CN111125550A - Interest point classification method, device, equipment and storage medium - Google Patents
Interest point classification method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN111125550A CN111125550A CN201811296498.5A CN201811296498A CN111125550A CN 111125550 A CN111125550 A CN 111125550A CN 201811296498 A CN201811296498 A CN 201811296498A CN 111125550 A CN111125550 A CN 111125550A
- Authority
- CN
- China
- Prior art keywords
- interest point
- interest
- information
- vector
- point
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 239000013598 vector Substances 0.000 claims abstract description 157
- 238000013145 classification model Methods 0.000 claims abstract description 68
- 238000012549 training Methods 0.000 claims abstract description 44
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 description 9
- 238000013527 convolutional neural network Methods 0.000 description 8
- 230000003287 optical effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000003058 natural language processing Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Abstract
The embodiment of the invention discloses a method, a device, equipment and a storage medium for classifying points of interest. Wherein, the method comprises the following steps: generating an interest point name vector according to the name information of the interest point; generating interest point label vectors according to the label information of the interest points; and obtaining the category information of the interest points according to the name vectors and the label vectors of the interest points by adopting an interest point classification model obtained by pre-training. According to the technical scheme provided by the embodiment of the invention, the classification information of the interest points is determined by combining the two dimensional information of the names and the labels of the interest points and adopting the interest point classification model obtained by pre-training, so that the classification accuracy of the interest points is improved. And then, information pushing is carried out based on the category information of the interest points, so that the pushed information is more in line with the demand of the user.
Description
Technical Field
The embodiment of the invention relates to the technical field of artificial intelligence, in particular to a method, a device, equipment and a storage medium for classifying points of interest.
Background
With the development of science and technology, the appearance of electronic maps provides convenience for the life of people. The graph displayed in the electronic map basically consists of a point line surface, and the interest points serving as important components of point data are indispensable components in the electronic map; in theory, any building, area and point of interest that can be named can be displayed as the point of interest data, such as a restaurant, a community, a parking lot, a bus station, etc.
However, the category information of the point of interest directly affects the service based on the point of interest, for example, information push is performed based on the category of the point of interest, and therefore, the category identification of the point of interest is very critical. The existing method for identifying the category of the interest point has low identification precision, so that the information pushed based on the identified category of the interest point cannot meet the actual requirements of users. Therefore, it is important to provide a new method for accurately identifying the category of the interest point.
Disclosure of Invention
The embodiment of the invention provides a method, a device, equipment and a storage medium for classifying interest points, which improve the classification accuracy of the interest points.
In a first aspect, an embodiment of the present invention provides a method for classifying points of interest, where the method includes:
generating an interest point name vector according to the name information of the interest point;
generating interest point label vectors according to the label information of the interest points;
and obtaining the category information of the interest points according to the name vectors and the label vectors of the interest points by adopting an interest point classification model obtained by pre-training.
In a second aspect, an embodiment of the present invention further provides an apparatus for classifying points of interest, where the apparatus includes:
the name vector generating module is used for generating an interest point name vector according to the name information of the interest point;
the tag vector generating module is used for generating interest point tag vectors according to the tag information of the interest points;
and the category information acquisition module is used for acquiring category information of the interest points according to the interest point name vectors and the interest point label vectors by adopting an interest point classification model obtained through pre-training.
In a third aspect, an embodiment of the present invention further provides an apparatus, where the apparatus includes:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the point of interest classification method of any of the first aspects.
In a fourth aspect, an embodiment of the present invention further provides a storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for classifying points of interest as described in any of the first aspects.
According to the method, the device, the equipment and the storage medium for classifying the interest points, the name information and the label information of the interest points are respectively processed to obtain the name vectors and the label vectors of the interest points, and the name vectors and the label vectors of the interest points are trained by adopting a pre-trained interest point classification model to obtain the category information of the interest points. According to the method and the device, through combining the two dimension information of the name and the label of the interest point, the classification information of the interest point is determined by adopting the interest point classification model obtained through pre-training, and the classification accuracy of the interest point is improved. And then, information pushing is carried out based on the category information of the interest points, so that the pushed information is more in line with the demand of the user.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 is a flowchart of a method for classifying points of interest according to an embodiment of the present invention;
fig. 2 is a flowchart of a method for classifying points of interest according to a second embodiment of the present invention;
fig. 3 is a flowchart of a method for classifying points of interest according to a third embodiment of the present invention;
fig. 4 is a block diagram of a structure of an interest point classifying device according to a fourth embodiment of the present invention;
fig. 5 is a schematic structural diagram of an apparatus provided in the fifth embodiment of the present invention.
