Based on the foregoing embodiment, as an alternative embodiment, training the positioning prediction neural network model using multiple sets of training samples includes:
for any group of training samples, inputting the training samples into a positioning prediction neural network model, and outputting the prediction probability corresponding to the training samples;
calculating a loss value according to the prediction probability corresponding to the training sample and the label in the training sample by using a preset loss function;
based on the loss value, adjusting model parameters of the positioning prediction neural network model until the model training times reach preset times;
and taking the model parameters obtained when the model training times reach the preset times as the model parameters of the trained positioning prediction neural network model.
Specifically, the preset loss function described in the embodiment of the present invention refers to a loss function preset in a positioning prediction neural network model, and is used for model evaluation; the preset threshold refers to a threshold preset by the model, and is used for obtaining a minimum loss value and completing model training; the preset times refer to the preset maximum times of model iterative training.
After a plurality of groups of training samples are obtained, for any group of training samples, text address samples and corresponding sample labels in each group of training samples are simultaneously input into a positioning prediction neural network model, and the prediction probability of the corresponding results of the training samples is output.
On the basis, a loss value is calculated according to the prediction probability corresponding to the training sample and the sample label corresponding to the training sample by utilizing a preset loss function.
Further, after the loss value is obtained by calculation, the training process ends. And then, the model parameters of the positioning prediction neural network model are adjusted based on the loss value by using a Back Propagation (BP) algorithm, so that the weight parameters of each layer of the model in the positioning prediction neural network model are updated, then, the next training is carried out, and the model training is carried out repeatedly.
In the training process, if the training result of a certain group of training samples meets the preset training termination condition, if the loss value obtained by corresponding calculation is smaller than the preset threshold value, or the current iteration number reaches the preset number, the loss value of the model can be controlled within the convergence range, and the model training is ended. At this time, the obtained model parameters can be used as the model parameters of the trained positioning prediction neural network model, and the positioning prediction neural network model is trained, so that the trained positioning prediction neural network model is obtained.
According to the method provided by the embodiment of the invention, the loss value of the positioning prediction neural network model is controlled within the convergence range by utilizing the repeated iterative training of the positioning prediction neural network model by utilizing the plurality of groups of training samples, so that the accuracy of the model output result is improved, and the accuracy of the address positioning prediction result is improved.
Based on the foregoing embodiment, as an optional embodiment, before determining the geographic grid range of the query address and embedding the representation vector based on the query address input by the user, the method further includes:
carrying out geographic grid division on the space region range, numbering each divided geographic grid, and determining the grid number of each geographic grid;
carrying out association processing on all address text data and corresponding geographic coordinate information of the space region range and grid numbers of each geographic grid, and acquiring association information after the association processing;
constructing a geographic grid space index rule for all address text data and corresponding geographic coordinate information based on the associated information, and determining embedded expression vectors of all address text data;
and storing the geographic grid space index rule, the embedded representation vectors of all address text data, all address text data and corresponding geographic coordinate information into a space database to obtain an address coordinate library.
Specifically, in the embodiment of the present invention, before determining the geographical grid range of the query address and embedding the representation vector based on the query address input by the user, an address coordinate library needs to be constructed in advance.
In the embodiment of the invention, the spatial region range refers to the spatial region range covering the required query address, and can be a specific spatial region range on the earth or a global region range.
In the embodiment of the invention, firstly, the space region is divided into the geographical grids, and each divided geographical grid is uniquely numbered, and the unique grid number of each geographical grid is sequentially set.
And further, carrying out association processing on all address text data and corresponding geographic coordinate information of the space region range and the grid number of each geographic grid, so that the grid number of each geographic grid corresponds to the relevant address data and geographic coordinate information (such as longitude and latitude information), and thus, the association information between the address text data and the corresponding geographic coordinate information and each geographic grid number is constructed.
