CN111782955A - Interest point representing and pushing method and device, electronic equipment and storage medium - Google Patents
Interest point representing and pushing method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN111782955A CN111782955A CN202010624873.5A CN202010624873A CN111782955A CN 111782955 A CN111782955 A CN 111782955A CN 202010624873 A CN202010624873 A CN 202010624873A CN 111782955 A CN111782955 A CN 111782955A
- Authority
- CN
- China
- Prior art keywords
- point
- interest
- user
- target user
- candidate
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 239000013598 vector Substances 0.000 claims abstract description 133
- 239000011159 matrix material Substances 0.000 claims abstract description 88
- 238000012549 training Methods 0.000 claims abstract description 59
- 230000006399 behavior Effects 0.000 claims description 24
- 238000004590 computer program Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 12
- 230000006870 function Effects 0.000 description 9
- 238000012545 processing Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—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
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- 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
- G06F16/90—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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
-
- 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
Abstract
The embodiment of the specification discloses a method, a device, electronic equipment and a storage medium for expressing and pushing interest points, wherein a historical track sequence of a user is constructed according to an interest point check-in record of a user set and a GPS (global positioning system) point acquisition record; and training a word expression model according to the historical track sequence of the user, wherein the obtained position point embedding matrix comprises position semantic vector expression of each interest point in the interest point sign-in record and position semantic vector expression of each position point in the GPS point collection record.
Description
Technical Field
The embodiment of the specification relates to the field of data processing, in particular to a method and a device for expressing and pushing a point of interest, electronic equipment and a storage medium.
Background
A POI (Point of interest) is a landmark or a sight spot on an electronic map, and is used to mark places such as commercial establishments, tourist spots, historic sites, traffic facilities, etc. of various industries represented by the place. Millions of interest points POIs exist in the real world, so that check-in data of user interest points are very sparse, with the rapid development of the mobile internet, Location-based Social networking service (LBSN) is more and more important in life, experience and experience are shared with friends or net friends through check-in behaviors and comments on a specified position, and interest point recommendation becomes the most important application in Location services, so that a user is guided to dig out a potentially interesting place and a very large scene exists in tourism, catering and life services.
Disclosure of Invention
The embodiment of the specification provides a method and a device for expressing and pushing a point of interest, an electronic device and a storage medium, so that the point of interest can be better pushed to a user.
In a first aspect, an embodiment of the present specification provides a method for representing a point of interest, including: acquiring an interest point check-in record and a GPS (global positioning system) point acquisition record of a user set, wherein a plurality of position points in the GPS point acquisition record are overlapped with interest points in the interest point check-in record; according to the interest point check-in record and the GPS point collecting record, constructing a historical track sequence of each user in the user set; and training a word expression model by using the historical track sequence of each user in the user set to obtain a position point embedding matrix, wherein the position point embedding matrix comprises position semantic vector expression of each interest point in the interest point check-in record and position semantic vector expression of each position point in the GPS point collection record.
In a second aspect, an embodiment of the present specification provides a point of interest pushing method, including: determining a target user, wherein the target user belongs to any user with interest point pushing requirements in all users;
and pushing interest points to the target user according to a position point embedding matrix, wherein the position point embedding matrix is obtained by the interest point representing method in the first aspect.
In a third aspect, an embodiment of the present specification provides an interest point representation apparatus, including a record acquisition unit, configured to acquire an interest point check-in record of a user set and a GPS collection point record, where a plurality of location points in the GPS collection point record coincide with interest points in the interest point check-in record; the track construction unit is used for constructing a historical track sequence of each user in the user set according to the interest point check-in record and the GPS point collection record; and the model training unit is used for training a word representation model by using the historical track sequence of each user in the user set to obtain a position point embedding matrix, wherein the position point embedding matrix comprises position semantic vector representation of each interest point in the interest point check-in record and position semantic vector representation of each position point in the GPS point collection record.
In a fourth aspect, an embodiment of the present specification provides a point of interest pushing apparatus, including: the system comprises a determining unit, a sending unit and a receiving unit, wherein the determining unit is used for determining a target user, and the target user belongs to any user with interest point pushing requirements in all users; and the interest point pushing module is used for pushing interest points to the target user according to a position point embedding matrix, and the position point embedding matrix is obtained by the interest point representing method of the first aspect.
In a fifth aspect, embodiments of the present specification provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor implements the steps of the method according to the first aspect or the second aspect when executing the program.
In a sixth aspect, embodiments of the present specification provide a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the method according to the first aspect or the second aspect.
One or more technical solutions provided in the embodiments of the present description at least achieve the following technical effects or advantages:
the method comprises the steps that a historical track sequence of a user is constructed according to a point of interest check-in record of a user set and a GPS point acquisition record; the word expression model is trained according to a historical track sequence of a user, an obtained position point embedding matrix contains position semantic vector expression of each position point in the GPS acquisition point record, so that the mapping of the interest points from a high-dimensional sparse space to a low-dimensional dense space is realized to obtain the vector expression of the interest points, the problem of sparse data of the interest points due to excessive interest points is solved, the GPS acquisition point information integrated into the vector expression of the interest points is endowed with stronger semantic information of the interest points on the geographic positions, and therefore, the interest points are pushed more accurately by using the position point embedding matrix.
Drawings
In order to more clearly illustrate the embodiments of the present specification or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only the embodiments of the present specification, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a schematic view of a scenario of a point of interest pushing method in an embodiment of the present disclosure;
FIG. 2 is a flow chart of a point of interest representation method in an embodiment of the present description;
FIG. 3 is a diagram of a training word representation model in an embodiment of the present specification;
FIG. 4 is a flowchart of a point of interest pushing method in an embodiment of the present disclosure;
FIG. 5 is a functional block diagram of a point of interest representation apparatus in an embodiment of the present disclosure;
FIG. 6 is a functional block diagram of a point of interest pushing apparatus in an embodiment of the present disclosure;
fig. 7 is a schematic structural diagram of an electronic device in an embodiment of the present specification.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are some, but not all embodiments of the present disclosure. All other embodiments obtained by a person skilled in the art based on the embodiments in the present specification without any inventive step are within the scope of the present specification.