Detailed Description
The embodiments of the present invention will be described in further detail with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the embodiments of the invention and that no limitation of the invention is intended. It should be further noted that, for convenience of description, only some structures, not all structures, relating to the embodiments of the present invention are shown in the drawings.
Example one
Fig. 1 is a flowchart of a method for classifying points of interest according to an embodiment of the present invention. The embodiment is suitable for the situation of accurately classifying the interest points. The method can be executed by the interest point classification device provided by the embodiment of the invention, the device can be realized in a software and/or hardware mode, and the device can be integrated in a computing device. Referring to fig. 1, the method specifically includes:
and S110, generating an interest point name vector according to the name information of the interest point.
The interest points, i.e., "information points", are points that the navigation software provider records and can directly find a corresponding destination on the navigation software (e.g., an electronic map). Bubble icons such as scenic spots, government agencies, companies, markets, restaurants and the like displayed on the electronic map all represent points of interest; the name information of the interest point may include a name of the interest point, such as "starbucks", and may also include keywords in the name of the interest point, and the like.
An interest point name vector refers to a representation of name information of an interest point in a vector space, which can be obtained by processing, e.g., modeling, the text. For example, the names of the interest points can be processed by word segmentation and the like to extract keywords, and then the keywords are mapped into an N-dimensional real number vector by training through a text depth representation model word2vector to obtain the name vector of the interest points. Where N is generally a hyper-parameter in the model.
For example, generating the point of interest name vector according to the name information of the point of interest may include: and taking the name information of the interest point as the input of a bag-of-words model BOW, a text depth representation model word2vector or a topic model LDA to obtain the name vector of the interest point.
Among them, the Bag of Words model (BOW) is a method that is commonly used for representing documents, which first appears in the fields of NLP (Natural language processing) and IR (Information retrieval), and omits the syntax and the order of text, and can be used to represent each document as an N-dimensional vector. The topic model (LDA) is a document topic generation model, also called a three-layer bayesian probability model, and it adopts the method of the BOW model, and can represent each document as a word frequency vector. The word2vector model is used to convert words in the corpus into vectors.
Specifically, the name information of the interest point can be input into any one of a BOW model, a word2vector model or an LDA model, and the model can output the name vector of the interest point after analyzing the input information by combining the parameters of the model.
Generating the point of interest name vector according to the name information of the point of interest may also be: and determining keywords according to the name information of the interest points, and inputting the keywords into a pre-established word-word vector corresponding table for matching to obtain the name vectors of the interest points. The pre-established word-word vector correspondence table is obtained by training by using a word2vector model based on massive encyclopedic entry data as a prediction.
And S120, generating interest point label vectors according to the label information of the interest points.
The tag information of the interest point refers to descriptive information associated with the interest point, and can be obtained from the related description of the interest point, such as introduction, comments, business hours and the like. The number of the tag information of the interest point associated with one interest point may be one or more. For example, the name information of the point of interest is "gym building", and the tag information of the point of interest may include: the parking space is reserved, the variety is rich, and the like.
The interest point tag vector is a representation of tag information of one interest point in a vector space, and can be obtained by processing the tag information. For example, the tag information of each interest point may be processed based on the tag information of the interest point to obtain an interest point tag vector; for the convenience of calculation, the tag information of the interest point can also be processed into a binary vector.
And S130, obtaining the category information of the interest points according to the name vectors of the interest points and the label vectors of the interest points by adopting the interest point classification model obtained through pre-training.
The category refers to a classification of the interest points based on their functions or uses, such as shopping malls, entertainment, lodging, hospitals, schools, and the like. Correspondingly, the category information refers to the category to which the interest point belongs; the interest point classification model can be obtained by training based on a machine learning model in advance, and can also be obtained by training based on an automatic attribute switch deep trust network; the interest point classification model can also be formed by combining machine learning with an automatic attribute switch deep trust network, namely a two-stage or multi-stage model.
