CN116401443A - Point recommendation method, device, electronic equipment and storage medium - Google Patents

Abstract

The invention discloses a point position recommending method, a point position recommending device, electronic equipment and a storage medium, wherein the point position recommending method comprises the following steps: acquiring current snapshot image information of a user; comparing the user snap image with a plurality of file information of the reference file set to determine target file information corresponding to the user snap image; determining a corresponding travel frequency matrix according to the target archive information and the current point location information; the target archive information comprises snapshot image information of a user collected at a plurality of travel points, and a travel frequency matrix is used for describing the frequency of the user at each travel point at different time, wherein the travel points comprise pre-calculated distance similarity and time similarity; and determining the target interest point based on the travel frequency matrix, and recommending the point according to the target interest point. Therefore, the interest point recommendation has obvious rationality, the interest demands of users corresponding to different time and different places can be solved, and the recommendation accuracy and recall rate are improved.

Description

Point recommendation method, device, electronic equipment and storage medium

technical field

The present invention relates to the technical field of data processing, in particular to a point recommendation method, device, electronic equipment and storage medium.

Background technique

At present, when different profile users arrive at certain places, point-of-interest recommendations can be made based on the information that the profile users have visited specific places in historical time, but the existing recommendation methods are usually based on geographic location information or for different time Recommendations based on information make the recommendation of points of interest obviously objective, and cannot solve the user's interest needs corresponding to different times and different locations, resulting in insufficient recommendation accuracy.

Contents of the invention

In the first aspect, the main purpose of the present invention is to provide a point recommendation method, including:

Acquiring the current snapshot image information of the user; the current snapshot image information includes the user snapshot image, the current time information corresponding to the user snapshot image, and the current point information corresponding to the user snapshot image;

Comparing the captured image of the user with a plurality of profile information of the reference profile set, and determining the target profile information corresponding to the captured image of the user;

Determine the corresponding trip times matrix according to the target file information and the current point information; the target file information includes the snapshot image information collected by the user at multiple travel points, and the trip times matrix is used for Describe the number of times the user appears at each travel point at different times, including pre-calculated distance similarity and time similarity between multiple travel points;

A target point of interest is determined based on the trip times matrix, and a point recommendation is performed according to the target point of interest.

Optionally, before acquiring the user's current snapshot image information, it includes:

Determine the set of similarities between the images to be archived according to the characteristics of the personnel images corresponding to the images to be archived captured at each travel point;

The images to be archived are clustered based on the similarity set, and the images to be archived that meet the similarity threshold are archived to obtain the reference archive set.

Optionally, clustering the images to be archived based on the similarity set, and archiving the images to be archived that meet the similarity threshold, after obtaining the reference archive set, includes:

Determine the location information corresponding to each travel point according to the set of reference files, and calculate the distance between any two travel points;

According to the distance between the two travel points, the distance similarity between the two travel points is determined.

Optionally, clustering the images to be archived based on the similarity set, and archiving the images to be archived that meet the similarity threshold, after obtaining the reference archive set, further includes:

Determine the number of occurrences of the user at the travel point in each time period according to the set of reference files;

Constructing the number of trips matrix according to the number of occurrences, and performing normalization processing based on the number of trips matrix, to obtain a matrix vector corresponding to the number of trips matrix;

Calculations are performed based on the matrix vectors to determine the time similarity between two travel points.

Optionally, the determining the corresponding travel times matrix according to the target file information and the current point information includes:

According to the target file information, determine the historical visit points that the user has visited;

determining a point recommendation range for the user according to the current point information;

Among the historical visiting points, the historical visiting points within the recommended range of the points are used as the points to be recommended;

Based on the snapshot image information collected by the user at the point to be recommended, a corresponding travel times matrix is determined.

Optionally, determining the target point of interest based on the trip times matrix, and making recommendations according to the target point of interest includes:

Based on the travel points corresponding to the travel times matrix, determine the time similarity and distance similarity corresponding to the travel points;

Calculate according to the time similarity and distance similarity to obtain the total similarity;

A target point of interest is determined according to the total similarity, and recommendations are made based on the target point of interest.

Optionally, the determining the target point of interest according to the total similarity, and making the recommendation according to the target point of interest includes:

Sorting each travel point according to the total similarity, and determining a target travel point whose total similarity satisfies a predetermined condition;

The address of the point of interest corresponding to the target travel point is used as the target point of interest, and recommendations are made based on the target point of interest.