The point of interest pushing method provided by the embodiment of the present specification is applied to the scenario shown in fig. 1. Referring to fig. 1, the scenario includes: the server 20 and the client 10 of the service provided by the server 20, the client 10 may be deployed on a mobile device, and the client 10 may be any target application based on the LBSN. The client 10 registers the interest points and reports the GPS acquisition point records to the server 20, the server 20 collects the interest point registration records and the GPS acquisition point records reported by the client 10, so as to obtain the user set interest point registration records and the GPS acquisition point records, and a plurality of position points in the GPS acquisition point records are overlapped with the interest points in the interest point registration records; the method comprises the steps of constructing a historical track sequence of each user in a user set according to interest point check-in records and GPS point collection records of the user set, training a word expression model by using the historical track sequence of each user in the user set, and obtaining a position point embedding matrix, wherein the position point embedding matrix comprises position semantic vector expression of each interest point in the interest point check-in records and position semantic vector expression of each position point in the GPS point collection records. And the server 20 pushes the interest points to the target user according to the position point embedding matrix. Or, the server 20 sends the obtained location point embedded matrix to other servers, and other server devices push the interest points to the target user according to the location point embedded matrix.
The vector representation of the interest points is obtained through mapping from a high-dimensional sparse space to a low-dimensional dense space, the problem of sparse sign-in data of the interest points caused by excessive interest points is solved, the GPS point acquisition information integrated in the vector representation of the interest points is endowed with stronger semantic information of the interest points on the geographical positions, the defect that the interest points are lack of sufficient text description is overcome, and therefore the interest points are pushed more accurately by using the position point embedding matrix.
In a first aspect, an embodiment of the present specification provides a method for representing a point of interest, which is applied to a server, and is shown with reference to fig. 2, where the method for representing a point of interest includes the following steps:
s200, acquiring an interest point check-in record and a GPS (global positioning system) point collection record of the user set, wherein a plurality of position points in the GPS point collection record are overlapped with interest points in the interest point check-in record.
If the point-of-interest check-in records and the GPS point-taking records of the users are collected for a certain geographic area only in order to realize local point-of-interest pushing in the geographic area, the user set only contains the users checking in the point-of-interest in the geographic area based on the LBSN. If the point of interest check-in records and the GPS point-taking records of the users need to be collected aiming at different geographical areas in order to realize the remote point of interest push. The user set may include users who check in the point of interest based on the LBSN under the target application, and even include users who check in the point of interest based on the LBSN under multiple target applications.
Specifically, the point of interest check-in record of the user set includes: the method comprises the steps that a point of interest check-in record of each user in a user set comprises more than one piece of point of interest check-in data, and specifically, any user checks in a current point of interest based on LBSN, and one piece of point of interest check-in data is correspondingly generated. Each generated point of interest check-in data may be represented by a triple (u, v, t), i.e., indicating that user u visited point of interest v at time t.
In order to give the interest point more semantic information on the geographical position, the GPS acquisition point record of each user in the user set is merged into the embodiment of the present specification. Specifically, the GPS acquisition point record of each user is acquired by the mobile device where the target application is located. The GPS acquisition record of each user is a series of location points with geographical identifications acquired by the mobile device through GPS, and each location point may represent that the user u is at the location point l at the time t by a triplet (u, l, t). The location points acquired by the GPS may be points of interest or common location points.
For example, if U ═ { U ═ is used1,u2,…,unDenotes a user set, then L ═ L1,l2,…,lmThe position point set in the GPS acquisition point record corresponding to the user set U is represented, and V is { V ═ V }1,v2,…,vpAnd expressing the interest point set in the interest point check-in record corresponding to the user set U. And the position point set L comprises the interest point set V, and m is more than or equal to p.
S202, according to the interest point check-in record and the GPS point collecting record of the user set, a historical track sequence of each user in the user set is constructed.
Specifically, for each user in the user set, a historical track sequence of the user is constructed according to the interest point check-in record of the user and the GPS point collection record of the user. Specifically, the location points of the user in a certain historical time range T may be sorted according to time sequence, or the location points of a certain number M of wanders of the user may be sorted according to time sequence, so as to construct a historical track sequence of the user, for example: by using
And representing the historical track sequence of the user u, wherein the historical track sequence of the user u comprises a plurality of interest points and other position points which are accessed by the user u in a historical mode, and therefore the historical track sequence of each user in the user set can be obtained. Referring to the history track sequence in fig. 3, black blocks represent location points that are points of interest, and white blocks represent other location points.
S204, training a word expression model by using a historical track sequence of each user in the user set to obtain a position point embedding matrix, wherein the position point embedding matrix comprises position semantic vector expression of each interest point in the interest point check-in record and position semantic vector expression of each position point in the GPS point collection record.
Therefore, the position semantic vector representation of each interest point in the interest point check-in record and the position semantic vector representation of other position points (position points other than the interest points) are obtained.
Specifically, step S204 includes the following refinement steps: generating training sample data according to the historical track sequence of each user in the user set, wherein the training sample data comprises a training sample for each interest point in the interest point check-in record; training the word representation model according to the training sample data to obtain a trained word representation model; and obtaining a position point embedding matrix according to the trained hidden layer weight vector of the word representation model.