The model comprises an adjustable Attribute layer control mechanism, can automatically introduce proper connection between the Attribute and the hidden layer, and compared with a Convolutional Neural Network model (CNN), the AG-DBN model has good performance on discrete vector input and higher classification precision. For example, the AG-DBN model can accurately classify pictures. Therefore, in this embodiment, an AG-DBN model is preferably used for training to obtain an interest point classification model; or the machine learning is combined with an automatic attribute switch deep trust network, namely a two-stage or multi-stage model to form the interest point classification model.
If the interest point classification model is obtained based on the automatic attribute switch deep trust network training. Specifically, the interest points with known category information are used as training sample data, and the names of the interest points of the samples and the sample label information are respectively processed to obtain name vectors of the interest points of the samples and label vectors of the interest points of the samples; inputting the name vector of the sample interest point, the label vector of the sample interest point and the category information of the sample interest point into an automatic attribute switch deep trust network for training until the name vector of the interest point and the label vector of the interest point of the unknown interest point are input into the model, and the model can accurately output the category information of the unknown interest point according to the existing parameters, and is the interest point classification model at the moment.
For example, obtaining the category information of the interest point according to the name vector of the interest point and the tag vector of the interest point by using the interest point classification model obtained by pre-training may include: and taking the interest point name vector and the interest point label vector as the input of the interest point classification model to obtain the category information of the interest point. The interest point classification model is obtained based on automatic attribute switch deep trust network training.
Specifically, the name vector and the label vector of the interest point are used as input variables and input into the interest point classification model, and the interest point classification model is trained by combining parameters of the interest point classification model and outputs category information of the interest point. Compared with the conventional model obtained by machine learning, the AG-DBN model which is excellent in discrete vector input performance and has higher classification accuracy is adopted, so that the category information of the interest point can be accurately obtained.
According to the technical scheme provided by the embodiment of the invention, the name information and the label information of the interest point are respectively processed to obtain the name vector and the label vector of the interest point, and the name vector and the label vector of the interest point are trained by adopting a pre-trained interest point classification model to obtain the category information of the interest point. According to the method and the device, through combining the two dimension information of the name and the label of the interest point, the classification information of the interest point is determined by adopting the interest point classification model obtained through pre-training, and the classification accuracy of the interest point is improved. And then, information pushing is carried out based on the category information of the interest points, so that the pushed information is more in line with the demand of the user.
Example two
Fig. 2 is a flowchart of a method for classifying points of interest according to a second embodiment of the present invention, where this embodiment further explains, on the basis of the first embodiment, the category information of the points of interest obtained by using a pre-trained point of interest classification model according to a point of interest name vector and a point of interest tag vector. Referring to fig. 2, the method specifically includes:
and S210, generating an interest point name vector according to the name information of the interest point.
And S220, generating interest point label vectors according to the label information of the interest points.
And S230, taking the name vector of the interest point as the input of the first classification model to obtain the preliminary classification information.
The preliminary classification information refers to a preliminary classification of the interest points, and may include at least one classification result. The first classification model may be obtained by training in advance based on a machine learning model such as a CNN model or an LTSM model. Specifically, the name vector of the sample interest point and the preliminary classification information of the sample can be used as a training sample set, the training sample set is input into a convolutional neural network to train the convolutional neural network, and a first classification model is obtained after training of each sample. When an interest point name vector is input into the first classification model, the model judges the input interest point name vector by combining the existing parameters of the model and outputs corresponding preliminary classification information.
For example, the name information of the interest point is "jili building", the term segmentation and other processing are performed on the "jili building" to obtain two keywords of "jili" and "building", then a word2vector model is used to obtain an interest point name vector corresponding to the "jili building", and the interest point name vector is input to the first classification model to obtain preliminary classification information, which may include: shopping malls, business buildings (or office buildings), etc.
Because the training complexity of each model one by one is relatively large, in order to reduce the training complexity, it is optional to: the method comprises the steps that an interest point with known category information is used as training sample data, and name information and sample label information of the sample interest point are processed respectively to obtain a sample interest point name vector and a sample interest point label vector; inputting the name vector of the sample interest point into a convolutional neural network model to obtain primary classification information, inputting the primary classification information, the label vector of the sample interest point and the class information of the sample interest point into an automatic attribute switch deep belief network model, and training the two models simultaneously until the class information of the interest point can be accurately output by the automatic attribute switch deep belief network model, wherein the convolutional neural network model is a first classification model corresponding to the first classification model; the automatic attribute switch deep trust network model is the second classification model. According to the method, the output result of the convolutional neural network model does not need to be concerned, the classification result can be accurately output only by the automatic attribute switch deep confidence network model, and the training is stopped.
S240, the preliminary classification information and the interest point label vector are used as the input of a second classification model, and the category information of the interest point is obtained.
And the second classification model is obtained based on the deep trust network training of the automatic attribute switch. Optionally, the second classification model may be trained together with the first classification model; or may be trained separately. The second classification model can be used for screening the primary classification information obtained by the first classification model, and then interested classification information is obtained.
Specifically, in order to make the obtained category information of the interest points more accurate, the scheme obtains preliminary classification information based on the names of the interest points; and inputting the preliminary classification information and the interest point label vector into a second classification model obtained based on the automatic attribute switch deep trust network training, wherein the second classification model outputs the classification information of the interest point.
For example, the name information of the point of interest is "gym building", and the tag information of the point of interest includes: "there is parking space", "variety is abundant"; inputting the interest point name vector into the first classification model, and obtaining preliminary classification information as follows: shopping mall, business building (or office building). And processing the preliminary classification information to obtain a corresponding vector, inputting the vector and the interest label vector into a second classification model together, and outputting the category information of the interest point as a market by the model finally, wherein the category information conforms to the actual condition, so that the identification accuracy is improved.
According to the technical scheme provided by the embodiment of the invention, the name information and the label information of the interest point are respectively processed to obtain the name vector and the label vector of the interest point, and the name vector and the label vector of the interest point are trained to obtain the category information of the interest point by adopting a first classification model and a second classification model, namely a two-stage model, which is constructed by an automatic attribute switch deep trust network with good performance on discrete vector input. According to the method and the device, through combining two dimension information of the name and the label of the interest point, the classification model constructed by the automatic attribute switch deep trust network which is good in discrete vector input performance is adopted to determine the category information of the interest point, and the classification accuracy of the interest point is improved. And then, information pushing is carried out based on the category information of the interest points, so that the pushed information is more in line with the demand of the user.
EXAMPLE III
Fig. 3 is a flowchart of a method for classifying points of interest according to a third embodiment of the present invention, and this embodiment further explains, on the basis of the foregoing embodiment, generation of a point of interest tag vector according to tag information of a point of interest. Referring to fig. 3, the method specifically includes:
s310, generating an interest point name vector according to the name information of the interest point.
And S320, taking the dimension total amount of the label information of each interest point as the dimension of the interest point label vector.
In this embodiment, the number of the tag information of each interest point is the dimension of the tag information of the interest point; the total number of dimensions may be determined based on the dimensions of the tag information for each point of interest. For example, may be the sum of dimensions of the tag information for each point of interest. For example, three points of interest A, B and C, the number of tag information of point of interest a is 5; the number of the label information of the interest point B is 5; the number of the label information of the interest point C is 4; the total number of dimensions is 14.
For convenience of subsequent calculation, the dimensionality of the interest point label vector can be reduced, and therefore, the sum of the numbers of the label information of the non-intersection parts in the label information of each interest point can be regarded as the total dimensionality of the label information of each interest point. For example, if the number of the label information overlapped among the interest point a, the interest point B, and the interest point C is 1, the number of the label information overlapped between the interest point a and the interest point B is 2, the number of the label information overlapped between the interest point a and the interest point C is 1, and the number of the label information overlapped between the interest point B and the interest point C is 1, the total dimensionality amount may be determined to be 11.
Specifically, the total dimension amount may be determined based on the dimension of the tag information of each interest point, and the total dimension amount is used as the dimension of the interest point tag vector corresponding to the tag information of each interest point.
S330, aiming at each dimension in the interest point label vector, if the interest point has label information of the dimension, determining that the dimension in the interest point label vector takes a first numerical value; otherwise, determining the value on the dimension as a second value.
The first numerical value is a preset value corresponding to a certain dimension in the interest point label vector when the interest point has label information of the dimension; correspondingly, the second numerical value is a preset value corresponding to a certain dimension in the interest point label vector when the interest point does not have label information of the dimension. Optionally, the first value is different from the second value. For ease of calculation, the first value may be set to 1 and the second value may be set to 0.
For example, the interest point label vector is represented by a vector b, and all label information is numbered as biIf a certain point of interest contains tag information biThen b isi1, otherwise bi0. Wherein i takes a value of 0 to n. It should be noted that the numbering sequence of the tag information of each interest point may be sequentially numbered in ascending order according to the sequence of the interest point, or may be numbered in advance according to the sequence of the interest pointThe tag information of each point of interest is numbered in a predetermined order.
It should be noted that, the tag information of the interest point is processed in steps S320 and S330, and the obtained interest point tag vector is a binary vector, so that the training complexity of obtaining the category information of the interest point based on the model training in the following step is reduced.
And S340, obtaining the category information of the interest points according to the name vectors and the label vectors of the interest points by adopting the interest point classification model obtained by pre-training.
According to the technical scheme provided by the embodiment of the invention, the name information and the label information of the interest point are respectively processed to obtain the name vector and the label vector of the interest point, and the name vector and the label vector of the interest point are trained by adopting a pre-trained interest point classification model to obtain the category information of the interest point. According to the method and the device, through combining the two dimension information of the name and the label of the interest point, the classification information of the interest point is determined by adopting the interest point classification model obtained through pre-training, and the classification accuracy of the interest point is improved. And then, information pushing is carried out based on the category information of the interest points, so that the pushed information is more in line with the demand of the user.
Example four
Fig. 4 is a block diagram of a structure of an interest point classifying device according to a fourth embodiment of the present invention, which is capable of executing the interest point classifying method according to any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the executing method. As shown in fig. 4, the apparatus may include:
a name vector generating module 410, configured to generate a name vector of the interest point according to the name information of the interest point;
a tag vector generating module 420, configured to generate an interest point tag vector according to tag information of the interest point;
the category information obtaining module 430 is configured to obtain category information of the interest point according to the interest point name vector and the interest point tag vector by using an interest point classification model obtained through pre-training.
According to the technical scheme provided by the embodiment of the invention, the name information and the label information of the interest point are respectively processed to obtain the name vector and the label vector of the interest point, and the name vector and the label vector of the interest point are trained by adopting a pre-trained interest point classification model to obtain the category information of the interest point. According to the method and the device, through combining the two dimension information of the name and the label of the interest point, the classification information of the interest point is determined by adopting the interest point classification model obtained through pre-training, and the classification accuracy of the interest point is improved. And then, information pushing is carried out based on the category information of the interest points, so that the pushed information is more in line with the demand of the user.
Illustratively, the category information obtaining module 430 may further be configured to:
taking the name vector and the label vector of the interest point as the input of an interest point classification model to obtain the category information of the interest point; the interest point classification model is obtained based on automatic attribute switch deep trust network training.
Illustratively, the category information obtaining module 430 may further be configured to:
taking the name vector of the interest point as the input of a first classification model to obtain primary classification information;
using the preliminary classification information and the interest point label vector as the input of a second classification model to obtain the category information of the interest point; and the second classification model is obtained based on the deep trust network training of the automatic attribute switch.
Illustratively, the name vector generation module 410 is specifically configured to:
and taking the name information of the interest point as the input of a bag-of-words model BOW, a text depth representation model word2vector or a topic model LDA to obtain the name vector of the interest point.
Illustratively, the tag vector generation module 420 is specifically configured to:
taking the dimension total amount of the label information of the interest points as the dimension of the label vector of the interest points;
for each dimension in the interest point label vector, if the interest point has label information of the dimension, determining that the value of the dimension in the interest point label vector is a first numerical value; otherwise, determining the value on the dimension as a second value.
EXAMPLE five
Fig. 5 is a schematic structural diagram of an apparatus according to a fifth embodiment of the present invention, and fig. 5 shows a block diagram of an exemplary apparatus suitable for implementing the embodiment of the present invention. The device 12 shown in fig. 5 is only an example and should not bring any limitations to the functionality and scope of use of the embodiments of the present invention. As shown in FIG. 5, device 12 is in the form of a general purpose computing device. The components of device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.
The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM)30 and/or cache memory 32. Device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 5, and commonly referred to as a "hard drive"). Although not shown in FIG. 5, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.
A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally carry out the functions and/or methodologies of embodiments described herein.
The processing unit 16 executes programs stored in the system memory 28 to perform various functional applications and data processing, such as implementing the point of interest classification method provided by the embodiments of the present invention.
EXAMPLE six
An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program (or referred to as computer-executable instructions) is stored, where the computer program, when executed by a processor, can implement the method for classifying points of interest described in any of the above embodiments.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for embodiments of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the embodiments of the present invention have been described in more detail through the above embodiments, the embodiments of the present invention are not limited to the above embodiments, and many other equivalent embodiments may be included without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (12)
1. A method for classifying points of interest, comprising:
generating an interest point name vector according to the name information of the interest point;
generating interest point label vectors according to the label information of the interest points;
and obtaining the category information of the interest points according to the name vectors and the label vectors of the interest points by adopting an interest point classification model obtained by pre-training.
2. The method of claim 1, wherein obtaining the category information of the interest point according to the interest point name vector and the interest point tag vector by using an interest point classification model obtained by pre-training comprises:
taking the name vector and the label vector of the interest point as the input of an interest point classification model to obtain the category information of the interest point; the interest point classification model is obtained based on automatic attribute switch deep trust network training.
3. The method of claim 1, wherein obtaining the category information of the interest point according to the interest point name vector and the interest point tag vector by using an interest point classification model obtained by pre-training comprises:
taking the name vector of the interest point as the input of a first classification model to obtain primary classification information;
taking the preliminary classification information and the interest point label vector as the input of a second classification model to obtain the category information of the interest point;
and the second classification model is obtained based on the deep trust network training of the automatic attribute switch.
4. The method of claim 1, wherein generating a point of interest name vector based on name information of the point of interest comprises:
and taking the name information of the interest point as the input of a bag-of-words model BOW, a text depth representation model word2vector or a topic model LDA to obtain the name vector of the interest point.
5. The method of claim 1, wherein generating a point of interest tag vector based on tag information of a point of interest comprises:
taking the dimension total amount of the label information of each interest point as the dimension of the label vector of the interest point;
for each dimension in the interest point label vector, if the interest point has label information of the dimension, determining that the value of the dimension in the interest point label vector is a first numerical value; otherwise, determining the value on the dimension as a second value.
6. An apparatus for classifying a point of interest, comprising:
the name vector generating module is used for generating an interest point name vector according to the name information of the interest point;
the tag vector generating module is used for generating interest point tag vectors according to the tag information of the interest points;
and the category information acquisition module is used for acquiring category information of the interest points according to the interest point name vectors and the interest point label vectors by adopting an interest point classification model obtained through pre-training.
7. The apparatus of claim 6, wherein the category information obtaining module is further configured to:
taking the name vector and the label vector of the interest point as the input of an interest point classification model to obtain the category information of the interest point; the interest point classification model is obtained based on automatic attribute switch deep trust network training.
8. The apparatus of claim 6, wherein the category information obtaining module is further configured to:
taking the name vector of the interest point as the input of a first classification model to obtain primary classification information;
taking the preliminary classification information and the interest point label vector as the input of a second classification model to obtain the category information of the interest point;
and the second classification model is obtained based on the deep trust network training of the automatic attribute switch.
9. The apparatus of claim 6, wherein the name vector generation module is specifically configured to:
and taking the name information of the interest point as the input of a bag-of-words model BOW, a text depth representation model word2vector or a topic model LDA to obtain the name vector of the interest point.
10. The apparatus of claim 6, wherein the tag vector generation module is specifically configured to:
taking the dimension total amount of the label information of the interest points as the dimension of the label vector of the interest points;
for each dimension in the interest point label vector, if the interest point has label information of the dimension, determining that the value of the dimension in the interest point label vector is a first numerical value; otherwise, determining the value on the dimension as a second value.
11. An apparatus, characterized in that the apparatus comprises:
one or more processors;
storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the point of interest classification method of any of claims 1-5.
12. A storage medium on which a computer program is stored which, when being executed by a processor, carries out the method of interest classification according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811296498.5A CN111125550B (en) | 2018-11-01 | 2018-11-01 | Point-of-interest classification method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811296498.5A CN111125550B (en) | 2018-11-01 | 2018-11-01 | Point-of-interest classification method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111125550A true CN111125550A (en) | 2020-05-08 |
| CN111125550B CN111125550B (en) | 2023-11-24 |
Family
ID=70494926
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811296498.5A Active CN111125550B (en) | 2018-11-01 | 2018-11-01 | Point-of-interest classification method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111125550B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111767359A (en) * | 2020-06-30 | 2020-10-13 | 北京百度网讯科技有限公司 | Interest point classification method, device, equipment and storage medium |
| CN112328896A (en) * | 2020-11-26 | 2021-02-05 | 北京百度网讯科技有限公司 | Method, apparatus, electronic device, and medium for outputting information |
| CN112579793A (en) * | 2020-12-24 | 2021-03-30 | 北京创鑫旅程网络技术有限公司 | Model training method and POI (Point of interest) label detection method and device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070110292A1 (en) * | 2005-11-14 | 2007-05-17 | Jinbo Bi | System and method for computer aided detection via asymmetric cascade of sparse linear classifiers |
| US20100172584A1 (en) * | 2009-01-07 | 2010-07-08 | Rastislav Lukac | Method Of Classifying Red-Eye Objects Using Feature Extraction And Classifiers |
| CN106682217A (en) * | 2016-12-31 | 2017-05-17 | 成都数联铭品科技有限公司 | Method for enterprise second-grade industry classification based on automatic screening and learning of information |
| CN107403198A (en) * | 2017-07-31 | 2017-11-28 | 广州探迹科技有限公司 | A kind of official website recognition methods based on cascade classifier |
| CN107679189A (en) * | 2017-09-30 | 2018-02-09 | 百度在线网络技术（北京）有限公司 | A kind of point of interest update method, device, server and medium |
| CN108021940A (en) * | 2017-11-30 | 2018-05-11 | 中国银联股份有限公司 | data classification method and system based on machine learning |
| CN108171276A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术（北京）有限公司 | For generating the method and apparatus of information |
| CN108363698A (en) * | 2018-03-13 | 2018-08-03 | 腾讯大地通途（北京）科技有限公司 | Point of interest relation recognition method and device |
-
2018
- 2018-11-01 CN CN201811296498.5A patent/CN111125550B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070110292A1 (en) * | 2005-11-14 | 2007-05-17 | Jinbo Bi | System and method for computer aided detection via asymmetric cascade of sparse linear classifiers |
| US20100172584A1 (en) * | 2009-01-07 | 2010-07-08 | Rastislav Lukac | Method Of Classifying Red-Eye Objects Using Feature Extraction And Classifiers |
| CN106682217A (en) * | 2016-12-31 | 2017-05-17 | 成都数联铭品科技有限公司 | Method for enterprise second-grade industry classification based on automatic screening and learning of information |
| CN107403198A (en) * | 2017-07-31 | 2017-11-28 | 广州探迹科技有限公司 | A kind of official website recognition methods based on cascade classifier |
| CN107679189A (en) * | 2017-09-30 | 2018-02-09 | 百度在线网络技术（北京）有限公司 | A kind of point of interest update method, device, server and medium |
| CN108021940A (en) * | 2017-11-30 | 2018-05-11 | 中国银联股份有限公司 | data classification method and system based on machine learning |
| CN108171276A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术（北京）有限公司 | For generating the method and apparatus of information |
| CN108363698A (en) * | 2018-03-13 | 2018-08-03 | 腾讯大地通途（北京）科技有限公司 | Point of interest relation recognition method and device |
Non-Patent Citations (10)
| Title |
|---|
| A. F. R. RAHMAN ET AL.: "Serial Combination of Multiple Experts:A Unified Evaluation", PATTERN ANALYSIS AND APPLICATION, pages 4 * |