In a second aspect, an embodiment of the present invention provides a point recommendation device, including:

An acquisition module, configured to acquire the current snapshot image information of the user; the current snapshot image information includes a user snapshot image, current time information corresponding to the user snapshot image, and current point information corresponding to the user snapshot image;

A comparison module, configured to compare the captured image of the user with a plurality of archive information of the reference archive set, and determine the target archive information corresponding to the captured image of the user;

A determining module, configured to determine a corresponding travel times matrix according to the target file information and the current point information; the target file information includes snapshot image information collected by the user at multiple travel points, the The number of trips matrix is used to describe the number of times the user appears at each travel point at different times, including pre-calculated distance similarity and time similarity between multiple travel points;

The recommending module is configured to determine a target point of interest in the travel times matrix based on the current time information and the current point information, and perform point recommendation according to the target point of interest.

In a third aspect, an embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, Steps for implementing the point recommendation method as described above.

In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above point recommendation method are implemented.

Above-mentioned scheme of the present invention comprises following beneficial effect at least:

The point recommendation method provided by the present invention first obtains the current snapshot image information of the user; the current snapshot image information includes the user snapshot image, the current time information corresponding to the user snapshot image, and the current point information corresponding to the user snapshot image; the user snapshot image Compare with multiple file information of the reference file set, determine the target file information corresponding to the user's captured image; determine the corresponding travel times matrix according to the target file information and current point information; the target file information includes the user in multiple The captured image information of travel points is collected, and the travel times matrix is used to describe the number of times users appear at each travel point at different times, including pre-calculated distance similarity and time similarity between multiple travel points; based on The current time information and the current point information determine the target point of interest in the trip times matrix, and make point recommendations based on the target point of interest. As a result, the recommendation of points of interest is obviously reasonable, and it can solve the user's interest needs corresponding to different times and different locations, and improve the accuracy and recall rate of the recommendation.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those skilled in the art, other drawings can also be obtained according to the structures shown in these drawings without creative effort.

FIG. 1 is a schematic diagram of the overall flow of a point recommendation method provided by an embodiment of the present invention;

FIG. 2 is a schematic flow diagram of a point recommendation method provided by an embodiment of the present invention;

FIG. 3 is a structural block diagram of a point recommendation device provided by an embodiment of the present invention;

Fig. 4 is a structural block diagram of an electronic device provided by an embodiment of the present invention.

The realization of the purpose of the present invention, functional characteristics and advantages will be further described in conjunction with the embodiments and with reference to the accompanying drawings.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

The terms "first", "second" and "third" in the specification and claims of the present invention and the above drawings are used to distinguish different objects, rather than to describe a specific order. Furthermore, the term "comprise", as well as any variations thereof, is intended to cover a non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally further includes For other steps or units inherent in these processes, methods, products or apparatuses.

Firstly, the solution of the embodiment of the present application will be introduced by way of example in conjunction with the relevant drawings.

As shown in Figure 1, a specific embodiment of the present invention provides a point recommendation method, including:

S10. Obtain current snapshot image information of the user, where the current snapshot image information includes the user snapshot image, current time information corresponding to the user snapshot image, and current point information corresponding to the user snapshot image.

In this embodiment, the user's current snapped image information can be obtained through the camera according to the user's recommendation request, or the user's current snapped image information can be obtained through the camera when the user arrives at a certain snapping location, and the camera can be set at a traffic intersection , Shopping malls, neighborhoods and other places with a large flow of people, so as to collect user snapshot images in time; the user snapshot images can be the user's face image, or the user's human body image, etc. It is understandable that the user's face image or The human body image can be the image data that has been archived in the historical time. For example, archiving the images taken by the user within one month or within 24 hours can determine the corresponding archived data of the user, so that the current captured image information of the user can be acquired Determine the archived data corresponding to the user.

Specifically, before obtaining the user's current captured image information according to the user's recommendation request, it includes: determining the similarity set between the images to be archived according to the characteristics of the personnel images corresponding to the images to be archived captured at each travel point; The degree set clusters the images to be archived, and archives the images to be archived that meet the similarity threshold to obtain a reference archive set.

In this embodiment, the images to be archived captured at each travel point may include face images or human body images of different users, and the similarity set may be calculated from the user's face images. When the user captures an image, the similarity calculation can be performed on the user's captured image. For example, the cosine similarity can be used to calculate the similarity between two images to be archived, that is, the closer the cosine value between the two images to be archived is to When 1, it means that the similarity between the two archived data is greater; it can be understood that after calculating the similarity set between the images to be archived, the data to be archived is clustered into the corresponding image through the similarity set Then, for each image pile, the image with better image quality is determined as the initial archive set, and then the captured images of users collected at subsequent travel points can be archived according to the initial archive set, so as to obtain the reference archive set; for example, in Fig. Each point shown in 2 can be represented as a file, and after the similarity calculation and clustering of each file, multiple image piles can be formed, thereby completing the filing operation between different files.