In the embodiment of the present specification, the word representation model may be, but is not limited to, a Skip-Gram model, and it should be noted that the Skip-Gram model is a three-layer neural network including an input layer, a hidden layer and an output layer, the hidden layer does not use an activation function, but the output layer uses a sotfmax function. In the embodiment of the description, a training sample for each interest point is generated by using a historical track sequence of each user in a user set to train the Skip-Gram model, and a weight matrix of a hidden layer in the Skip-Gram model obtained by training is a position point embedding matrix. And each row of hidden layer weight vectors in the position point embedding matrix respectively corresponds to the position semantic vector representation of one position point.
More specifically, the implementation process of generating training sample data according to the historical track sequence of each user in the user set is as follows:
selecting an ith interest point from a history track sequence of an nth user in a user set, wherein N is 1 to N, i is 1 to Z, N is the number of users in the user set, and Z is the number of interest points in the history track sequence of the nth user; determining a context position point of an ith interest point from the history track sequence of the nth user according to a preset context window, and constructing a sample according to the context position point of the ith interest point and the ith interest point; c interest points different from the ith interest point are selected, a corresponding negative sample is constructed according to the context position point of the ith interest point and the C interest points, and a positive sample is constructed according to the context position point of the ith interest point and the ith interest point; a set of training examples for the ith point of interest is formed based on the positive examples and the negative examples.
In a specific implementation process, the form of the training sample for the ith interest point may be represented as follows:
(li,context(li) And training a three-layer neural network through the training sample of each i interest point, predicting the context position point of the i interest point, and enabling the output probability of the context position point to be maximum.
A preset context window d and the number of neurons of the hidden layer k, which determine the vector dimension for representing the interest point to select the secondi the point of interest is the point of interest l0For example, the construction is directed to the point of interest l0The process of the set of training examples of (1) is described:
selecting a point of interest l02d front and back position points of the point of interest l0Is expressed as context (l)0) According to the point of interest l0Context with context location point (l)0) Constructing a sample;
because the data of the position points is usually too much, the training pressure needs to be reduced by adopting a negative sampling method, and therefore, the point l of interest is randomly selected0Different C interest points respectively form a point of interest l with front and back 2d position points according to the C interest points0Negative examples of (3). For example, the user a and the user B both access and sign in the same interest point, or the user a accesses the same interest point successively. Therefore, one or more sets of training examples may be obtained for each point of interest in the point of interest check-in record. Predict the point of interest l0The front and rear 2d position points of (2) are set so that the output probability of the front and rear 2d position points is maximized. Based on this, a probability P (context (l) is output0),l0) Output probability P (l) of 10),li) 0, i-1, 2, …, c, i.e. it is desired to maximize the probability:
in the above formula, σ denotes sigmoid function, context (l)0) Represents a point of interest l02d front and back position points of the front and back,
is the linear objective function of the neural network, w represents the weight, and θ is a parameter of the linear objective function.
Training the Skip-Gram model based on more than one group of training samples of each interest point, wherein the training of the Skip-Gram model is to obtain a hidden layer weight vector of the Skip-Gram model, and the hidden layer weight vector of the Skip-Gram model after the training is finished is used as a position point Embedding matrix (Location Embedding), and is represented as follows:
wherein W ═ W1,w2,…,wm)TEmbedding a matrix for the location points, including the hidden layer weight vector for each location point in the GPS acquisition point record, see FIG. 3, wi＝(wi1,wi2,…,wik)TThe hidden layer weight vector of the position point i is the position semantic vector representation of the position point i, m is the number of the position points, k is the vector dimension of the position points and corresponds to the number k of the neurons of the hidden layer.
For example, if the obtained location point embedding matrix is 10000 rows × 300 columns, and each row corresponds to a location semantic vector representation representing one location point, the location point embedding matrix of 10000 rows × 300 columns contains location semantic vector representations for 10000 location points (containing several interest points), and the location semantic vector representation for each location point is a 300-dimensional vector.
In a specific application scenario, the following are corresponding to the point of interest check-in record and the GPS acquisition record acquired in step S200: the obtained position point embedding matrix comprises position semantic vector representation of each position point in a target geographical area, for example, the target geographical area can be a city, a province or a city area; if it is necessary to push a different point of interest to the user in addition to pushing a local point of interest to the user, the location point embedding matrix includes a location semantic vector representation of each location point in a plurality of different geographic areas, for example, a location semantic vector representation of each location point in a national geographic area.
Therefore, the POI matrix sparseness problem caused by excessive interest points and sparse interest point check-in data can be solved through the Location point Embedding matrix (Location Embedding), and on the other hand, semantic information of the interest points on the geographic Location is given.
In a second aspect, an embodiment of the present disclosure provides a method for pushing a point of interest, which is applied to a server, and is used for pushing the point of interest to a user according to the location point embedding matrix obtained in the first aspect. Specifically, referring to fig. 4, the method for pushing a point of interest provided in the embodiment of the present specification includes the following steps:
s400: determining a target user, wherein the target user belongs to any user with interest point pushing requirements in the whole users.
Specifically, the target user may be any interest point push demand user in or out of the user set. For example, if the point of interest push method is applied to a target application, all users may register users for each of the target applications.
S402, pushing interest points to the target user according to the position point embedding matrix. Wherein, the position point embedding matrix is obtained by the interest point representation method of the first aspect.
Specifically, step S402 includes the following specific steps:
s4021, calculating push basis information aiming at a target user according to the position point embedding matrix, wherein the push basis information comprises any one of the following information aiming at the target user:
and (1) similarity information of interest points. Specifically, the interest point similarity information includes a similarity between each candidate interest point and the interest point where the target user is currently located. Specifically, the interest point similarity between each candidate interest point and the interest point where the target user is currently located can be obtained through the following steps a1 to a 4:
step A1, detecting the current interest point accessed by the target user; specifically, the current point of interest visited by the target user may be the point of interest that the target user currently checks in based on the LBSN.