Further, in the embodiment of the present invention, based on the above-mentioned association information, a geographic grid spatial index rule is constructed for all address text data and corresponding geographic coordinate information, so that under the geographic grid spatial index rule, the corresponding geographic coordinate information can be searched and queried according to the grid number of the geographic grid.
And simultaneously, inputting all address texts in the address coordinate library into a basic backbone sub-network and a query matching sub-network of the positioning prediction neural network model, thereby obtaining embedded representation vectors of all address text data.
Finally, in the embodiment of the invention, the geographic grid space index rule, the embedded expression vector of all the address text data, all the address text data and the corresponding geographic coordinate information are stored in a space database, so that an address coordinate library is constructed.
The method of the embodiment of the invention can directly store the address data and the geographic coordinate data by utilizing the strong learning capacity and the characteristic extraction capacity of the deep neural network, and can perform query matching by utilizing the semantic characteristic embedded expression vector without constructing complex manual matching rules.
Fig. 2 is a second flow chart of the address locating method provided in the present invention, as shown in fig. 2, in a specific embodiment of the present invention, first, a query address input by a front end user is received, and the query address is first input to a backbone network to obtain a text semantic vector of the query address. And then, respectively inputting the text semantic vector of the query address into a geographic grid prediction sub-network and a query matching sub-network, predicting the geographic grid number to which the query address belongs through the geographic grid prediction sub-network, and outputting the embedded representation vector of the query address through the query matching sub-network.
Further, in this embodiment, the geographical grid coordinate range of the grid is obtained according to the geographical grid number to which the query address belongs. And then, selecting all address data in the geographic grid coordinate range area from an address coordinate library by utilizing space inquiry as candidate address information, and embedding a representative vector set of candidate addresses corresponding to the candidate address information, wherein the embedded representative vector set contains the embedded representative vectors of the candidate address information.
Further, in this embodiment, the embedded representation vector of the query address and the candidate address embedded representation vector set are input to a similarity calculation network layer to perform similarity matching analysis, the similarity between the embedded representation vector of the query address and the embedded representation vector of each candidate address information is calculated, and the best matching address and the coordinate calculation result are determined, so as to complete the positioning of the query address input by the user.
Compared with the existing text address positioning technology, the method of the embodiment of the invention can simultaneously calculate the geographic grid information of the query address and the semantic feature embedded representation vector for query matching, and compared with the method for predicting the geographic grid and query matching by respectively utilizing two same-scale deep neural network models, the method of the embodiment of the invention can reduce processing steps and save calculation resources.
Compared with a method for global retrieval by only using a query matching model based on a deep neural network, the method can effectively reduce the number of candidate addresses to be queried and matched through screening of the geographic grid, and improves the calculation efficiency. In addition, if the query address is not stored in the address coordinate library, the method can output the geographic range to which the query address belongs according to the predicted geographic grid number.
The address positioning device provided by the invention is described below, and the address positioning device described below and the address positioning method described above can be referred to correspondingly.
Fig. 3 is a schematic structural diagram of an address location device provided by the present invention, as shown in fig. 3, including:
a processing module 310, configured to determine a geographic grid range of the query address and an embedded representation vector based on the query address input by the user;
a matching module 320, configured to determine a plurality of candidate address information that matches the geographic grid range from the address coordinate library, and determine an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information;
The positioning module 330 is configured to perform matching analysis on the embedded representation vector of the query address and the embedded representation vector of each candidate address information, and determine geographic location information of the query address.
The address positioning device in this embodiment may be used to execute the address positioning method embodiment, and its principle and technical effects are similar, and will not be described herein.
According to the address positioning device, the address coordinate library is constructed in advance by utilizing all address text data in the space area range and corresponding geographic coordinate information, and the semantic feature embedded representation of the address text is utilized to improve the address text query matching efficiency, so that after a query address input by a user is received, a plurality of candidate address information matched with the geographic grid range and the embedded representation vector of each candidate address information can be quickly matched from the address coordinate library by determining the geographic grid range and the embedded representation vector of the query address, further, the embedded representation vector of the query address and the embedded representation vector of each candidate address information are subjected to matching analysis, the geographic position information of the query address is obtained according to the successfully matched candidate address information, the effective address positioning is realized, the address query matching efficiency is improved, and the address positioning accuracy is improved.