| XIAOMING JIN ET AL.: ""Automatic gating of attributes in deep structure"", 《ACM》 * |
| XIAOMING JIN ET AL.: ""Automatic gating of attributes in deep structure"", 《ACM》, 13 July 2018 (2018-07-13), pages 3 * |
| XIAOMING JIN ET AL.: "Automatic Gating of Attributes in Deep Structure", 《ACM》, pages 4 * |
| XU L.: "Methods of Combining MultiPle Classifiers and Their APPlication to Handwriting ReCognitio", IEEE TRANSACTIONS ON SYSTEMS，MAN，AND CYBERNETIES * |
| 刘清泉: "基于混合支持向量机多分类器的交通事件检测方法研究", 《中国优秀硕士学位论文全文数据库》 * |
| 康琦: "《机器学习中的不平衡分类方法》", 31 October 2017 * |
| 柳杨: "《数字图像物体识别理论详解与实战》", 31 January 2018, pages: 62 - 63 * |
| 胡志坚: "《技术前瞻与评价 第1卷 第2辑 2015版》", 31 October 2015 * |
| 蒋艳凰,杨学军: "多层组合分类器研究", 计算机工程与科学 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111767359A (en) * | 2020-06-30 | 2020-10-13 | 北京百度网讯科技有限公司 | Interest point classification method, device, equipment and storage medium |
| CN111767359B (en) * | 2020-06-30 | 2023-09-01 | 北京百度网讯科技有限公司 | Point-of-interest classification method, device, equipment and storage medium |
| CN112328896A (en) * | 2020-11-26 | 2021-02-05 | 北京百度网讯科技有限公司 | Method, apparatus, electronic device, and medium for outputting information |
| CN112328896B (en) * | 2020-11-26 | 2024-03-15 | 北京百度网讯科技有限公司 | Method, apparatus, electronic device, and medium for outputting information |
| CN112579793A (en) * | 2020-12-24 | 2021-03-30 | 北京创鑫旅程网络技术有限公司 | Model training method and POI (Point of interest) label detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111125550B (en) | 2023-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110795543B (en) | Unstructured data extraction method, device and storage medium based on deep learning | |
| CN107908635B (en) | Method and device for establishing text classification model and text classification | |
| CN108052577B (en) | Universal text content mining method, device, server and storage medium | |
| US11698261B2 (en) | Method, apparatus, computer device and storage medium for determining POI alias | |
| CN107210035B (en) | Generation of language understanding systems and methods | |
| CN112329467A (en) | Address recognition method and device, electronic equipment and storage medium | |
| CN112712069B (en) | Question judging method and device, electronic equipment and storage medium | |
| CN116795973B (en) | Text processing method and device based on artificial intelligence, electronic equipment and medium | |
| CN111767359B (en) | Point-of-interest classification method, device, equipment and storage medium | |
| CN113094550A (en) | Video retrieval method, device, equipment and medium | |
| CN113220835B (en) | Text information processing method, device, electronic equipment and storage medium | |
| CN111125550B (en) | Point-of-interest classification method, device, equipment and storage medium | |
| CN102930048A (en) | Data abundance automatically found by semanteme and using reference and visual data | |
| CN111259262A (en) | Information retrieval method, device, equipment and medium | |
| CN111126084B (en) | Data processing method, device, electronic equipment and storage medium | |
| CN114120166B (en) | Video question-answering method and device, electronic equipment and storage medium | |
| CN110263340B (en) | Comment generation method, comment generation device, server and storage medium | |
| CN111310065A (en) | Social contact recommendation method and device, server and storage medium | |
| CN109858024B (en) | Word2 vec-based room source word vector training method and device | |
| CN111143534A (en) | Method and device for extracting brand name based on artificial intelligence and storage medium | |
| CN114462357A (en) | Entity linking method, entity linking device, entity constructing method and entity linking device and storage medium | |
| CN108491387B (en) | Method and apparatus for outputting information | |
| WO2021104274A1 (en) | Image and text joint representation search method and system, and server and storage medium | |
| CN113254814A (en) | Network course video labeling method and device, electronic equipment and medium | |
| CN110362688A (en) | Examination question mask method, device, equipment and computer readable storage medium |
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 |