Step A2, recalling a plurality of candidate interest points according to the current interest point. Specifically, each interest point except the current interest point in the position point embedding matrix may be used as a candidate interest point, or a part of interest points may be recalled as candidate interest points according to some preset rule, so as to reduce the amount of computation. Such as random recall or recall of the same type of points of interest, etc.
Step A3: and acquiring the position semantic vector representation of the current interest point and the position semantic vector representation of each candidate interest point from the position point embedding matrix. And according to the sequence identification of the current interest point, finding out a corresponding row hidden layer weight vector from the position point embedding matrix, and using the vector as a position semantic vector representation of the current interest point. For example, if the current interest point is the 100 th interest point in the interest point set, the position semantic vector of the 100 th interest point obtained from the position point embedding matrix is the position semantic vector representation of the current interest point. The manner of obtaining the semantic vector representation of the position of the candidate interest point is similar, and for the sake of brevity of the description, the details are not repeated herein.
Step A4: and aiming at each candidate interest point, obtaining the interest point similarity between the current interest point and the candidate interest point according to the position semantic vector representation of the current interest point and the position semantic vector representation of the candidate interest point. The similarity of the points of interest may refer to: the cosine distance between the location semantic vector representation of the candidate point of interest and the location semantic vector representation of the current point of interest.
② user similarity information. The user similarity information comprises the user similarity between each candidate user and the target user. Specifically, the user similarity information may be obtained through the following steps B1 to B3:
and step B1, acquiring the position semantic vector representation of each position point in the historical track sequence of the target user from the position point embedding matrix, and generating the track vector representation of the target user according to the position semantic vector representation of each position point in the historical track sequence of the target user.
Specifically, the trajectory vector representation of the target user is to: and carrying out average calculation on the position semantic vector representation of each position point in the historical track sequence of the target user.
Step B2, determining a plurality of candidate users aiming at the target user from the user set;
specifically, each user in the user set may be regarded as a candidate user. In order to reduce the amount of computation, a plurality of users associated with a target user are selected from a set of users as candidate users. Such as: selecting a friend of the target user under the target application as a candidate user, such as: a plurality of users belonging to the same age group as the target user are selected as candidate users, and the like. However, the selection manner for the candidate user is not limited to the above example.
Step B3, the user similarity between each candidate user and the target user is obtained by the following method: obtaining the position semantic vector representation of each position point in the historical track sequence of the candidate user from the position point embedding matrix; generating a track vector representation of the candidate user according to the position semantic vector representation of each position point in the historical track sequence of the candidate user; and obtaining the user similarity between the candidate user and the target user according to the track vector representation of the candidate user and the track vector representation of the target user.
Specifically, the trajectory vector representation of each candidate user is obtained by performing average calculation on the position semantic vector representation of each position point in the history trajectory sequence of the candidate user.
The method comprises the following steps of obtaining historical track similarity between a candidate user and a target user by calculating the cosine distance between the track vector representation of the candidate user and the track vector representation of the target user, wherein the historical track similarity obtained by calculation is the user similarity between the candidate user and the target user:
cosθijrepresenting the similarity of the historical tracks between the candidate users and the target users, and respectively enabling the track vector representation of the candidate users and the track vector representation of the target users to be avg (l)i)、avg(lj)。
Potential behavior tracks. Specifically, the potential behavior trace is a result of predicting a future behavior trace of the target user according to the historical trace sequence of the target user. Specifically, the potential behavior track of the target user can be predicted through the following steps C1-C3:
and step C1, obtaining the position semantic vector representation of each position point in the historical track sequence of the target user from the position point embedding matrix.
Specifically, the semantic vector representation of the position of each position point in the historical track sequence of the target user in a certain historical time range may be triggered and obtained according to a preset period or when the target user is detected to be currently located at a certain interest point
And step C2, obtaining a track vector matrix of the target user according to the position semantic vector representation of each position point in the historical track sequence of the target user.
And C3, predicting a potential behavior track of the target user according to the track vector matrix, wherein the potential behavior track comprises more than one interest point.
For example, a trajectory prediction model is obtained through a CNN (conditional Neural Networks) model or an RNN (Recurrent Neural Networks) model training, specifically, a trajectory vector matrix of the target user is used as an input of the trajectory prediction model, and the trajectory vector matrix of the target user is processed by using the trajectory prediction model to obtain a potential behavior trajectory of the target user.
S4022: and pushing the interest points to the target user according to the pushing basis information.
Specifically, the interest points pushed to the target user include one or more interest points. According to different information, the interest points pushed to the target user are correspondingly different, which is described below:
if the push basis information is the interest point similarity between each candidate interest point and the current interest point. Correspondingly, step S4022 specifically includes: selecting similar interest points with the current interest points from the candidate interest points according to the interest point similarity between each candidate interest point and the current interest point; and pushing the selected similar interest points to the target user.
Specifically, one or more interest points are selected as similar interest points according to the similarity of the interest points between each candidate interest point and the current interest point from high to low, for example, if the current interest point of the target user is a mall a, the similar interest point pushed to the target user may be a mall B. Thereby being capable of selecting other interest points similar to the current interest point to be pushed to the target user.
In the case that the push basis information is the implementation mode (i), it is required to use the semantic vector representation of the position point which is only the interest point embedded in the matrix, and the semantic vector representation of other position points can be ignored.
Through the vector correlation of different interest points, more similar interest points can be obtained, the limitation of regions is eliminated, and not only can local interest point push be realized, but also remote interest point push can be realized.