Fig. 4 is a schematic physical structure of an electronic device according to the present invention, as shown in fig. 4, the electronic device may include: processor 410, communication interface (Communications Interface) 420, memory 430 and communication bus 440, wherein processor 410, communication interface 420 and memory 430 communicate with each other via communication bus 440. The processor 410 may invoke logic instructions in the memory 430 to perform the address location method provided by the methods described above, the method comprising: determining a geographic grid range and an embedded representation vector of a query address based on the query address input by a user; determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information; and carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address.
Further, the logic instructions in the memory 430 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of performing the address locating method provided by the methods described above, the method comprising: determining a geographic grid range and an embedded representation vector of a query address based on the query address input by a user; determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information; and carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the address location method provided by the above methods, the method comprising: determining a geographic grid range and an embedded representation vector of a query address based on the query address input by a user; determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information; and carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (7)
1. An address locating method, comprising:
determining a geographic grid range and an embedded representation vector of a query address based on the query address input by a user;
determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library, and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information;
matching analysis is carried out on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and geographic position information of the query address is determined;
Wherein, based on the query address input by the user, determining the geographic grid range and the embedded representation vector of the query address comprises:
inputting the query address input by the user into a positioning prediction neural network model to obtain a geographic grid range and an embedded representation vector of the query address output by the positioning prediction neural network model;
the positioning prediction neural network model is obtained by training according to a text address sample and a corresponding sample label thereof;
the positioning prediction neural network model comprises a basic backbone sub-network, a first prediction sub-network and a second prediction sub-network; the step of inputting the query address input by the user to a positioning prediction neural network model to obtain the geographic grid range and the embedded representation vector of the query address output by the positioning prediction neural network model comprises the following steps:
inputting the query address input by the user to the basic backbone sub-network to obtain a text semantic vector of the query address output by the basic backbone sub-network; the basic backbone sub-network is constructed based on a transducer encoder;
inputting the text semantic vector of the query address to the first prediction sub-network to obtain a geographic grid range of the query address output by the first prediction sub-network; the first prediction sub-network is constructed based on a full-connection layer neural network;
Inputting the text semantic vector of the query address to the second prediction sub-network to obtain an embedded representation vector of the query address output by the second prediction sub-network; the second prediction sub-network is constructed based on a sentence vector pooling layer neural network and a full-connection layer neural network;
wherein, before the determining the geographic grid range of the query address and embedding the representation vector based on the query address input by the user, the method further comprises:
carrying out geographic grid division on the space region range, numbering each divided geographic grid, and determining the grid number of each geographic grid;
carrying out association processing on all address text data and corresponding geographic coordinate information of the space region range and grid numbers of each geographic grid, and acquiring association information after the association processing;
constructing a geographic grid space index rule for all address text data and corresponding geographic coordinate information based on the association information, and determining embedded representation vectors of all address text data;
and storing the geographic grid space index rule, the embedded representation vector of all the address text data, all the address text data and the corresponding geographic coordinate information into a space database to obtain the address coordinate library.
2. The address location method of claim 1, wherein said matching the embedded representation vector of the query address with the embedded representation vector of each of the candidate address information, determining the geographic location information of the query address, comprises:
calculating the distance between the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information by using a similarity matching method so as to determine similarity information of the embedded representation vector of the query address relative to the embedded representation vector of each piece of candidate address information;
determining target candidate address information corresponding to the maximum value in the similarity information from the candidate address information;
and obtaining the geographic position information of the query address according to the target candidate address information.
