A lightweight point of interest recommendation method and device integrating user movement direction

Abstract

The present invention discloses a lightweight point of interest recommendation method and device integrating user movement direction, which relates to the technical field of point of interest recommendation, and its main purpose is to solve the problem of low data processing efficiency for point of interest recommendation. It mainly includes obtaining the real-time location, movement direction and global points of interest of the target user in response to the point of interest recommendation demand of the target user, wherein the global point of interest is the location matching the point of interest recommendation demand in the global location check-in record data set; screening the global points of interest according to the real-time location and the check-in records of each point of interest to obtain the candidate check-in record data set; calculating the moving direction similarity of each candidate point of interest according to the moving direction and the real-time location, and calculating the collaborative filtering similarity of each candidate point of interest according to the check-in user information and the check-in record of the target user; generating the point of interest recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity. It is mainly used for point of interest recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of interest point recommendation, and in particular to a lightweight interest point recommendation method and device integrating user movement direction. Background Art

[0002] The advancement of mobile computing, wireless communication and location acquisition technology has greatly promoted the popularity and development of location-based social networks (LBSNs), such as X (formerly Twitter), WeChat, Foursquare and Facebook. POI recommendation for LBSNs can provide mobile users with diverse, personalized and never-visited places, thereby effectively alleviating the choice confusion caused by information overload for users, helping to improve users' experience in social networks and real life, and also helping businesses to tap potential customers for advertising push. Therefore, POI recommendation is an extension of traditional recommendation technology in social networks, and has become a hot topic in current research, attracting extensive attention from many scholars.

[0003] The key issue in the study of POI recommendation is how to mine the implicit multi-dimensional feature factors from the historical check-in data of massive LBSNs and recommend POIs based on these feature factors. Existing POI recommendation algorithms mainly make recommendations based on all users and all locations (for example, Brightkite has 50,687 users and 702,401 locations). The huge amount of data causes a large amount of computation in the recommendation operation process and an excessive computational load on the recommendation system, which in turn affects the data processing efficiency of POI recommendation. Summary of the invention

[0004] In view of this, the present invention provides a lightweight POI recommendation method and device integrating user movement direction, the main purpose of which is to solve the problem of low data processing efficiency of existing POI recommendation.

[0005] According to one aspect of the present invention, a lightweight POI recommendation method integrating user movement direction is provided, comprising:

[0006] In response to the target user's point of interest recommendation requirement, obtain the target user's real-time location, movement direction, and global points of interest, wherein the global points of interest are locations in the global location check-in record data set that match the point of interest recommendation requirement;

[0007] The global points of interest are screened according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet check-in and distance conditions and check-in user information of the candidate points of interest;

[0008] Calculating the similarity of the moving directions of the candidate points of interest based on the moving directions and the real-time positions, and calculating the collaborative filtering similarity of the candidate points of interest based on the check-in user information and the check-in record of the target user;

[0009] Generate interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

[0010] Furthermore, obtaining the moving direction of the target user includes:

[0011] According to a preset sampling frequency and a preset sampling duration, raw compass data of the mobile terminal corresponding to the target user is collected to obtain a one-dimensional angle sequence;

[0012] One-dimensional median filtering is performed on the one-dimensional angle sequence according to a preset filtering window length, and the median of the one-dimensional angle sequence is used as the moving direction of the target user.

[0013] Furthermore, the global points of interest are screened according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, including:

[0014] Calculating a longitude and latitude screening range of points of interest based on the real-time position and the preset distance range, and screening out preliminary screening points of interest that meet the longitude and latitude screening range of points of interest from the global points of interest;

[0015] Counting the global user historical check-in times of each of the preliminarily screened interest points, and determining the preliminarily screened interest points whose historical check-in times are greater than or equal to a preset check-in times threshold as candidate interest points;

[0016] A candidate check-in record data set is generated according to the candidate points of interest and the historical check-in record data of the candidate points of interest.

[0017] Furthermore, the step of calculating the similarity of the moving directions of any of the candidate interest points comprises:

[0018] According to the latitude and longitude coordinates of the real-time position and the latitude and longitude coordinates of the candidate point of interest, using the tangent theorem of a right triangle, a first angle between a line connecting the real-time position of the target user and the candidate point of interest and a preset direction is calculated;

[0019] Calculate a third angle between a line connecting the real-time position of the target user to the candidate point of interest and the moving direction based on the first angle and a second angle between the moving direction and the preset direction;

[0020] According to the third angle, the direction similarity calculation formula is used to calculate the moving direction similarity of the candidate interest point, wherein the direction similarity calculation formula is: ;in, is the direction similarity, is the third angle, Indicates real-time location. Candidate points of interest.

[0021] Furthermore, the calculating the collaborative filtering similarity of each of the candidate interest points based on the check-in user information and the check-in record of the target user includes:

[0022] Calculating the user similarity between each associated user in the signed-in user information and the target user respectively;

[0023] Adding up all the user similarities to obtain a global user similarity value;

[0024] For each candidate point of interest, the sum of the user similarities between the associated users who checked in to the candidate point of interest and the target user is calculated to obtain the single-point user similarity and value of different candidate points of interest, and the ratio of the single-point user similarity and value to the global user similarity and value is calculated to obtain the collaborative filtering similarity.

[0025] Furthermore, the step of calculating the user similarity between any of the associated users and the target user includes:

[0026] Calculate the number of candidate interest points that the associated user and the target user check in at together to obtain the number of intersection interest points;

[0027] Calculate the average of the number of candidate points of interest that the associated user has checked in and the number of candidate points of interest that the target user has checked in to obtain the average number of points of interest check-in;

[0028] The quotient of the number of the intersection points of interest and the average check-in value of the points of interest is calculated to obtain the user similarity.

[0029] Furthermore, generating the interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity includes:

[0030] For each of the candidate interest points, linearly fitting the moving direction similarity and the collaborative filtering similarity is performed respectively to obtain a recommendation value of each of the candidate interest points;

[0031] The candidate interest points are sorted according to the recommendation values, and interest point recommendation information for the target user is generated based on the candidate interest points whose order numbers meet a preset recommendation number.

[0032] According to another aspect of the present invention, a lightweight POI recommendation device integrating user movement direction is provided, comprising:

[0033] An acquisition module, configured to acquire the real-time location and movement direction of the target user in response to the target user's interest point recommendation requirement, wherein the global interest point is a location in the global location check-in record data set that matches the interest point recommendation requirement;

[0034] a screening module, configured to screen the global points of interest according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet check-in and distance conditions and check-in user information of the candidate points of interest;

[0035] A calculation module, configured to calculate the similarity of the moving directions of the candidate points of interest based on the moving directions and the real-time positions, and to calculate the collaborative filtering similarity of the candidate points of interest based on the check-in user information and the check-in record of the target user;

[0036] A generating module is used to generate the interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

[0037] According to another aspect of the present invention, a storage medium is provided, in which at least one executable instruction is stored, and the executable instruction enables a processor to perform operations corresponding to the above-mentioned lightweight point of interest recommendation method integrating user movement direction.

[0038] According to another aspect of the present invention, there is provided a terminal, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0039] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned lightweight interest point recommendation method integrating the user movement direction.

[0040] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0041] The present invention provides a lightweight point of interest recommendation method and device integrating user movement direction. The embodiment of the present invention obtains the real-time location, movement direction and global points of interest of the target user in response to the point of interest recommendation demand of the target user, wherein the global points of interest are places in the global place check-in record data set that match the point of interest recommendation demand; the global points of interest are screened according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet the check-in and distance conditions and the check-in user information of the candidate points of interest; according to the movement direction and the real-time position, calculate the similarity of the moving direction of each of the candidate points of interest, and calculate the collaborative filtering similarity of each of the candidate points of interest based on the signed-in user information and the sign-in record of the target user; generate the point of interest recommendation information of the target user based on the moving direction similarity and the collaborative filtering similarity, avoid calculating the recommendation value of the global points of interest, greatly reduce the data processing amount of the point of interest recommendation process, and realize lightweight recommendation. At the same time, the point of interest recommendation is performed by integrating the moving direction features, enriching the recommended feature information, ensuring the accuracy of the recommendation, thereby greatly improving the data processing efficiency of the point of interest recommendation.

[0042] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0044] Figure 1 A flowchart of a lightweight point of interest recommendation method integrating user movement direction provided by an embodiment of the present invention is shown;

[0045] Figure 2 A schematic diagram showing the angle relationship between the position direction and the moving direction of a point of interest provided by an embodiment of the present invention is shown;

[0046] Figure 3 A block diagram showing a lightweight point of interest recommendation device integrating user movement direction provided by an embodiment of the present invention;

[0047] Figure 4 A schematic diagram of the structure of a terminal provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0048] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0049] In view of the low data processing efficiency of existing POI recommendation, the present invention provides a lightweight POI recommendation method integrating user movement direction, such as Figure 1 As shown, the method includes:

[0050] 101. In response to a target user's demand for recommendation of points of interest, obtain the target user's real-time location, moving direction, and global points of interest.

[0051] In an embodiment of the present invention, a point of interest recommendation system generates relevant recommendation information according to the needs of the target user. The current execution subject is a server of the point of interest recommendation system, which can be a cloud server or a local server. The point of interest recommendation system can be an application that specifically provides recommendation services, or it can be an application involving point of interest recommendation scenarios such as catering, entertainment, shopping, and tourism, which is not specifically limited in the embodiment of the present invention. When the target user triggers the point of interest recommendation function through the corresponding application, the current execution subject obtains the real-time location and moving direction of the target user through the mobile terminal device of the target user, and obtains the global point of interest according to the point of interest recommendation needs entered by the target user. Among them, the mobile terminal device is a mobile portable device with a GPS positioning function and an electronic compass sensor, such as a smart phone, a tablet computer, a smart wearable device, etc. The real-time location can be obtained based on the GPS positioning function, and the moving direction can be determined based on the deflection angle reading collected by the electronic compass sensor.

[0052] It should be noted that the global points of interest are places in the global place check-in record dataset that match the point of interest recommendation requirements. The global place check-in record dataset includes the historical check-in record data of all users in the social network for all places. The historical check-in record data includes the basic information of the check-in users, the type and location information of the check-in place, the check-in time and number of check-ins of the same user in the historical time period. The point of interest recommendation requirements can include the types of places of interest, such as catering, entertainment, shopping, tourism, culture, and the names of places of interest, such as CCB ATM, a certain restaurant, etc.

[0053] In an embodiment of the present invention, for further explanation and limitation, obtaining the moving direction of the target user includes:

[0054] According to a preset sampling frequency and a preset sampling duration, raw compass data of the mobile terminal corresponding to the target user is collected to obtain a one-dimensional angle sequence;

[0055] One-dimensional median filtering is performed on the one-dimensional angle sequence according to a preset filtering window length, and the median of the one-dimensional angle sequence is used as the moving direction of the target user.

[0056] Normally, the azimuth range of the electronic compass in a mobile terminal device is 0-359°, which represents the angle between the user's moving direction and the direction of the geomagnetic North Pole. When a pedestrian walks toward due north, the reading is 0°, due east is 90°, due south is 180°, and due west is 270°. However, the angle value fluctuates over time, especially during adjacent sampling times, when there will be significant angle changes. Factors that lead to the above phenomenon include irregular walking of users, habitual swinging back and forth, the influence of the surrounding magnetic field, and the internal bias of the sensor during the movement of a handheld smartphone. The influence of these factors leads to an error between the electronic compass reading and the actual azimuth, reducing the calculation accuracy of the electronic compass. In an embodiment of the present invention, in order to ensure the accuracy of the user's moving direction, the original electronic compass data is subjected to noise reduction processing to improve the accuracy of the moving direction. According to the preset acquisition frequency, the original compass data within the preset time length is collected to form a one-dimensional angle sequence that changes with time, and the median filtering method is used to perform noise reduction processing on the one-dimensional angle sequence. With the preset sampling frequency of 1Hz and the preset sampling time of a One-dimensional angle sequence collected in seconds Take the one-dimensional median filtering process as an example. Set the preset filter window length to b ( b is an odd number), extracted from the sequence b Number: in, is the window center value, , and then Arrange them by numerical value and take the median value as the moving direction of the target user. Set the median value to , The calculation formula is:

[0057] (1);

[0058] in, Indicates taking the middle value. The above-mentioned preset sampling frequency, preset sampling time and preset filter window length can be customized according to actual application requirements, and are not specifically limited in the embodiment of the present invention. By performing noise reduction processing on the original compass data, the accuracy of the moving direction can be improved, thereby ensuring the calculation accuracy of the moving direction similarity of the subsequent candidate interest points.

[0059] 102. Screen the global points of interest according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set.

[0060] The problem of low computational efficiency and accuracy caused by the participation of all locations in the calculation. In an embodiment of the present invention, global points of interest are screened based on the real-time location of the target user and the historical check-in records of different points of interest. From the distance dimension, points of interest that are within the user's tolerance distance relative to the current target user are found. From the popularity dimension of the points of interest, the historical check-in records of the points of interest by global users are used to filter out points of interest where few users have checked in. That is, the candidate check-in record data set includes candidate points of interest that meet the check-in and distance conditions and the check-in user information of the candidate points of interest. By filtering the global points of interest from the distance dimension from the target user and the popularity dimension of the points of interest, the number of points of interest participating in the subsequent point of interest recommendation calculation can be greatly reduced, and the system's operating resource consumption can be reduced, thereby achieving lightweight point of interest recommendation and improving the data processing efficiency of point of interest recommendation.

[0061] In one embodiment of the present invention, for further explanation and limitation, the global points of interest are screened according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, including:

[0062] Calculating a longitude and latitude screening range of points of interest based on the real-time position and the preset distance range, and screening out preliminary screening points of interest that meet the longitude and latitude screening range of points of interest from the global points of interest;

[0063] Counting the global user historical check-in times of each of the preliminarily screened interest points, and determining the preliminarily screened interest points whose historical check-in times are greater than or equal to a preset check-in times threshold as candidate interest points;

[0064] A candidate check-in record data set is generated according to the candidate points of interest and the historical check-in record data of the candidate points of interest.

[0065] In the embodiment of the present invention, the current real-time location of the target user is set as location U( , ) ,in, Indicates real-time location The longitude, Indicates real-time location In order to filter the global points of interest based on the distance relative to the target user, the longitude and latitude screening range of the points of interest is calculated according to the preset tolerance distance, so as to filter out the points of interest whose corresponding longitude and latitude are outside the longitude and latitude screening range of the points of interest, and determine the points of interest within the longitude and latitude screening range of the points of interest as the initial screening points of interest. Among them, the derivation process of the longitude and latitude screening range of the points of interest includes: according to the distance between any point of interest and the real-time location The calculation formula to establish the relationship between the longitude and latitude of the two locations is expressed as:

[0066]

[0067] in, is the average radius of the Earth (6371 km), Candidate points of interest The longitude coordinates of Points of interest The latitude coordinate of . The above distance If the preset tolerance distance needs to be met, the preset tolerance distance can be used as the maximum , calculate the longitude and latitude screening range of the point of interest. Taking 20 km as an example, according to formula (2), it can be calculated that the longitude difference between the longitude boundary of the longitude and latitude screening range of the point of interest and the real-time position is less than 0.18, and the latitude difference between the latitude boundary of the longitude and latitude screening range of the point of interest and the real-time position is also less than 0.18. Then the longitude and latitude screening range of the point of interest can be expressed as:

[0068] (3);

[0069] (4).

[0070] After the global points of interest are screened based on the distance and the preliminary screened points of interest are obtained, the global user historical check-in times of each preliminary screened point of interest are continued to be screened. If the historical check-in times of a certain preliminary screened point of interest are greater than or equal to the preset check-in times threshold, the preliminary screened point of interest is determined as a candidate point of interest. If the historical check-in times are less than the preset check-in times threshold, the preliminary screened point of interest is filtered out. Among them, the preset check-in times threshold can be 3 times, and can also be customized according to specific application requirements. The embodiment of the present invention does not make specific limitations. After completing the screening of all the preliminary screened points of interest, a candidate check-in record data set is generated based on the screened candidate points of interest and the historical check-in record data of the candidate points of interest. That is to say, the candidate check-in record data set includes both candidate points of interest and user information that checks in at each candidate point of interest.

[0071] It should be noted that by adding two lightweight processes, namely, candidate POI screening and check-in record data screening, to the POI recommendation algorithm integrating the user's moving direction based on the user's tolerated distance range and the similarity characteristics between users, the algorithm's lightweight performance can be achieved, making the newly designed POI recommendation algorithm have better relevance while reducing the consumption of computing resources. As shown in Table 1 below, it is not difficult to see from the statistical results of the data volume of the two social network data sets of Foursquare and Gowalla before and after lightweight processing that the effect of lightweight processing based on the above method is relatively ideal.

[0072] Table 1: Data volume statistics before and after lightweight processing

[0073]

[0074] 103. Calculate the moving direction similarity of each of the candidate interest points based on the moving direction and the real-time position, and calculate the collaborative filtering similarity of each of the candidate interest points based on the check-in user information and the check-in record of the target user.

[0075] In an embodiment of the present invention, after obtaining the candidate points of interest in the candidate check-in record data set, in order to determine the suitability of different candidate points of interest with the target user, each candidate point of interest is evaluated from two dimensions: the correlation between the candidate point of interest and the target user's moving direction, and the historical check-in of the associated user at the candidate point of interest. The moving direction similarity of any candidate point of interest is used to characterize the similarity between the direction of the target user's real-time position pointing to the current candidate point of interest and the moving direction of the target user. It can be determined according to the angle between the two directions. The larger the angle, the lower the similarity, and the smaller the angle, the higher the similarity. The collaborative filtering similarity of the candidate point of interest, that is, based on the check-in information of the associated user who has the same interest tendency as the target user on the candidate point of interest, indirectly judges the degree of match between the interest point and the target user's preference. By calculating the moving direction similarity, the user's moving direction characteristics can be learned more accurately, and the moving direction context factors can be more fully and effectively used to improve the recommendation effect.

[0076] In one embodiment of the present invention, for further explanation and limitation, the step of calculating the similarity of the moving direction of any candidate interest point includes:

[0077] According to the latitude and longitude coordinates of the real-time position and the latitude and longitude coordinates of the candidate point of interest, using the tangent theorem of a right triangle, a first angle between a line connecting the real-time position of the target user and the candidate point of interest and a preset direction is calculated;

[0078] Calculate a third angle between a line connecting the real-time position of the target user to the candidate point of interest and the moving direction based on the first angle and a second angle between the moving direction and the preset direction;

[0079] According to the third angle, the moving direction similarity of the candidate interest point is calculated using a direction similarity calculation formula.

[0080] In the embodiment of the present invention, the angle between the position of the point of interest and the direction of the user's movement is calculated based on the user's movement direction. The size of the angle can be used to effectively determine the relevance of the point of interest in terms of direction factors, and the similarity between the direction and the movement direction between the candidate point of interest and the user's current position is used as the direction similarity. Figure 2 As shown in , it is the angle relationship between the position direction of the interest point and the user's moving direction. Among them, the position direction of the interest point is the direction of the line connecting the candidate interest point and the real-time position. Figure 2 In the default direction, set the north direction as the default direction, corresponding to X Axis, real-time position is U( , ) ,in Indicate location The longitude, Indicate location Candidate points of interest The coordinates are expressed as ( , ) , ( , )，…， ( , )，…， ( , ) In the figure ( , ） Indicates the real-time location of the target user U With candidate points of interest According to the tangent theorem of a right triangle, the line segment is calculated and X Axis Angle , the formula is:

[0081] (5);

[0082] ray Represents the moving direction. The angle of this ray relative to the preset direction (due north) is .from Figure 2 It can be seen ,but ( , ） The calculation formula is as follows:

[0083] (6);

[0084] in, The range is [0º, 180º].

[0085] set up Indicates the real-time location of the target user With candidate points of interest The direction similarity is calculated as follows:

[0086] (7);

[0087] in, is the direction similarity, is the third angle, Indicates real-time location. is a candidate interest point. From the above formula, we can conclude that when hour, ;when =180º, Angle The smaller it is, the closer the location of the candidate POI is to the target user's moving direction. ( , ） The larger the value of , the greater the correlation. That is, the range of directional similarity is a decimal between [0,1]. The candidate interest points that are close to the moving direction of the target user have a greater correlation, and the greater the value of directional similarity.

[0088] In one embodiment of the present invention, for further explanation and limitation, the calculating the collaborative filtering similarity of each of the candidate interest points based on the check-in user information and the check-in record of the target user includes:

[0089] Calculating the user similarity between each associated user in the signed-in user information and the target user respectively;

[0090] Adding up all the user similarities to obtain a global user similarity value;

[0091] For each candidate point of interest, the sum of the user similarities between the associated users who checked in to the candidate point of interest and the target user is calculated to obtain the single-point user similarity and value of different candidate points of interest, and the ratio of the single-point user similarity and value to the global user similarity and value is calculated to obtain the collaborative filtering similarity.

[0092] In the embodiment of the present invention, the similarity between the associated users other than the target user in the candidate check-in record data set and the target user in the dimension of check-in points of interest, that is, the user similarity, is first calculated. User similarity is the degree of association between the places visited by the target user and the places visited by the associated users. If the points of interest visited by the target user and a certain associated user are exactly the same, the similarity result is equal to 1; if there are no common points of interest checked in between the two, the intersection is empty and the similarity value is 0. After calculating the similarity between the target user and the associated user, the recommendation value from the user dimension is considered, that is, the calculation of collaborative filtering similarity is performed. Specifically, the candidate points of interest are recommended to the target user. Collaborative filtering similarity Have you been to a candidate point of interest? Users v and target users The sum of the similarities between users and all associated users v and target users The ratio of the sum of similarities between them. The formula is expressed as:

[0093] (8);

[0094] in, Indicates associated user At the candidate point of interest Boolean value of sign-in. If a sign-in occurs, ,otherwise , represents the candidate check-in record dataset, i.e. v Represents all users in the candidate check-in record dataset.

[0095] In one embodiment of the present invention, for further explanation and limitation, the step of calculating the user similarity between any of the associated users and the target user includes:

[0096] Calculate the number of candidate interest points that the associated user and the target user check in at together to obtain the number of intersection interest points;

[0097] Calculate the average of the number of candidate points of interest that the associated user has checked in and the number of candidate points of interest that the target user has checked in to obtain the average number of points of interest check-in;

[0098] The quotient of the number of the intersection points of interest and the average check-in value of the points of interest is calculated to obtain the user similarity.

[0099] In the embodiment of the present invention, any associated user is calculated by the following formula With target users The similarity between users :

[0100] (9);

[0101] in, Indicates the target user At the candidate point of interest Boolean value of the sign-in, Indicates associated user At the candidate point of interest Boolean value for signed in. express Record the candidate points of interest in the dataset for the candidate check-in. The value range of is [0,1]. The larger the value, the higher the similarity between the two users.

[0102] 104. Generate interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

[0103] In an embodiment of the present invention, after obtaining the moving direction similarity and the collaborative filtering similarity, the recommendation value of each candidate point of interest can be calculated based on the above two similarities, and the recommendation values ​​can be arranged in descending order. And the candidate points of interest with higher recommendation values ​​are extracted from the arrangement results according to the preset recommendation quantity as the final recommended points of interest for the target user. Then, the target user's point of interest recommendation information is generated based on the relevant information of the recommended points of interest. The point of interest recommendation information may include address information, place name, related pictures, historical check-in data, related evaluations, scores and other information of the recommended points of interest, which is not specifically limited in the embodiment of the present invention.

[0104] By evaluating the matching degree between points of interest and target users based on feature information of two dimensions, namely moving direction similarity and collaborative filtering similarity, the dimension of spatial information is expanded, and the feature information of point of interest evaluation is enriched, so that the points of interest are more compatible with the user's preference, thereby improving the accuracy of personalized point of interest recommendation for users.

[0105] Generating the interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity, including:

[0106] For each of the candidate interest points, linearly fitting the moving direction similarity and the collaborative filtering similarity is performed respectively to obtain a recommendation value of each of the candidate interest points;

[0107] The candidate interest points are sorted according to the recommendation values, and interest point recommendation information for the target user is generated based on the candidate interest points whose order numbers meet a preset recommendation number.

[0108] In the embodiment of the present invention, in order to combine the pixel points in the moving direction dimension and the similarity in the user dimension to evaluate the recommendation value of the candidate interest point, the moving direction similarity and the collaborative filtering similarity are linearly fitted to obtain the final recommendation value of the candidate interest point. The recommendation value is set to , the calculation formula is expressed as:

[0109] (10);

[0110] in, Representing candidate interest points The direction similarity of Representing candidate interest points After calculating the recommendation value, sort the candidate points of interest in descending order according to the recommendation value, and select ( …) points of interest are recommended to the user as results.

[0111] The present invention provides a lightweight point of interest recommendation method integrating user movement direction. In an embodiment of the present invention, by responding to the point of interest recommendation demand of the target user, the real-time position, movement direction and global points of interest of the target user are obtained, wherein the global points of interest are places matching the point of interest recommendation demand in a global place check-in record data set; the global points of interest are screened according to the real-time position and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest meeting check-in and distance conditions and the check-in user information of the candidate points of interest; the global points of interest are screened according to the real-time position and the check-in records of each point of interest to obtain a candidate check-in record data set; ... global points of interest are screened according to the real-time position and the check-in records of each point of interest The real-time position is used to calculate the moving direction similarity of each of the candidate points of interest, and the collaborative filtering similarity of each of the candidate points of interest is calculated based on the check-in user information and the check-in record of the target user; based on the moving direction similarity and the collaborative filtering similarity, the point of interest recommendation information of the target user is generated, and the recommendation value calculation for the global points of interest is avoided, which greatly reduces the data processing amount of the point of interest recommendation process and realizes lightweight recommendation. At the same time, the point of interest recommendation is performed by integrating the moving direction features, which enriches the recommended feature information and ensures the accuracy of the recommendation, thereby greatly improving the data processing efficiency of the point of interest recommendation.

[0112] Furthermore, as a response to the above Figure 1 The implementation of the method shown in the figure, the embodiment of the present invention provides a lightweight interest point recommendation device integrating the user's moving direction, such as Figure 3 As shown, the device comprises:

[0113] An acquisition module 31 is used to acquire the real-time location and movement direction of the target user in response to the target user's interest point recommendation requirement, wherein the global interest point is a location in the global location check-in record data set that matches the interest point recommendation requirement;

[0114] A screening module 32, configured to screen the global points of interest according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet check-in and distance conditions and check-in user information of the candidate points of interest;

[0115] A calculation module 33, configured to calculate the similarity of the moving directions of the candidate points of interest based on the moving directions and the real-time positions, and to calculate the collaborative filtering similarity of the candidate points of interest based on the check-in user information and the check-in record of the target user;

[0116] The generating module 34 is used to generate the interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

[0117] Furthermore, the acquisition module 31 includes:

[0118] A collection unit, used to collect original compass data of the mobile terminal corresponding to the target user according to a preset sampling frequency and a preset sampling duration to obtain a one-dimensional angle sequence;

[0119] A filtering unit is used to perform one-dimensional median filtering on the one-dimensional angle sequence according to a preset filtering window length, and use the median of the one-dimensional angle sequence as the moving direction of the target user.

[0120] Furthermore, the screening module 32 includes:

[0121] A first calculation unit is used to calculate a longitude and latitude screening range of interest points according to the real-time position and a preset distance range, and screen out preliminary interest points that meet the longitude and latitude screening range of interest points from the global interest points;

[0122] A counting unit, configured to count the global user historical check-in times of each of the preliminarily screened interest points, and determine the preliminarily screened interest points whose historical check-in times are greater than or equal to a preset check-in times threshold as candidate interest points;

[0123] The first generating unit is configured to generate a candidate check-in record data set according to the candidate point of interest and the historical check-in record data of the candidate point of interest.

[0124] Furthermore, the calculation module 33 includes:

[0125] A second calculation unit is used to calculate a first angle between a line connecting the real-time position of the target user and the candidate point of interest and a preset direction based on the latitude and longitude coordinates of the real-time position and the latitude and longitude coordinates of the candidate point of interest and using the tangent theorem of a right triangle;

[0126] A third calculation unit, configured to calculate a third angle between a line connecting the real-time position of the target user to the candidate point of interest and the moving direction according to the first angle and a second angle between the moving direction and the preset direction;

[0127] The fourth calculation unit is used to calculate the moving direction similarity of the candidate interest point according to the third angle using a direction similarity calculation formula, wherein the direction similarity calculation formula is: ;in, is the direction similarity, is the third angle, Indicates real-time location. Candidate points of interest.

[0128] Furthermore, the calculation module 33 further includes:

[0129] A fifth calculation unit, used to respectively calculate the user similarity between each associated user in the signed-in user information and the target user;

[0130] A sixth calculation unit, configured to sum all the user similarities to obtain a global user similarity sum value;

[0131] The seventh calculation unit is used to calculate, for each candidate point of interest, the sum of the user similarities between the associated users who check in to the candidate point of interest and the target user, obtain the single-point user similarities and values ​​of different candidate points of interest, calculate the ratio of the single-point user similarity and value to the global user similarity and value, and obtain the collaborative filtering similarity.

[0132] Furthermore, in a specific application scenario, the fifth calculation unit is specifically used to calculate the number of candidate points of interest that the associated user and the target user have signed in together to obtain the number of intersection points of interest; calculate the average number of candidate points of interest that the associated user has signed in and the number of candidate points of interest that the target user has signed in to obtain the average number of interest point sign-ins; calculate the quotient of the intersection number of interest points and the average number of interest point sign-ins to obtain user similarity.

[0133] Furthermore, the generating module 34 includes:

[0134] A fitting unit, configured to perform linear fitting on the moving direction similarity and the collaborative filtering similarity for each of the candidate interest points, to obtain a recommendation value for each of the candidate interest points;

[0135] The second generating unit is used to sort the candidate interest points according to the recommendation values, and generate interest point recommendation information for the target user based on the candidate interest points whose order numbers meet a preset recommendation number.

[0136] The present invention provides a lightweight point of interest recommendation device integrating user movement direction. The embodiment of the present invention obtains the real-time position, movement direction and global points of interest of the target user in response to the point of interest recommendation demand of the target user, wherein the global points of interest are places matching the point of interest recommendation demand in a global place check-in record data set; the global points of interest are screened according to the real-time position and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest meeting check-in and distance conditions and the check-in user information of the candidate points of interest; the global points of interest are screened according to the real-time position and the check-in records of each point of interest to obtain a candidate check-in record data set; ... The real-time position is used to calculate the moving direction similarity of each of the candidate points of interest, and the collaborative filtering similarity of each of the candidate points of interest is calculated based on the check-in user information and the check-in record of the target user; based on the moving direction similarity and the collaborative filtering similarity, the point of interest recommendation information of the target user is generated, and the recommendation value calculation for the global points of interest is avoided, which greatly reduces the data processing amount of the point of interest recommendation process and realizes lightweight recommendation. At the same time, the point of interest recommendation is performed by integrating the moving direction features, which enriches the recommended feature information and ensures the accuracy of the recommendation, thereby greatly improving the data processing efficiency of the point of interest recommendation.

[0137] According to an embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the lightweight POI recommendation method integrating the user movement direction in any of the above method embodiments.

[0138] Figure 4 A schematic diagram of the structure of a terminal provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the terminal.

[0139] like Figure 4 As shown, the terminal may include: a processor (processor) 402 , a communication interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

[0140] The processor 402 , the communication interface 404 , and the memory 406 communicate with each other via a communication bus 408 .

[0141] The communication interface 404 is used for network communication with other devices such as a client or other servers.

[0142] The processor 402 is used to execute the program 410, and specifically can execute the relevant steps in the above-mentioned embodiment of the lightweight interest point recommendation method integrating the user movement direction.

[0143] Specifically, the program 410 may include program codes, which include computer operation instructions.

[0144] The processor 402 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiment of the present invention. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0145] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0146] The program 410 may be specifically configured to enable the processor 402 to perform the following operations:

[0147] In response to the target user's point of interest recommendation requirement, obtain the target user's real-time location, movement direction, and global points of interest, wherein the global points of interest are locations in the global location check-in record data set that match the point of interest recommendation requirement;

[0148] The global points of interest are screened according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet check-in and distance conditions and check-in user information of the candidate points of interest;

[0149] Calculating the similarity of the moving directions of the candidate points of interest based on the moving directions and the real-time locations, and calculating the collaborative filtering similarity of the candidate points of interest based on the check-in user information and the check-in record of the target user;

[0150] Generate interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

[0151] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0152] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A lightweight POI recommendation method integrating user movement direction, characterized in that: include: In response to the target user's point of interest recommendation requirement, obtain the target user's real-time location, movement direction, and global points of interest, wherein the global points of interest are locations in the global location check-in record data set that match the point of interest recommendation requirement; The global points of interest are screened according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet check-in and distance conditions and check-in user information of the candidate points of interest; According to the moving direction and the real-time position, the moving direction similarity of each of the candidate points of interest is calculated, wherein the step of calculating the moving direction similarity of any of the candidate points of interest comprises: according to the latitude and longitude coordinates of the real-time position and the latitude and longitude coordinates of the candidate point of interest, using the tangent theorem of a right triangle, calculating a first angle between a line connecting the real-time position of the target user and the candidate point of interest and a preset direction; according to the first angle and the second angle between the moving direction and the preset direction, calculating a third angle between a line connecting the real-time position of the target user to the candidate point of interest and the moving direction; according to the third angle, using a direction similarity calculation formula, calculating the moving direction similarity of the candidate points of interest, wherein the direction similarity calculation formula is: ;in, is the direction similarity, is the third angle, Indicates real-time location. Representing candidate interest points; Calculating the collaborative filtering similarity of each of the candidate interest points based on the check-in user information and the check-in record of the target user, specifically including: respectively calculating the user similarity between each associated user in the check-in user information and the target user; summing up all the user similarities to obtain a global user similarity sum value; for each candidate interest point, calculating the sum of the user similarities between the associated users who check in at the candidate interest point and the target user to obtain single-point user similarities and values ​​of different candidate interest points, and calculating the ratio of the single-point user similarity sum value to the global user similarity sum value to obtain the collaborative filtering similarity; Generate interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

2. The method according to claim 1, characterized in that Obtaining the moving direction of the target user includes: According to a preset sampling frequency and a preset sampling duration, raw compass data of the mobile terminal corresponding to the target user is collected to obtain a one-dimensional angle sequence; One-dimensional median filtering is performed on the one-dimensional angle sequence according to a preset filtering window length, and the median of the one-dimensional angle sequence is used as the moving direction of the target user.

3. The method according to claim 1, characterized in that The screening of the global points of interest according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set includes: Calculating a longitude and latitude screening range of points of interest based on the real-time position and the preset distance range, and screening out preliminary screening points of interest that meet the longitude and latitude screening range of points of interest from the global points of interest; Counting the global user historical check-in times of each of the preliminarily screened interest points, and determining the preliminarily screened interest points whose historical check-in times are greater than or equal to a preset check-in times threshold as candidate interest points; A candidate check-in record data set is generated according to the candidate points of interest and the historical check-in record data of the candidate points of interest.

4. The method according to claim 1, characterized in that The step of calculating the user similarity between any of the associated users and the target user comprises: Calculate the number of candidate interest points that the associated user and the target user check in at together to obtain the number of intersection interest points; Calculate the average of the number of candidate points of interest that the associated user has checked in and the number of candidate points of interest that the target user has checked in to obtain the average number of points of interest check-in; The quotient of the number of the intersection points of interest and the average check-in value of the points of interest is calculated to obtain the user similarity.

5. The method according to any one of claims 1 to 4, characterized in that The generating the interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity includes: For each of the candidate interest points, linearly fitting the moving direction similarity and the collaborative filtering similarity is performed respectively to obtain a recommendation value of each of the candidate interest points; The candidate interest points are sorted according to the recommendation values, and interest point recommendation information for the target user is generated based on the candidate interest points whose order numbers meet a preset recommendation number.

6. A lightweight point of interest recommendation device integrating user movement direction, characterized in that: include: An acquisition module, configured to acquire the real-time location, movement direction, and global points of interest of the target user in response to the point of interest recommendation requirement of the target user, wherein the global points of interest are places in the global place check-in record data set that match the point of interest recommendation requirement; a screening module, configured to screen the global points of interest according to the real-time location and the check-in records of each point of interest to obtain a candidate check-in record data set, wherein the candidate check-in record data set includes candidate points of interest that meet check-in and distance conditions and check-in user information of the candidate points of interest; A calculation module is used to calculate the similarity of the moving directions of each of the candidate points of interest based on the moving direction and the real-time position, wherein the step of calculating the similarity of the moving directions of any of the candidate points of interest comprises: calculating the first angle between the line between the real-time position of the target user and the candidate point of interest and the preset direction using the tangent theorem of a right triangle based on the latitude and longitude coordinates of the real-time position and the latitude and longitude coordinates of the candidate point of interest; calculating the third angle between the line between the real-time position of the target user and the candidate point of interest and the moving direction based on the first angle and the second angle between the moving direction and the preset direction; calculating the similarity of the moving directions of the candidate points of interest based on the third angle using a direction similarity calculation formula, wherein the direction similarity calculation formula is: ;in, is the direction similarity, is the third angle, Indicates real-time location. Representing candidate interest points; The calculation module is further used to calculate the collaborative filtering similarity of each of the candidate interest points based on the check-in user information and the check-in record of the target user, specifically including: respectively calculating the user similarity between each associated user in the check-in user information and the target user; summing up all the user similarities to obtain a global user similarity sum value; for each candidate interest point, calculating the sum of the user similarities between the associated users who check in at the candidate interest point and the target user to obtain single-point user similarities and values ​​of different candidate interest points, and calculating the ratio of the single-point user similarity sum value to the global user similarity sum value to obtain the collaborative filtering similarity; A generating module is used to generate the interest point recommendation information of the target user according to the moving direction similarity and the collaborative filtering similarity.

7. A storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the lightweight point of interest recommendation method integrating user movement direction as described in any one of claims 1 to 5.

8. A terminal, comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the lightweight point of interest recommendation method integrating user movement direction as described in any one of claims 1-5.

Images