If the push basis information is the user similarity between each candidate user and the target user, correspondingly, step S4022 specifically includes: selecting a reference user from a plurality of candidate users according to the user similarity between each candidate user and the target user, wherein one or more candidate users can be selected as the reference user from high to low according to the user similarity; and pushing the current interest points accessed by the reference user to the target user.
By pushing the current interest points visited by the reference user to the target user, the interest points which are possibly interested by the target user can be mined, and the remote recommendation of the interest points can be realized.
And if the potential behavior track of the user according to the information target is pushed, the potential behavior track comprises one or more interest points. Therefore, step S4022 specifically is: determining potential interest points aiming at the target user from the potential behavior tracks; and pushing the potential interest points to the target user. And predicting potential interest points by combining a deep learning algorithm, thereby providing interest point pushing according with the historical behavior habits and interest preferences of users.
In a third aspect, based on the same inventive concept as the foregoing interest point representing method, an embodiment of the present specification provides an interest point representing apparatus, shown with reference to fig. 5, including:
a record obtaining unit 501, configured to obtain an interest point check-in record and a GPS collection point record of a user set, where a plurality of location points in the GPS collection point record coincide with interest points in the interest point check-in record;
the track construction unit 502 is used for constructing a historical track sequence of each user in the user set according to the interest point sign-in record and the GPS point collection record;
the model training unit 503 is configured to train a word representation model using the historical track sequence of each user in the user set to obtain a location point embedding matrix, where the location point embedding matrix includes a location semantic vector representation of each interest point in the interest point sign-in record and a location semantic vector representation of each location point in the GPS collection point record.
In an alternative embodiment, the model training unit 503 includes:
the training sample generating subunit is used for generating training sample data according to the historical track sequence of each user in the user set, wherein the training sample data comprises a training sample for each interest point in the interest point check-in record;
the training subunit is used for training the word representation model according to the training sample data to obtain a trained word representation model;
and the weight obtaining subunit is used for obtaining the position point embedded matrix according to the trained hidden layer weight vector of the word representation model.
In an optional implementation manner, the training sample generation subunit is specifically configured to:
selecting an ith interest point from the history track sequence of the nth user in the user set, wherein N is 1 to N, i is 1 to Z, N is the number of users in the user set, and Z is the number of interest points in the history track sequence of the nth user;
determining a context position point of the ith interest point from the history track sequence of the nth user according to a preset context window;
c interest points different from the ith interest point are selected, a corresponding negative sample is constructed according to the context position point of the ith interest point and the C interest points, and a positive sample is constructed according to the context position point of the ith interest point and the ith interest point;
a set of training examples for the ith point of interest is formed based on the positive examples and the negative examples.
In a fourth aspect, an embodiment of the present specification provides a point of interest pushing apparatus, which is shown in fig. 6, and includes:
a determining unit 601, configured to determine a target user, where the target user belongs to any user with an interest point pushing requirement in all users;
the interest point pushing module 602 is configured to push an interest point to a target user according to the location point embedding matrix, where the location point embedding matrix is obtained by using the foregoing interest point representation method embodiment, and details are not described here for brevity of the description.
Under an optional implementation manner, the point of interest pushing module 602 includes:
the information calculating unit 6021 is configured to obtain push basis information for the target user according to the location point embedding matrix, where the push basis information includes one of the following information for the target user: interest point similarity information, user similarity information and potential behavior tracks;
the pushing unit 6022 is configured to push the interest point to the target user according to the pushing basis information.
In an optional implementation, the information calculating unit 6021 includes:
the detection subunit is used for detecting the current interest point accessed by the target user;
the recalling subunit is used for recalling a plurality of candidate interest points according to the current interest point;
a first vector obtaining subunit, configured to obtain, from the location point embedding matrix, a location semantic vector representation of the current interest point, and a location semantic vector representation of each candidate interest point of the multiple candidate interest points;
and the computing subunit is used for obtaining the interest point similarity between the current interest point and the candidate interest point according to the position semantic vector representation of the current interest point and the position semantic vector representation of the candidate interest point aiming at each candidate interest point in the candidate interest points.
In an optional embodiment, the pushing unit 6022 includes:
the interest point selecting subunit is used for selecting similar interest points of the current interest points from the candidate interest points according to the interest point similarity between each candidate interest point and the current interest point;
and the first pushing subunit is used for pushing the similar interest points to the target user.
In an alternative embodiment, the information calculating unit 6021 includes:
the track expression subunit is used for acquiring the position semantic vector expression of each position point in the historical track sequence of the target user from the position point embedding matrix and generating the track vector expression of the target user according to the position semantic vector expression of each position point in the historical track sequence of the target user;
a user determination subunit, configured to determine a plurality of candidate users for a target user from a user set;
the similarity operator unit is used for acquiring the position semantic vector representation of each position point in the historical track sequence of the candidate users from the position point embedding matrix aiming at each candidate user; generating track vector representation of the candidate user according to the position semantic vector representation of each position point in the historical track sequence of the candidate user; and according to the track vector representation of the candidate user and the track vector representation of the target user, obtaining the user similarity between the candidate user and the target user.
In an alternative embodiment, the pushing unit 6022 comprises:
the user selecting subunit is used for selecting a reference user from the candidate users according to the user similarity between each candidate user and the target user;
and the second pushing subunit is used for pushing the current interest point accessed by the reference user to the target user.
In an optional implementation, the information calculating unit 6021 includes:
the second vector acquisition subunit is used for acquiring the position semantic vector representation of each position point in the historical track sequence of the target user from the position point embedding matrix;
the track obtaining subunit is used for obtaining a track vector matrix for the target user according to the position semantic vector representation of each position point in the historical track sequence of the target user;
and the track prediction subunit is used for predicting a potential behavior track of the target user according to the track vector matrix, wherein the potential behavior track comprises more than one interest point.
In an optional embodiment, the pushing unit 6022 includes:
the interest point determining subunit is used for determining potential interest points aiming at the target user from the potential behavior track;
and the third pushing subunit is used for pushing the potential interest points to the target user.
In a fifth aspect, based on the same inventive concept as that of the foregoing point of interest representing method and point of interest pushing method embodiments, an embodiment of the present specification further provides an electronic device, as shown in fig. 7, including a memory 704, a processor 702, and a computer program stored on the memory 704 and being executable on the processor 702, where the processor 702 implements the steps of the foregoing point of interest representing method embodiments when executing the program, or implements the steps of the foregoing point of interest pushing method embodiments.
Where in fig. 7 a bus architecture (represented by bus 700) is shown, bus 700 may include any number of interconnected buses and bridges, and bus 700 links together various circuits including one or more processors, represented by processor 702, and memory, represented by memory 704. The bus 700 may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. A bus interface 706 provides an interface between the bus 700 and the receiver 701 and transmitter 703. The receiver 701 and the transmitter 703 may be the same element, i.e., a transceiver, providing a means for communicating with various other apparatus over a transmission medium. The processor 702 is responsible for managing the bus 700 and general processing, and the memory 704 may be used for storing data used by the processor 702 in performing operations.
In a sixth aspect, based on the same inventive concept as the foregoing point of interest representation method and point of interest pushing method embodiments, this specification embodiment further provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the foregoing point of interest representation method embodiments or implements the steps of the foregoing point of interest pushing method embodiments.
The description has been presented with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the description. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.
It will be understood that the invention is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is defined only by the appended claims, which are not intended to limit the present invention, and any modifications, equivalents, improvements, etc. made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (24)
1. A point of interest representation method, comprising:
acquiring an interest point check-in record and a GPS (global positioning system) point acquisition record of a user set, wherein a plurality of position points in the GPS point acquisition record are overlapped with interest points in the interest point check-in record;
according to the interest point check-in record and the GPS point collecting record, constructing a historical track sequence of each user in the user set;
and training a word expression model by using the historical track sequence of each user in the user set to obtain a position point embedding matrix, wherein the position point embedding matrix comprises position semantic vector expression of each interest point in the interest point check-in record and position semantic vector expression of each position point in the GPS point collection record.
2. The method of claim 1, wherein the training of the word representation model using the historical track sequence of each user in the set of users to obtain the location point embedding matrix comprises:
generating training sample data according to a historical track sequence of each user in a user set, wherein the training sample data comprises a training sample for each interest point in the interest point check-in record;
training a word representation model according to the training sample data to obtain a trained word representation model;
and obtaining the position point embedded matrix according to the hidden layer weight vector of the trained word expression model.
3. The method of claim 2, wherein generating training sample data according to the historical track sequence of each user in the set of users comprises:
selecting an ith interest point from the history track sequence of the nth user in the user set, wherein N is 1 to N, i is 1 to Z, N is the number of users in the user set, and Z is the number of interest points in the history track sequence of the nth user;
determining a context position point of the ith interest point from the history track sequence of the nth user according to a preset context window;
c interest points different from the ith interest point are selected, a corresponding negative sample is constructed according to the context position point of the ith interest point and the C interest points, and a positive sample is constructed according to the context position point of the ith interest point and the ith interest point;
forming a set of training examples for the ith point of interest based on the positive examples and the negative examples.
4. A point of interest pushing method comprises the following steps:
determining a target user, wherein the target user belongs to any user with interest point pushing requirements in all users;
and pushing interest points to the target user according to a position point embedding matrix, wherein the position point embedding matrix is obtained by the interest point representation method of any one of claims 1 to 3.
5. The method of claim 4, the pushing of points of interest to the target user according to the location point embedding matrix, comprising:
obtaining push basis information aiming at the target user according to the position point embedding matrix, wherein the push basis information comprises the following information aiming at the target user: interest point similarity information, user similarity information and potential behavior tracks;
and pushing the interest points to the target user according to the pushing basis information.
6. The method of claim 5, wherein the obtaining push dependency information for the target user according to the location point embedding matrix comprises:
detecting a current interest point accessed by the target user;
recalling a plurality of candidate interest points according to the current interest point;
obtaining a position semantic vector representation of the current interest point and a position semantic vector representation of each candidate interest point in the plurality of candidate interest points from the position point embedding matrix;
and aiming at each candidate interest point in the candidate interest points, obtaining the interest point similarity between the current interest point and the candidate interest point according to the position semantic vector representation of the current interest point and the position semantic vector representation of the candidate interest point.
7. The method of claim 6, the pushing points of interest to the target user according to the push-by-information, comprising:
selecting a similar interest point of the current interest point from the candidate interest points according to the interest point similarity between each candidate interest point and the current interest point;
and pushing the similar interest points to the target user.
8. The method of claim 5, wherein the obtaining push dependency information for the target user according to the location point embedding matrix comprises:
acquiring the position semantic vector representation of each position point in the historical track sequence of the target user from the position point embedding matrix, and generating the track vector representation of the target user according to the position semantic vector representation of each position point in the historical track sequence of the target user;
determining a plurality of candidate users for the target user from the set of users;
for each candidate user, acquiring a position semantic vector representation of each position point in the historical track sequence of the candidate user from the position point embedding matrix; generating a track vector representation of the candidate user according to the position semantic vector representation of each position point in the historical track sequence of the candidate user; and obtaining the user similarity between the candidate user and the target user according to the track vector representation of the candidate user and the track vector representation of the target user.
9. The method of claim 8, the pushing points of interest to the target user according to the push-by-information, comprising:
selecting a reference user from the candidate users according to the user similarity between each candidate user and the target user;
and pushing the current interest points accessed by the reference user to the target user.