3. The address location method of claim 1, wherein prior to said entering the user-entered query address into a location-predicting neural network model, the method further comprises:
taking the text address sample and the corresponding sample label thereof as a group of training samples, and obtaining a plurality of groups of training samples;
and training the positioning prediction neural network model by utilizing the plurality of groups of training samples.
4. The address location method of claim 3, wherein training the location prediction neural network model using the plurality of sets of training samples comprises:
for any group of training samples, inputting the training samples into a positioning prediction neural network model, and outputting the prediction probability corresponding to the training samples;
calculating a loss value according to the prediction probability corresponding to the training sample and the sample label in the training sample by using a preset loss function;
based on the loss value, adjusting model parameters of the positioning prediction neural network model until the model training times reach preset times;
and taking the model parameters obtained when the model training times reach the preset times as the model parameters of the trained positioning prediction neural network model.
5. An address locating apparatus, comprising:
the processing module is used for determining the geographic grid range and the embedded representation vector of the query address based on the query address input by the user;
the matching module is used for determining a plurality of candidate address information matched with the geographical grid range from an address coordinate library and determining an embedded representation vector of each candidate address information; the address coordinate library is constructed based on all address text data of the space region range and corresponding geographic coordinate information;
The positioning module is used for carrying out matching analysis on the embedded representation vector of the query address and the embedded representation vector of each piece of candidate address information, and determining the geographic position information of the query address;
the processing module is specifically configured to:
inputting the query address input by the user into a positioning prediction neural network model to obtain a geographic grid range and an embedded representation vector of the query address output by the positioning prediction neural network model;
the positioning prediction neural network model is obtained by training according to a text address sample and a corresponding sample label thereof;
the positioning prediction neural network model comprises a basic backbone sub-network, a first prediction sub-network and a second prediction sub-network; the processing module is specifically further configured to:
inputting the query address input by the user to the basic backbone sub-network to obtain a text semantic vector of the query address output by the basic backbone sub-network; the basic backbone sub-network is constructed based on a transducer encoder;
inputting the text semantic vector of the query address to the first prediction sub-network to obtain a geographic grid range of the query address output by the first prediction sub-network; the first prediction sub-network is constructed based on a full-connection layer neural network;
Inputting the text semantic vector of the query address to the second prediction sub-network to obtain an embedded representation vector of the query address output by the second prediction sub-network; the second prediction sub-network is constructed based on a sentence vector pooling layer neural network and a full-connection layer neural network;
wherein, the device is specifically used for:
carrying out geographic grid division on the space region range, numbering each divided geographic grid, and determining the grid number of each geographic grid;
carrying out association processing on all address text data and corresponding geographic coordinate information of the space region range and grid numbers of each geographic grid, and acquiring association information after the association processing;
constructing a geographic grid space index rule for all address text data and corresponding geographic coordinate information based on the association information, and determining embedded representation vectors of all address text data;
and storing the geographic grid space index rule, the embedded representation vector of all the address text data, all the address text data and the corresponding geographic coordinate information into a space database to obtain the address coordinate library.
6. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the address location method of any of claims 1 to 4 when the program is executed by the processor.