In a preferred embodiment, the set of reference archives can be aidN{aid1, aid2, aid3...aidN}, and the captured images in each image archive set aidN can be expressed as b _n {b ₁ , b ₂ , b ₃ ,..b _n }, calculate the similarity between two captured images by performing n×n on the captured images collected at each travel point, therefore, the calculated similarity set can be sim _ij {sim ₁₂ ,sim ₁₃ ....sim _ij }, i, j represent the data b _i , b _j ; after comparing the similarity set with the first similarity threshold, the snapshots of users corresponding to the similarity set greater than the first similarity threshold The image is then retained, so it can be determined that the initial archive set can be expressed as aidN{aid1,aid2,aid3...aidN}; understandably, the first similarity threshold can be expressed as α _sim , when sim _ij -α When _sim > 0, the corresponding user captured images are retained to form the initial archive collection. When the unarchived images collected at each subsequent travel point need to be archived, the similarity comparison between the initial archive collection and the unarchived images can be performed , and then compared by the second similarity threshold, when it is determined that the similarity is greater than the second similarity threshold, the user’s captured image can be retained in the corresponding initial archive set to form a reference archive set; it can be understood that the first The second similarity threshold can be expressed as β _ij , and the similarity set determined after comparing the subsequent collected unarchived images with the initial archive set can be expressed as sim _ij {sim ₁₂ ,sim ₁₃ ....sim _ij }, Therefore, the captured image of the user corresponding to sim _ij −β _ij >0 can be archived in the corresponding archive data.

Further, the above-mentioned clustering of the images to be archived based on the similarity set, and archiving the images to be archived that meet the similarity threshold, after obtaining the reference archive set, includes: determining the location information corresponding to each travel point according to the reference archive set , and calculate the distance between the two travel points; according to the distance between the two travel points, determine the distance similarity between the two travel points.

In this embodiment, each user's snapped image in the reference file collection includes corresponding travel point information and corresponding point of interest. The travel point information may include latitude and longitude, map coordinates, etc., and the point of interest may be travel Scenic spots, schools, restaurants and other locations around the point, through the above-mentioned reference file collection, the distance similarity between each travel point is calculated. The larger the distance between the travel points, the smaller the distance similarity. The smaller the distance between travel points, the greater the distance similarity; among them, since users usually choose to visit POIs that are closer to their location at each travel point, the distance between each travel point The similarity can be inversely proportional to the distance between the user’s visits to various points of interest. The larger the distance, the smaller the distance similarity, and the smaller the distance, the greater the distance similarity.

It can be understood that the above distance similarity can be calculated using the following formula:

distance(li,lj)=R*arccos[sin(lat _i )*sin(lat _j )+cos(lat _i )*cos(lat _j )*cos(lat _i -lon _j )];

表示两个出行点位li和lj之间的距离相似度，distiance(li,lj)表示两个出行点位li和lj之间的距离，lat_i和lon_i表示出行点位的经纬度，R为地球半径:R＝6378.137km；由上公式可以确定出两个出行点位之间的距离，两个出行点位之间的距离可以是道路通行距离，通过两个出行点位之间的距离可以确定出对应的距离相似度。in,

Indicates the distance similarity between two travel points li and lj, distance(li,lj) represents the distance between two travel points li and lj, lat _i and lon _i represent the latitude and longitude of the travel point, and R is Radius of the Earth: R=6378.137km; the distance between the two travel points can be determined by the above formula, the distance between the two travel points can be the road travel distance, and the distance between the two travel points can be Determine the corresponding distance similarity.

Further, the above-mentioned clustering of the images to be archived based on the similarity set, and archiving the images to be archived that meet the similarity threshold, after obtaining the reference archive set, further includes: determining the travel point of the user in each time period according to the reference archive set The number of occurrences; the trip times matrix is constructed according to the number of occurrences, and the matrix vector is obtained based on the normalization process based on the trip times matrix; the time similarity between two trip points is determined by calculation based on the matrix vector.

In this embodiment, the trip times matrix can be a two-dimensional table matrix, and it is obtained by statistics that users appear at different travel points in different time periods. After normalizing the trip times matrix, multiple ones are determined. One-dimensional vector matrix, so multiple one-dimensional vector matrices can be represented by matrix vectors, and then the cosine similarity calculation is performed based on the matrix vectors, so that the time similarity between two travel points can be determined.

It is understandable that the table matrix corresponding to different time periods is as follows:

Among them, the table matrix corresponding to the number of occurrences of users in different time periods in each travel point carema _n is as follows:

It can be understood from the above table matrix that 24 hours can be divided into different time periods, and then the number of occurrences of multiple users at each travel point in each time period can be counted. For example, in the above table matrix, t ₄ corresponds to In the period of time, the number of occurrences of the travel point carema ₁ corresponding to multiple users is 56 times, thus determining the captured images of the user collected by the user in different time periods within 24 hours, and then counting the corresponding occurrence times. After the above table matrix is determined, normalization calculation is performed to determine the matrix vector.

It can be understood that, when performing normalized calculation through the above table matrix, the following formula can be used for calculation:

表示出行点位carema_li在时间点tj中所有档案出现的次数，N_t表示出行点位carema_li所有出现的次数，在计算出一维的向量矩阵后，向量矩阵可以表示为/>

由此，矩阵向量可以表示为/>

在确定出矩阵向量后，通过余弦相似度将上述的矩阵向量计算出两两出行点位之间的时间相似度，并可以采用以下公式进行计算：in,

Indicates the number of occurrences of the travel point carema _li in all files at the time point tj, and N _t represents the number of occurrences of the travel point carema _li . After calculating the one-dimensional vector matrix, the vector matrix can be expressed as />

Thus, a matrix-vector can be expressed as />

After the matrix vector is determined, the above-mentioned matrix vector is used to calculate the time similarity between two travel points through the cosine similarity, and the following formula can be used for calculation:

t_sim_li,lj表示时间相似度。

t_sim _li,lj represent time similarity.

由此通过计算出所有出行点位的矩阵向量后，可以将所有出行点位的矩阵向量进行余弦相似度计算，由此可以确定出两两出行点位之间的时间相似度，时间相似度可以表示为两两出行点位之间的出行时间和出行次数的相似度结果，因此，通过将两两出行点位计算得到的矩阵向量进行余弦相似度计算后，可以确定出所有出行点位之间的时间相似度，在后续进行点位推荐时，当用户到达某一个出行点位时，可以根据该出行点位与各个出行点位之间的时间相似度进行推荐对应点位，提升了点位推荐的准确性；举例来说，在计算时间相似度时，出行点位包括A、B、C、D，当用户在上午11点半出现在A点位时，A点位可以采集到用户在A点位的抓拍图像信息，进而可以根据A点位的抓拍图像信息确定出对应的时间信息，并匹配确定出处于t4时间段，并可确定出对应的点位信息为A点位，由此在计算时间相似度时，可以将t4时间段和A点位对应的出行次数进行归一化计算，由此得到对应的归一化值，进而根据该归一化值通过矩阵向量表示后再计算出在t4时间段A点位与B、C、D出行点位之间的时间相似度。Among them, when calculating the time similarity, the travel point corresponding to the current point information can be expressed as carema _li , therefore, the time similarity between the travel point carema _li corresponding to the current point information and each travel point carema _n It can be calculated by the above formula; when calculating, the travel time and the number of trips corresponding to each travel point can be normalized to the mean and variance, that is to say, the travel time and the number of trips in the above-mentioned trip times matrix are normalized to In the distribution with a mean value of 0 and a variance of 1, the mean value of the obtained data is 0 and a variance of 1. Therefore, the number of trips and the travel time corresponding to the travel point carema _li corresponding to the current point information can be calculated first. Calculate, and then calculate the normalized value corresponding to carema _li according to the variance result and the normalization formula; it is understandable that after calculating the normalized value corresponding to the carema _li at the travel point, multiple normalized It can be expressed as a matrix vector of travel point carema _li , that is, the above

Therefore, after calculating the matrix vectors of all travel points, the cosine similarity calculation can be performed on the matrix vectors of all travel points, so that the time similarity between two travel points can be determined, and the time similarity can be It is expressed as the similarity result of the travel time and the number of trips between two travel points. Therefore, after calculating the cosine similarity of the matrix vector obtained by calculating the two travel points, it can be determined that the distance between all travel points time similarity, in the follow-up point recommendation, when the user arrives at a certain travel point, the corresponding point can be recommended according to the time similarity between the travel point and each travel point, which improves the point Recommendation accuracy; for example, when calculating time similarity, travel points include A, B, C, and D. When the user appears at point A at 11:30 in the morning, point A can collect the user's The captured image information at point A, and then the corresponding time information can be determined according to the captured image information at point A, and matched to determine that it is in the t4 time period, and the corresponding point information can be determined to be point A, thus When calculating the time similarity, the number of trips corresponding to the t4 time period and point A can be normalized and calculated, thereby obtaining the corresponding normalized value, and then calculated according to the normalized value through the matrix vector representation Show the time similarity between point A and travel points B, C, and D in the t4 time period.

S20. Comparing the captured image of the user with a plurality of archive information in the reference archive set, and determining the target archive information corresponding to the captured image of the user.

In this embodiment, when the user appears at a travel point, the travel point can collect the user's current snapshot image information and the above-mentioned reference file set to perform similarity calculation, so as to determine the corresponding target in the reference file set Therefore, the target file information and the current captured image information can be searched to determine the corresponding travel times matrix, and then determine the corresponding recommended point.

S30. Determine the corresponding trip times matrix according to the target file information and the current point information; the target file information includes the snapshot image information collected by the user at multiple travel points, and the trip times matrix is used to describe the user's appearance at different times The number of travel points, including the pre-calculated distance similarity and time similarity between multiple travel points.

In this embodiment, when the target file information is determined, the historical visit points that the user has visited can be found according to the target file information, and then the corresponding time similarity and distance similarity can be determined through the trip times matrix; It is understood that the current captured image information includes the user's captured image, the current time information corresponding to the user's captured image, and the current point information corresponding to the user's captured image; The historical visit points are determined to obtain.

Specifically, the above-mentioned determination of the corresponding travel times matrix based on the target file information and the current point information includes: determining the historical visiting point that the user has visited according to the target file information; determining the user's point according to the current point information Recommended range: Among the historical visited points, the historical visited points within the recommended point range are used as the points to be recommended; based on the captured image information collected by the user at the points to be recommended, the corresponding travel times matrix is determined.

Among them, the historical visit point indicates the travel point that the user has visited before the current time, and the point recommendation range indicates the range of interest points corresponding to the user at the current point, which can be collected by combining the current captured image information and reference files The similarity calculation is performed for each file in the file. When the current captured image information can determine the target file information with high similarity in the reference file set, the historical visit points that the user has visited can be determined according to the target file information. Optionally, the travel point has a corresponding point recommendation range, and the point recommendation range can be divided according to points of interest, such as schools, supermarkets, restaurants, hotels, scenic spots, etc., so points of interest can include learning, shopping, eating, and accommodation , play, etc., through the current point information to determine the point recommendation range, and then can determine the historical visiting points within the recommended point range of the current point among the historical visiting points, and use them as points to be recommended According to the captured image information of the user at the point to be recommended, the trip times matrix can be determined, and then the trip times corresponding to the points to be recommended can be determined through the trip times matrix, so that it is more accurate when recommending users ;For example, the historical visiting points are A, B, C, D, and the current point that the user arrives at is A, and the point recommendation range of point A includes two points, B and C. The two points C are determined as the points to be recommended, and the corresponding trip times are determined from the snapshot image information collected by the user at the two points B and C, and included in the trip times matrix for point recommendation.

S40. Determine the target point of interest based on the trip times matrix, and recommend according to the target point of interest.

In this embodiment, the target point of interest can be determined according to the above-mentioned time similarity and distance similarity, so the corresponding travel point can be determined according to the travel times matrix, which is the same as the current travel point of the user. The distance between locations can be relatively close, and the number of arrivals of the user in the historical time is relatively large, so the distance similarity and time similarity can be comprehensively calculated to determine the location of the travel point, and through the travel The point determines the associated target point of interest.

In an optional embodiment, when the target interest point is determined by the distance similarity and time similarity, the distance similarity and time similarity can be screened first, and the distance similarity and time similarity of the maximum result can be determined. degree, and then carry out comprehensive calculation according to the distance similarity and time similarity after screening, so as to determine the corresponding travel point according to the result of comprehensive calculation.

Specifically, the above-mentioned target point of interest is determined based on the trip times matrix, and the recommendation is made according to the target point of interest, including: based on the travel point corresponding to the trip times matrix, the time similarity and distance similarity corresponding to the trip point are determined; Calculate according to the time similarity and distance similarity to obtain the total similarity; determine the target point of interest according to the total similarity, and make recommendations based on the target point of interest.

In this embodiment, the time similarity can be calculated according to the calculation formula of the above-mentioned time similarity, and the distance similarity can be calculated according to the above-mentioned calculation formula of the distance similarity. Finally, the travel point corresponding to the current captured image information can be determined, the distance similarity can be determined according to the travel point and each travel point, and the time similarity can be determined according to the above trip times matrix, and then the distance similarity can be used The weighted sum of time similarity and time similarity is obtained to obtain the total similarity, and the target point of interest is determined through the total similarity, which can comprehensively consider the user interest needs corresponding to different times and different locations, and improve the accuracy of point of interest recommendation.

It can be understood that when calculating the above total similarity, the following formula can be used for calculation:

sim(li,lj)=α*t_sim _li,lj +(1-α)d_sim _li,lj ,

t_sim _li,lj : time similarity;

d_sim _li,lj : distance similarity

表示两个出行点位li和lj之间的距离相似度，t_sim_li,lj表示上述当前抓拍图像信息对应的出行次数矩阵计算得到的时间相似度，sim(li,lj)表示为总相似度，通过总相似度确定出目标兴趣点位，由此能够结合不同时间段的信息进行综合推荐，推荐的准确性更高。Among them, the above α∈[0,1] is expressed as a weight value,

Indicates the distance similarity between two travel points li and lj, t_sim _li,lj represents the time similarity calculated by the travel number matrix corresponding to the current captured image information above, sim(li,lj) represents the total similarity, The target point of interest is determined by the total similarity, so that the information of different time periods can be combined for comprehensive recommendation, and the accuracy of the recommendation is higher.

Further, determine the target point of interest according to the total similarity, and recommend according to the target point of interest, including: sorting each travel point according to the total similarity, and determining the target travel point whose total similarity meets the predetermined condition ; Use the address of the point of interest corresponding to the target travel point as the target point of interest, and make recommendations based on the target point of interest.

In this embodiment, after the total similarity is determined, the total similarity of each travel point can be sorted in descending order, and the total similarity can be compared with the predetermined similarity after sorting, When the total similarity is greater than the predetermined similarity, it means that the travel point corresponding to the total similarity can be used as the target travel point; when the total similarity is less than the predetermined similarity, it means that the travel point corresponding to the total similarity can be Elimination, so that when recommending to users, recommendations can be made according to the address of the point of interest corresponding to the target travel point, thereby improving the accuracy of point of interest recommendation and making the user's options more diverse.

In an optional embodiment, after sorting according to the total similarity of each travel point in descending order, determine the travel point corresponding to the maximum total similarity as the target travel point, Therefore, the address of the point of interest can be determined according to the target travel point corresponding to the maximum total similarity, and then recommended to the user, which can significantly improve the accuracy and recall rate of the point of interest recommendation.

The point recommendation method provided by the present invention first obtains the current snapshot image information of the user; the current snapshot image information includes the user snapshot image, the current time information corresponding to the user snapshot image, and the current point information corresponding to the user snapshot image; the user snapshot image Compare with multiple file information of the reference file set, determine the target file information corresponding to the user's captured image; determine the corresponding travel times matrix according to the target file information and current point information; the target file information includes the user in multiple The captured image information of travel points is collected, and the travel times matrix is used to describe the number of times users appear at each travel point at different times, including pre-calculated distance similarity and time similarity between multiple travel points; based on The current time information and the current point information determine the target point of interest in the trip times matrix, and make point recommendations based on the target point of interest. As a result, the recommendation of points of interest is obviously reasonable, and it can solve the user's interest needs corresponding to different times and different locations, and improve the accuracy and recall rate of the recommendation.

It can be understood that, in the specific implementation manner of this application, related data such as snapshot image information and file information are involved, when the embodiment in this application is applied to a specific product or technology, it is necessary to obtain the user's permission or consent, and The collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

As shown in Fig. 3, the embodiment of the present invention provides a point recommendation device 10, including:

The acquiring module 11 is used to acquire the current snapshot image information of the user, the current snapshot image information including the user snapshot image, the current time information corresponding to the user snapshot image and the current point information corresponding to the user snapshot image;

The comparison module 12 is used to compare the user's captured image with a plurality of file information of the reference file set, and determine the target file information corresponding to the user's captured image;

The determination module 13 is used to determine the corresponding trip times matrix according to the target file information and the current point information; the target file information includes the snapshot image information collected by the user at multiple travel points, and the trip times matrix is used to describe the user in different places. The number of times the time appears at each travel point;

The recommendation module 14 is configured to determine the target point of interest based on the travel times matrix, and perform point recommendation according to the target point of interest.

The point recommendation device 10 provided by the present invention first obtains the current snapshot image information of the user; the current snapshot image information includes the user snapshot image, the current time information corresponding to the user snapshot image, and the current point information corresponding to the user snapshot image; the user snapshot The image is compared with the multiple archive information of the reference archive collection, and the target archive information corresponding to the captured image of the user is determined; the corresponding travel times matrix is determined according to the target archive information and the current point information; the target archive information includes the user in multiple The captured image information of each travel point, the travel times matrix is used to describe the number of times the user appears at each travel point at different times, including the pre-calculated distance similarity and time similarity between multiple travel points; Based on the current time information and the current point information, the target point of interest is determined in the trip times matrix, and the point of interest is recommended according to the target point of interest. As a result, the recommendation of points of interest is obviously reasonable, and it can solve the user's interest needs corresponding to different times and different locations, and improve the accuracy and recall rate of the recommendation.

It should be noted that the point recommendation device 10 provided in the specific embodiment of the present invention is a device corresponding to the above point recommendation method, and all embodiments of the above point recommendation method are applicable to the point recommendation device 10. In the embodiment of the recommendation device 10, there are corresponding modules corresponding to the steps in the above-mentioned point recommendation method, which can achieve the same or similar beneficial effects. In order to avoid too much repetition, each module in the point recommendation device 2 is not performed here. Too much to repeat.

As shown in FIG. 4 , the specific embodiment of the present invention also provides an electronic device 20, including a memory 202, a processor 201, and a computer program stored in the memory 202 and operable on the processor 201. The processor 201 The steps of implementing the above point recommendation method when executing the computer program.

Specifically, the processor 201 is used to call the computer program stored in the memory 202, and perform the following steps:

Obtain the user's current snapshot image information; the current snapshot image information includes the user snapshot image, the current time information corresponding to the user snapshot image, and the current point information corresponding to the user snapshot image;

Comparing the image captured by the user with multiple file information in the reference file set, and determining the target file information corresponding to the image captured by the user;

Determine the corresponding trip times matrix according to the target file information and the current point information; the target file information includes the snapshot image information collected by the user at multiple travel points, and the trip times matrix is used to describe the user's appearance at each trip at different times The number of points, including pre-calculated distance similarity and time similarity between multiple travel points;

Based on the current time information and the current point information, the target point of interest is determined in the trip times matrix, and the point of interest is recommended according to the target point of interest.

Optionally, before the processor 201 executes acquiring the user's current snapshot image information, it includes:

Determine the similarity set between the images to be archived according to the characteristics of the personnel images corresponding to the images to be archived captured at each travel point;

Based on the similarity set, the images to be archived are clustered, and the images to be archived that meet the similarity threshold are archived to obtain a reference archive set.

Optionally, the processor 201 performs clustering of the images to be archived based on the similarity set, and archives the images to be archived that meet the similarity threshold, and after obtaining the reference archive set, includes:

Determine the location information corresponding to each travel point according to the reference file set, and calculate the distance between any two travel points;

According to the distance between any two travel points, the distance similarity between any two travel points is determined.

Optionally, the processor 201 performs clustering of the images to be archived based on the similarity set, and archives the images to be archived that meet the similarity threshold, and after obtaining the reference archive set, further includes:

Determine the number of occurrences of the user at the travel point in each time period according to the reference file collection;

Construct a trip times matrix according to the number of occurrences, and perform normalization processing based on the trip times matrix to obtain a matrix vector corresponding to the trip times matrix;

Calculate based on the matrix vector to determine the time similarity between two travel points.

Optionally, the processor 201 determines the corresponding travel times matrix according to the target file information and the current point information, including:

According to the target file information, determine the historical visit points that the user has visited;

According to the current point information, determine the user's point recommendation range;

Among the historical visiting points, the historical visiting points within the point recommendation range are regarded as the points to be recommended;

Based on the captured image information collected by the user at the point to be recommended, the corresponding trip times matrix is determined.

Optionally, the processor 201 determines the target point of interest based on the number of trips matrix, and makes recommendations according to the target point of interest, including:

Based on the travel points corresponding to the travel times matrix, determine the time similarity and distance similarity corresponding to the travel points;

Calculate according to time similarity and distance similarity to get the total similarity;

Determine the target POI according to the total similarity, and make recommendations based on the target POI.

Optionally, the processor 201 determines the target point of interest according to the total similarity, and makes recommendations according to the target point of interest, including:

Sort each travel point according to the total similarity, and determine the target travel point whose total similarity meets the predetermined condition;

The address of the point of interest corresponding to the target travel point is used as the target point of interest, and recommendations are made based on the target point of interest.

That is, in a specific embodiment of the present invention, when the processor 201 of the electronic device 20 executes the computer program, the steps of the above point recommendation method are implemented, so that the point of interest recommendation has obvious rationality, and can solve the problem of corresponding points at different times and different places. The user's interest needs improve the accuracy and recall rate of the recommendation.

It should be noted that since the processor 201 of the electronic device 20 executes the computer program to implement the steps of the above point recommendation method, all embodiments of the above point recommendation method are applicable to the electronic device 20, and can achieve the same or similar beneficial effects.

In the computer-readable storage medium provided in the embodiment of the present invention, a computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the point recommendation method or the application endpoint point recommendation method provided in the embodiment of the present invention is implemented. Each process can achieve the same technical effect, so in order to avoid repetition, it will not be repeated here.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware, and the programs can be stored in a computer-readable storage medium. During execution, it may include the processes of the embodiments of the above-mentioned methods. Wherein, the storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM) or a random access memory (Random Access Memory, RAM for short).

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Under the conception of the present invention, the equivalent structural transformation made by using the description of the present invention and the contents of the accompanying drawings, or directly/indirectly used in Other relevant technical fields are all included in the patent protection scope of the present invention.

Claims (10)

1. A point location recommendation method, comprising:

acquiring current snapshot image information of a user; the current snapshot image information comprises a user snapshot image, current time information corresponding to the user snapshot image and current point location information corresponding to the user snapshot image;

Comparing the user snap image with a plurality of file information of a reference file set to determine target file information corresponding to the user snap image;

determining a corresponding travel frequency matrix according to the target archive information and the current point location information; the target archive information comprises snapshot image information of the user collected at a plurality of travel points, the travel frequency matrix is used for describing the frequency of the user at each travel point at different time, and the travel points comprise pre-calculated distance similarity and time similarity;

and determining a target point of interest in the travel frequency matrix based on the current time information and the current point information, and performing point location recommendation according to the target point of interest.

2. The point recommendation method according to claim 1, wherein before the obtaining the current snap image information of the user, the method comprises:

according to personnel image features corresponding to the images to be archived, which are captured by each travel point, determining a similarity set between the images to be archived;

and clustering the images to be archived based on the similarity set, archiving the images to be archived which meet the similarity threshold, and obtaining the reference archive set.

3. The point recommendation method according to claim 2, wherein the clustering the images to be archived based on the similarity set and archiving the images to be archived that meet a similarity threshold value, after obtaining the reference archive set, includes:

determining position information corresponding to each travel point according to the reference file set, and calculating the distance between every two travel point;

and determining the distance similarity between the two travel points according to the distance between the two travel points.

4. The point recommendation method according to claim 2, wherein the clustering the images to be archived based on the similarity set, archiving the images to be archived that meet a similarity threshold, and after obtaining the reference archive set, further comprises:

determining the occurrence times of the user at the trip point location in each time period according to the reference file set;

constructing a travel frequency matrix according to the occurrence frequency, and carrying out normalization processing based on the travel frequency matrix to obtain a matrix vector corresponding to the travel frequency matrix;

and calculating based on the matrix vector, and determining the time similarity between every two travel points.

5. The point location recommendation method according to claim 1, wherein the determining the corresponding trip count matrix according to the target profile information and the current point location information includes:

determining a historical visit point of the user according to the target archive information;

determining the point position recommendation range of the user according to the current point position information;

taking the historical visit points in the point recommending range as points to be recommended;

and determining a corresponding travel frequency matrix based on the captured image information of the user acquired at the point to be recommended.

6. The point location recommendation method according to claim 1, wherein determining a target point of interest based on the trip count matrix and recommending the target point of interest includes:

based on the travel point positions corresponding to the travel times matrix, determining the time similarity and the distance similarity corresponding to the travel point positions;

calculating according to the time similarity and the distance similarity to obtain total similarity;

and determining a target point of interest according to the total similarity, and recommending according to the target point of interest.

7. The method of point recommendation according to claim 6, wherein determining a target point of interest according to the total similarity and recommending according to the target point of interest comprises:

sequencing all travel points according to the total similarity, and determining a target travel point position of which the total similarity meets a preset condition;

and taking the interest point address corresponding to the target trip point as a target interest point, and recommending according to the target interest point.

8. A point location recommendation device, comprising:

the acquisition module is used for acquiring the current snapshot image information of the user; the current snapshot image information comprises a user snapshot image, current time information corresponding to the user snapshot image and current point location information corresponding to the user snapshot image;

the comparison module is used for comparing the user snap image with the plurality of file information of the reference file set to determine target file information corresponding to the user snap image;

the determining module is used for determining a corresponding travel frequency matrix according to the target archive information and the current point location information; the target archive information comprises snapshot image information of the user collected at a plurality of travel points, the travel frequency matrix is used for describing the frequency of the user at each travel point at different time, and the travel points comprise pre-calculated distance similarity and time similarity;

And the recommending module is used for determining a target interest point in the travel frequency matrix based on the current time information and the current point information and recommending the point according to the target interest point.

9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the spot recommendation method according to any one of claims 1 to 7 when the computer program is executed by the processor.

10. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the point location recommendation method according to any one of claims 1 to 7.

Images