10. The method of claim 5, wherein the obtaining push dependency information for the target user according to the location point embedding matrix comprises:
obtaining the position semantic vector representation of each position point in the historical track sequence of the target user from the position point embedding matrix;
obtaining a track vector matrix aiming at the target user according to the position semantic vector representation of each position point in the historical track sequence of the target user;
and predicting a potential behavior track of the target user according to the track vector matrix, wherein the potential behavior track comprises more than one interest point.
11. The method of claim 10, said pushing points of interest to the target user according to the push-by-information, comprising:
determining potential interest points aiming at the target user from the potential behavior tracks;
and pushing the potential interest points to the target user.
12. A point of interest representation apparatus comprising:
the system comprises a record acquisition unit, a GPS acquisition unit and a control unit, wherein the record acquisition unit is used for acquiring an interest point check-in record of a user set and a GPS acquisition point record, and a plurality of position points in the GPS acquisition point record are overlapped with interest points in the interest point check-in record;
the track construction unit is used for constructing a historical track sequence of each user in the user set according to the interest point check-in record and the GPS point collection record;
and the model training unit is used for training a word representation model by using the historical track sequence of each user in the user set to obtain a position point embedding matrix, wherein the position point embedding matrix comprises position semantic vector representation of each interest point in the interest point check-in record and position semantic vector representation of each position point in the GPS point collection record.
13. The apparatus of claim 12, the model training unit, comprising:
a training sample generating subunit, configured to generate training sample data according to a historical track sequence of each user in a user set, where the training sample data includes a training sample for each point of interest in the point of interest check-in record;
the training subunit is used for training the word representation model according to the training sample data to obtain a trained word representation model;
and the weight obtaining subunit is used for obtaining the position point embedded matrix according to the weight vector of the hidden layer of the trained word expression model.
14. The apparatus of claim 13, the training sample generation subunit being specifically configured to:
selecting an ith interest point from the history track sequence of the nth user in the user set, wherein N is 1 to N, i is 1 to Z, N is the number of users in the user set, and Z is the number of interest points in the history track sequence of the nth user;
determining a context position point of the ith interest point from the history track sequence of the nth user according to a preset context window;
c interest points different from the ith interest point are selected, a corresponding negative sample is constructed according to the context position point of the ith interest point and the C interest points, and a positive sample is constructed according to the context position point of the ith interest point and the ith interest point;
forming a set of training examples for the ith point of interest based on the positive examples and the negative examples.
15. A point of interest push apparatus, comprising:
the system comprises a determining unit, a sending unit and a receiving unit, wherein the determining unit is used for determining a target user, and the target user belongs to any user with interest point pushing requirements in all users;
an interest point pushing module, configured to push an interest point to the target user according to a location point embedding matrix, where the location point embedding matrix is obtained by using the interest point representing method according to any one of claims 1 to 3.
16. The apparatus of claim 15, the point of interest push module comprising:
an information calculating unit, configured to obtain push basis information for the target user according to the location point embedding matrix, where the push basis information includes one of the following information for the target user: interest point similarity information, user similarity information and potential behavior tracks;
and the pushing unit is used for pushing the interest points to the target user according to the pushing basis information.
17. The apparatus of claim 16, the information calculating unit, comprising:
the detection subunit is used for detecting the current interest point accessed by the target user;
the recalling subunit is used for recalling a plurality of candidate interest points according to the current interest point;
a first vector obtaining subunit, configured to obtain, from the location point embedding matrix, a location semantic vector representation of the current interest point and a location semantic vector representation of each candidate interest point in the multiple candidate interest points;
and the computing subunit is used for obtaining the interest point similarity between the current interest point and each candidate interest point in the candidate interest points according to the position semantic vector representation of the current interest point and the position semantic vector representation of the candidate interest point.
18. The apparatus of claim 17, the pushing unit comprising:
an interest point selecting subunit, configured to select, according to interest point similarity between each candidate interest point and the current interest point, a similar interest point of the current interest point from the multiple candidate interest points;
and the first pushing subunit is used for pushing the similar interest points to the target user.
19. The apparatus of claim 16, the information calculating unit comprising:
the track expression subunit is used for acquiring the position semantic vector expression of each position point in the historical track sequence of the target user from the position point embedding matrix, and generating the track vector expression of the target user according to the position semantic vector expression of each position point in the historical track sequence of the target user;
a user determination subunit, configured to determine, from the user set, a plurality of candidate users for the target user;
the similarity operator unit is used for acquiring the position semantic vector representation of each position point in the historical track sequence of the candidate users from the position point embedding matrix aiming at each candidate user; generating a track vector representation of the candidate user according to the position semantic vector representation of each position point in the historical track sequence of the candidate user; and obtaining the user similarity between the candidate user and the target user according to the track vector representation of the candidate user and the track vector representation of the target user.
20. The apparatus of claim 19, the pushing unit comprising:
a user selecting subunit, configured to select a reference user from the multiple candidate users according to a user similarity between each candidate user and the target user;
and the second pushing subunit is used for pushing the current interest point accessed by the reference user to the target user.
21. The apparatus of claim 16, the information calculating unit, comprising:
the second vector acquisition subunit is used for acquiring the position semantic vector representation of each position point in the historical track sequence of the target user from the position point embedding matrix;
the track obtaining subunit is used for obtaining a track vector matrix for the target user according to the position semantic vector representation of each position point in the historical track sequence of the target user;
and the track prediction subunit is used for predicting a potential behavior track of the target user according to the track vector matrix, wherein the potential behavior track comprises more than one interest point.
22. The apparatus of claim 21, the pushing unit comprising:
an interest point determining subunit, configured to determine, from the potential behavior trajectory, a potential interest point for the target user;
and the third pushing subunit is used for pushing the potential interest points to the target user.
23. 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 steps of the method of any of claims 1-11 when executing the program.
24. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010624873.5A CN111782955A (en) | 2020-07-01 | 2020-07-01 | Interest point representing and pushing method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010624873.5A CN111782955A (en) | 2020-07-01 | 2020-07-01 | Interest point representing and pushing method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111782955A true CN111782955A (en) | 2020-10-16 |
Family
ID=72757829
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010624873.5A Pending CN111782955A (en) | 2020-07-01 | 2020-07-01 | Interest point representing and pushing method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111782955A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569149A (en) * | 2021-07-30 | 2021-10-29 | 拉扎斯网络科技（上海）有限公司 | Information processing method and device and electronic equipment |
| CN113722605A (en) * | 2021-11-03 | 2021-11-30 | 北京奇岱松科技有限公司 | Method and system for calculating real-time interest information |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104123398A (en) * | 2014-08-15 | 2014-10-29 | 百度在线网络技术（北京）有限公司 | Information pushing method and device |
| CN110110244A (en) * | 2019-04-26 | 2019-08-09 | 广东技术师范大学 | A kind of point of interest recommended method merging multi-source information |
| CN110929162A (en) * | 2019-12-04 | 2020-03-27 | 腾讯科技（深圳）有限公司 | Recommendation method and device based on interest points, computer equipment and storage medium |
| CN111143676A (en) * | 2019-12-26 | 2020-05-12 | 斑马网络技术有限公司 | Interest point recommendation method and device, electronic equipment and computer-readable storage medium |
| CN111177565A (en) * | 2019-12-31 | 2020-05-19 | 杭州电子科技大学 | Interest point recommendation method based on correlation matrix and word vector model |
-
2020
- 2020-07-01 CN CN202010624873.5A patent/CN111782955A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104123398A (en) * | 2014-08-15 | 2014-10-29 | 百度在线网络技术（北京）有限公司 | Information pushing method and device |
| CN110110244A (en) * | 2019-04-26 | 2019-08-09 | 广东技术师范大学 | A kind of point of interest recommended method merging multi-source information |
| CN110929162A (en) * | 2019-12-04 | 2020-03-27 | 腾讯科技（深圳）有限公司 | Recommendation method and device based on interest points, computer equipment and storage medium |
| CN111143676A (en) * | 2019-12-26 | 2020-05-12 | 斑马网络技术有限公司 | Interest point recommendation method and device, electronic equipment and computer-readable storage medium |
| CN111177565A (en) * | 2019-12-31 | 2020-05-19 | 杭州电子科技大学 | Interest point recommendation method based on correlation matrix and word vector model |
Non-Patent Citations (1)
| Title |
|---|
| SHANSHAN FENG 等: "POI2Vec: Geographical Latent Representation for Predicting Future Visitors", THIRTY-FIRST AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, 10 February 2017 (2017-02-10), pages 103 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569149A (en) * | 2021-07-30 | 2021-10-29 | 拉扎斯网络科技（上海）有限公司 | Information processing method and device and electronic equipment |
| CN113569149B (en) * | 2021-07-30 | 2024-01-19 | 拉扎斯网络科技（上海）有限公司 | Information processing method and device and electronic equipment |
| CN113722605A (en) * | 2021-11-03 | 2021-11-30 | 北京奇岱松科技有限公司 | Method and system for calculating real-time interest information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110929162B (en) | Recommendation method and device based on interest points, computer equipment and storage medium | |
| CN106462627B (en) | Analyzing semantic places and related data from multiple location data reports | |
| Xu et al. | A survey for mobility big data analytics for geolocation prediction | |
| CN105532030B (en) | For analyzing the devices, systems, and methods of the movement of target entity | |
| CN110119475B (en) | POI recommendation method and system | |
| CN108875007A (en) | The determination method and apparatus of point of interest, storage medium, electronic device | |
| Liu et al. | A two-stage destination prediction framework of shared bicycles based on geographical position recommendation | |
| US20150005901A1 (en) | Iterative learning for reliable sensor sourcing systems | |
| CN106767835B (en) | Positioning method and device | |
| CN112035591A (en) | Road network matching method, device, equipment and storage medium | |
| CN112199609A (en) | POI recommendation system and method for sensing space-time semantic interval under self-attention | |
| CN111782955A (en) | Interest point representing and pushing method and device, electronic equipment and storage medium | |
| Willekens | Evidence-based monitoring of international migration flows in Europe | |
| Tang et al. | Discovering travel community for POI recommendation on location-based social networks | |
| JP2014203191A (en) | Information providing system, information request terminal, information processing system, control method, and program | |
| CN113139140A (en) | Tourist attraction recommendation method based on space-time perception GRU and combined with user relation preference | |
| Steiger et al. | Research on social media feeds–A GIScience perspective | |
| EP3192061B1 (en) | Measuring and diagnosing noise in urban environment | |
| Zhou et al. | Spatial–Temporal Federated Transfer Learning with multi-sensor data fusion for cooperative positioning | |
| Shi et al. | An adaptive approach for modelling the movement uncertainty in trajectory data based on the concept of error ellipses | |
| CN117435819A (en) | Method, system and storage medium for recommending interest points through space-time dynamic perception | |
| Alia et al. | On the exploitation of GPS-based data for real-time visualisation of pedestrian dynamics in open environments | |
| CN106779181A (en) | Method is recommended by a kind of medical institutions based on linear regression factor Non-negative Matrix Factorization model | |
| CN113918802B (en) | Navigation method, device, equipment, medium and product | |
| CN111782973A (en) | Interest point state prediction method and device, electronic equipment and 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 | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 40039073 Country of ref document: HK |