7. A non-transitory computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the address location method according to any of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311618045.0A CN117312478B (en) | 2023-11-30 | 2023-11-30 | Address positioning method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311618045.0A CN117312478B (en) | 2023-11-30 | 2023-11-30 | Address positioning method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117312478A CN117312478A (en) | 2023-12-29 |
| CN117312478B true CN117312478B (en) | 2024-03-22 |
Family
ID=89255704
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311618045.0A Active CN117312478B (en) | 2023-11-30 | 2023-11-30 | Address positioning method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117312478B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008090334A (en) * | 2006-09-29 | 2008-04-17 | Nippon Telegr & Teleph Corp <Ntt> | Location analyzer, location analyzing method, its program, and recording medium |
| CN112347118A (en) * | 2021-01-08 | 2021-02-09 | 阿里云计算有限公司 | Data storage, query and generation method, database engine and storage medium |
| CN115129804A (en) * | 2022-07-19 | 2022-09-30 | 广州欢聚时代信息科技有限公司 | Address association method, device, equipment, medium and product |
| CN116431746A (en) * | 2023-03-03 | 2023-07-14 | 中国科学院空天信息创新研究院 | Address mapping method and device based on coding library, electronic equipment and storage medium |
| CN116955541A (en) * | 2023-09-21 | 2023-10-27 | 园测信息科技股份有限公司 | Address matching method and system integrating geographic element semantic segmentation and similarity |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10769222B2 (en) * | 2017-03-20 | 2020-09-08 | Facebook, Inc. | Search result ranking based on post classifiers on online social networks |
-
2023
- 2023-11-30 CN CN202311618045.0A patent/CN117312478B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008090334A (en) * | 2006-09-29 | 2008-04-17 | Nippon Telegr & Teleph Corp <Ntt> | Location analyzer, location analyzing method, its program, and recording medium |
| CN112347118A (en) * | 2021-01-08 | 2021-02-09 | 阿里云计算有限公司 | Data storage, query and generation method, database engine and storage medium |
| CN115129804A (en) * | 2022-07-19 | 2022-09-30 | 广州欢聚时代信息科技有限公司 | Address association method, device, equipment, medium and product |
| CN116431746A (en) * | 2023-03-03 | 2023-07-14 | 中国科学院空天信息创新研究院 | Address mapping method and device based on coding library, electronic equipment and storage medium |
| CN116955541A (en) * | 2023-09-21 | 2023-10-27 | 园测信息科技股份有限公司 | Address matching method and system integrating geographic element semantic segmentation and similarity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117312478A (en) | 2023-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111783419B (en) | Address similarity calculation method, device, equipment and storage medium | |
| CN108711422B (en) | Speech recognition method, speech recognition device, computer-readable storage medium and computer equipment | |
| CN112100529B (en) | Search content ordering method and device, storage medium and electronic equipment | |
| CN111475637B (en) | Data processing and training method and device for pushing knowledge points | |
| CN111382212B (en) | Associated address acquisition method and device, electronic equipment and storage medium | |
| CN107203526B (en) | Query string semantic demand analysis method and device | |
| CN111931077B (en) | Data processing method, device, electronic equipment and storage medium | |
| CN112381227B (en) | Neural network generation method and device, electronic equipment and storage medium | |
| CN112463944B (en) | Search type intelligent question-answering method and device based on multi-model fusion | |
| CN112052670B (en) | Address text word segmentation method, device, computer equipment and storage medium | |
| CN116629338B (en) | Model data processing method, related device, equipment and storage medium | |
| CN114461943B (en) | Deep learning-based multi-source POI semantic matching method and device and storage medium thereof | |
| CN116719520A (en) | Code generation method and device | |
| CN112711944B (en) | Word segmentation method and system, and word segmentation device generation method and system | |
| CN117312478B (en) | Address positioning method and device, electronic equipment and storage medium | |
| CN118095341A (en) | SimRank similarity calculation method based on deep neural network | |
| CN113535928A (en) | Service discovery method and system of long-term and short-term memory network based on attention mechanism | |
| CN117909754A (en) | Auxiliary power plant equipment defect elimination method and system based on twin neural network | |
| CN112380861A (en) | Model training method and device and intention identification method and device | |
| CN117010373A (en) | Recommendation method for category and group to which asset management data of power equipment belong | |
| CN116431746A (en) | Address mapping method and device based on coding library, electronic equipment and storage medium | |
| CN104572820B (en) | The generation method and device of model, importance acquisition methods and device | |
| CN116363469A (en) | Method, device and system for detecting infrared target with few samples | |
| CN113297456B (en) | Searching method, searching device, electronic equipment and storage medium | |
| CN112801271B (en) | Method for generating neural network, data processing method and intelligent driving control method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |