A Location Prediction Method and System Based on Multi-resolution Context Association

Abstract

This invention discloses a location prediction method and system based on multi-resolution context association. The method includes the following steps: Step S1, obtaining multi-resolution temporal information based on user historical trajectory data; Step S2, obtaining multi-resolution spatial information based on user historical trajectory data; Step S3, obtaining the multi-resolution temporal and spatial contextual patterns of the sequence based on the obtained multi-resolution temporal and spatial information; Step S4, obtaining location prediction information based on the obtained multi-resolution temporal and spatial contextual patterns of the sequence. The location prediction method based on multi-resolution context association provided by this invention obtains richer user temporal and spatial sequence behavioral patterns through multi-resolution temporal and spatial information, resulting in higher accuracy in location prediction.

Description

Technical Field

[0001] This invention relates to the field of information prediction technology, specifically to a location prediction method and system based on multi-resolution context association. Background Technology

[0002] Location prediction technology refers to predicting a user's potential future location based on their historical trajectory data over a period of time. It has significant application value in fields such as intelligent transportation, intelligent advertising, tourism recommendations, and criminal tracking; therefore, location prediction is a hot research topic in current big data analytics and smart city construction.

[0003] In recent years, there has been a large amount of research on location prediction. These studies can be mainly divided into three categories: matrix factorization-based methods (e.g., Xingyi Ren, Meina Song, E Haihong, and Junde Song. Context-aware probabilistic matrix factorization modeling for point-of-interest recommendation. Neurocomputing, 241:38-5, 2017.), Markov chain-based methods (e.g., Jia-Dong Zhang, Chi-Yin Chow, and Yanhua Li. Lore: Exploiting sequential influence for location recommendations. In ACM SIGSPATIAL, pages 103-12, 2014.), and deep network-based methods (e.g., Antonios Karatzoglou, Adrian Jablonski, and Michael Beigl. A seq2seq learning approach for modeling semantictrajectories and predicting the next location. In ACM SIGSPATIAL, pages 528-31, 2018.).

[0004] However, these methods have certain limitations. When learning from sequential data, they only consider the absolute time interval and absolute spatial distance between adjacent points. Although these methods have some effect on location prediction, user check-in data is unevenly distributed, and using absolute distance as a resolution can only reflect part of the patterns in user behavior. Therefore, location prediction remains a challenging task. Summary of the Invention

[0005] The purpose of this invention is to overcome the problems of uneven data distribution and low accuracy in existing similar algorithms, and to provide a location prediction method and system based on multi-resolution context association.

[0006] In a first aspect, the present invention provides a location prediction method based on multi-resolution context association, comprising the following steps:

[0007] Step S1: Obtain multi-resolution time information based on the user's historical trajectory data;

[0008] Step S2: Obtain multi-resolution spatial information based on the user's historical trajectory data;

[0009] Step S3: Based on the acquired multi-resolution temporal and spatial information, obtain the multi-resolution temporal and spatial contextual patterns of the sequence;

[0010] Step S4: Obtain location prediction information based on the multi-resolution temporal and spatial context patterns of the acquired sequence.

[0011] According to the first aspect, in a first possible implementation of the first aspect, prior to step S1, the following steps are further included:

[0012] Step S0: Obtain user's historical trajectory data.

[0013] According to the first aspect, in a second possible implementation of the first aspect, step S1 specifically includes the following steps:

[0014] Step S10: Based on the user's historical trajectory data, obtain the time features represented by multi-level resolution time windows.

[0015] According to the second possible implementation of the first aspect, in the third possible implementation of the first aspect, step S2 specifically includes the following steps:

[0016] Step S20: Based on the user's historical trajectory data, obtain spatial features represented by geographic coordinates and multi-level resolution regions.

[0017] Based on the third possible implementation of the first aspect, in the fourth possible implementation of the first aspect,

[0018] Step S31 specifically includes the following steps:

[0019] Step S311: Obtain the fusion hidden layer state value based on the time features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions.

[0020] Step S32: Based on the obtained spatial features represented by geographic coordinates, obtain multi-resolution spatial context rules.

[0021] According to the fourth possible implementation of the first aspect, in the fifth possible implementation of the first aspect, step S31 specifically includes the following steps:

[0022] Step S311: Input the temporal features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions into the embedding expression layer to obtain the corresponding embedding expression;

[0023] Step S312: Input the corresponding embedding expression into the multi-resolution spatiotemporal feature connection layer to obtain the concatenated vector of the corresponding embedding expression;

[0024] Step S313: Input the concatenated vector into the RNN network for sequence pattern learning and obtain the updated state values ​​of the hidden layer of the RNN network;

[0025] Step S314: Obtain the fused hidden layer state value based on the obtained update state value of the RNN network hidden layer.

[0026] According to the fourth possible implementation of the first aspect, in the sixth possible implementation of the first aspect, step S32 specifically includes the following steps:

[0027] The spatial correlation coefficient between two adjacent points in a spatial feature represented by geographic coordinates can be obtained using the following formula:

[0028] ω=e -Δd ;

[0029] Where Δd is the absolute spatial distance between two adjacent points calculated from geographic coordinates, and ω is the spatial correlation coefficient.

[0030] According to the fifth possible implementation of the first aspect, in the seventh possible implementation of the first aspect, step S4 specifically includes the following steps:

[0031] Step S40: Obtain location prediction information based on the following formula, the obtained fusion hidden layer state values, and the spatial correlation coefficient between two adjacent points:

[0032]

[0033] Among them, e u W represents the embedded expression of user information. P These are the parameters that need to be learned; the softmax() function is used for normalization.

[0034] Secondly, the present invention provides a location prediction system based on multi-resolution context association, comprising:

[0035] The multi-resolution time information acquisition module is used to acquire multi-resolution time information based on the user's historical trajectory data;

[0036] The multi-resolution spatial information acquisition module is used to acquire multi-resolution spatial information based on the user's historical trajectory data.

[0037] The time and space context pattern acquisition module is communicatively connected to the multi-resolution time information acquisition module and the multi-resolution space information acquisition module, and is used to acquire the multi-resolution time and space context patterns of the sequence based on the acquired multi-resolution time habits and multi-resolution space information;

[0038] The location prediction information acquisition module is communicatively connected to the time and space context pattern acquisition module, and is used to acquire location prediction information based on the acquired multi-resolution time and space context patterns.

[0039] According to the second aspect, in a first possible implementation of the second aspect, the time and space context pattern acquisition module includes:

[0040] The fusion hidden layer state value acquisition module is communicatively connected to the multi-resolution time information acquisition module and the multi-resolution spatial information acquisition module, and is used to acquire the fusion hidden layer state value based on the time features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions.

[0041] The spatial context pattern acquisition unit is communicatively connected to the multi-resolution spatial information acquisition module and is used to acquire multi-resolution spatial context patterns based on the acquired spatial features represented by geographic coordinates.

[0042] Compared with the prior art, the advantages of the present invention are as follows:

[0043] The location prediction method based on multi-resolution context association provided by this invention obtains richer user temporal and spatial sequence behavior patterns through multi-resolution temporal and spatial information, thereby improving the accuracy of location prediction. Attached Figure Description

[0044] Figure 1 This is a flowchart of the location prediction method based on multi-resolution contextual patterns according to an embodiment of the present invention;

[0045] Figure 2 This is another method flowchart of the location prediction method based on multi-resolution contextual patterns according to an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of a location prediction method based on multi-resolution contextual patterns according to an embodiment of the present invention;

[0047] Figure 4 This is a functional block diagram of the location prediction system based on multi-resolution contextual rules according to an embodiment of the present invention;

[0048] Figure 5 This is another functional module block diagram of the location prediction system based on multi-resolution contextual rules in an embodiment of the present invention. Detailed Implementation

[0049] Referring now to specific embodiments of the invention, examples of which are illustrated in the accompanying drawings. Although the invention will be described in conjunction with specific embodiments, it will be understood that it is not intended to limit the invention to the described embodiments. Rather, it is intended to cover variations, modifications, and equivalents included within the spirit and scope of the invention as defined by the appended claims. It should be noted that the method steps described herein can be implemented by any functional block or functional arrangement, and any functional block or functional arrangement can be implemented as a physical entity or a logical entity, or a combination of both.

[0050] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] Note: The examples described below are merely specific examples and are not intended to limit the embodiments of the present invention to the specific steps, values, conditions, data, order, etc. Those skilled in the art can utilize the concept of the present invention to construct more embodiments not mentioned herein by reading this specification.

[0052] Location prediction has important applications in fields such as intelligent transportation, intelligent advertising, tourism recommendation, and criminal tracking. Existing location prediction methods only consider the absolute time interval and absolute spatial interval between adjacent points when learning from sequence data. Since user check-in data is evenly distributed, using absolute time interval and absolute spatial interval can only reflect some of the patterns of user behavior, resulting in low location prediction accuracy.

[0053] In view of this, see Figure 1 and Figure 3 As shown, this embodiment of the invention provides a location prediction method based on multi-resolution context association, including the following steps:

[0054] Step S1: Obtain multi-resolution time information based on the user's historical trajectory data or sequence;

[0055] Step S2: Obtain multi-resolution spatial information based on the user's historical trajectory data or sequence;

[0056] Step S3: Based on the acquired multi-resolution temporal and spatial information, obtain the multi-resolution temporal and spatial contextual patterns of the sequence;

[0057] Step S4: Obtain location prediction information based on the multi-resolution temporal and spatial context patterns of the acquired sequence.

[0058] The location prediction method based on multi-resolution context management provided by this invention obtains richer user temporal and spatial sequence behavior patterns through multi-resolution temporal and spatial information, thereby improving the accuracy of location prediction.

[0059] The sequence is a historical trajectory sequence composed of the user's historical check-in records, sorted by time.

[0060] In one embodiment, prior to step S1, the following steps are further included:

[0061] Step S0: Obtain user historical trajectory data or sequences within a preset time period.

[0062] In a more specific embodiment, step S0 is implemented by acquiring the user's historical trajectory data over a past period of time.

[0063] In one embodiment, step S1 specifically includes the following steps:

[0064] Step S10: Based on the user's historical trajectory data, obtain the time features represented by multi-level resolution time windows.

[0065] In a more specific embodiment, step S10 is implemented by obtaining time features represented by three levels of time windows based on the user's historical trajectory data, starting from the time factor. The time window sizes are 4 hours, 6 hours, and 12 hours, respectively, and the time features corresponding to each resolution are represented as S1, s2, and s3. The multi-level resolution in the time features refers to different time window sizes. That is, the multi-level resolution time features are time features with different time window sizes. In other embodiments of this application, the time window size can also be implemented as other different hourly time windows.

[0066] In one embodiment, step S2 specifically includes the following steps:

[0067] Step S20: Based on the user's historical trajectory data or sequence, obtain spatial features represented by geographic coordinates and multi-level resolution regions.

[0068] In a more specific embodiment, step S20 is specifically implemented by, based on spatial factors and user historical trajectory data, obtaining spatial features including representations in specific geographic coordinates and three-level resolution regions. The region division method involves dividing the area into grids of the same size according to a Tile Map System. Different grid sizes correspond to different division levels. In this application, the grid levels are 10, 12, and 14, respectively. The spatial features represented by geographic coordinates and multi-level resolution regions are represented as l, v1, v2, and v3, respectively. The multi-level resolution in the spatial features refers to different grid sizes; that is, multi-level resolution spatial features are spatial features with different grid sizes. In other embodiments of this application, the division levels can also be implemented as other different division levels.

[0069] In one embodiment, please refer to Figure 2 Step S3 specifically includes the following steps:

[0070] Step S31: Obtain the fusion hidden layer state value based on the temporal features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions.

[0071] Step S32: Based on the obtained spatial features represented by geographic coordinates, obtain multi-resolution spatial context rules.

[0072] In one embodiment, step S31 specifically includes the following steps:

[0073] Step S311: Input the temporal features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions into the embedding expression layer to obtain the corresponding embedding expression. e l ,

[0074] Step S312: The corresponding embedded expression e l , The input is fed into a multi-resolution spatiotemporal feature concatenation layer to obtain the concatenated vector of the corresponding embedded representation.

[0075] Step S313: Concatenate the vectors The input is fed into an RNN network for sequence pattern learning, and the updated state values ​​of the hidden layers of the RNN network are obtained. The formula for updating the hidden layer state values ​​of the RNN network is as follows:

[0076]

[0077] In the formula, W represents the activation function.l W h and b h These are the parameters to be learned. It is t in RNN network i The hidden layer state value at time t.

[0078] Step S314: Obtain the fused hidden layer state value based on the obtained update state value of the RNN network hidden layer.

[0079] In a more specific embodiment, step S314 is specifically implemented as follows:

[0080] The fused hidden layer state value is obtained by weighted averaging the hidden layer state values ​​of all points representing spatial features in geographic coordinates according to the following formula.

[0081]

[0082] in, This refers to the weighting coefficient, i.e., the spatial correlation coefficient ω. For t in an RNN network i The hidden layer state value at time i, where j is the time in the sequence before time i.

[0083] In one embodiment, step S32 specifically includes the following steps:

[0084] The spatial correlation coefficient ω between two adjacent points in a spatial feature represented by geographic coordinates is obtained using the following formula:

[0085] ω=e -Δd ;

[0086] Where Δd is the absolute spatial distance between two adjacent points in a spatial feature represented by geographic coordinates, calculated from geographic coordinates, and ω is the spatial correlation coefficient.

[0087] In one embodiment, step S4 specifically includes the following steps:

[0088] Step S40: Merge the hidden layer state values Input is fed into a fully connected layer for personalized location prediction:

[0089]

[0090] Among them, e u W represents the embedded expression of user information. PThese are the parameters that need to be learned. The softmax() function is used for normalization. P is the probability result of the user visiting each location in the location database at the next time step, or the probability distribution of the probability of the user visiting each location in the location database. Obtaining the probability of the user visiting each location in the location database and sorting it is equivalent to obtaining a list of the most likely locations in the location database that the user will visit.

[0091] Based on the same inventive concept, please refer to Figure 4 This invention provides a location prediction system based on multi-resolution context association, comprising:

[0092] The multi-resolution time information acquisition module 100 is used to acquire multi-resolution time information based on the user's historical trajectory data.

[0093] The multi-resolution spatial information acquisition module 200 is used to acquire multi-resolution spatial information based on the user's historical trajectory data.

[0094] The time and space context pattern acquisition module 300 is communicatively connected to the multi-resolution time information acquisition module and the multi-resolution space information acquisition module, and is used to acquire the multi-resolution time and space context patterns of the sequence based on the acquired multi-resolution time habits and multi-resolution space information;

[0095] The location prediction information acquisition module 400 is communicatively connected to the time and space context pattern acquisition module, and is used to acquire location prediction information based on the acquired multi-resolution time and space context patterns.

[0096] In one embodiment, please refer to Figure 5 The time and space context pattern acquisition module 300 includes:

[0097] The fusion hidden layer state value acquisition module 310 is communicatively connected to the multi-resolution time information acquisition module and the multi-resolution spatial information acquisition module, and is used to acquire the fusion hidden layer state value based on the time features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions.

[0098] The spatial context pattern acquisition unit 320 is communicatively connected to the multi-resolution spatial information acquisition module and is used to acquire multi-resolution spatial context patterns based on the acquired spatial features represented by geographic coordinates.

[0099] Based on the same inventive concept, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements all or part of the method steps of the above method.

[0100] The present invention can implement all or part of the processes in the above methods, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0101] Based on the same inventive concept, embodiments of this application also provide an electronic device, including a memory and a processor. The memory stores a computer program that runs on the processor. When the processor executes the computer program, it implements all or part of the method steps described above.

[0102] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting all parts of the computer device through various interfaces and lines.

[0103] Memory can be used to store computer programs and / or modules. The processor performs various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. Memory can primarily include a program storage area and a data storage area. The program storage area can store the operating system and at least one application program required for a function (e.g., sound playback, image playback, etc.); the data storage area can store data created based on the use of the mobile phone (e.g., audio data, video data, etc.). Furthermore, memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, SmartMedia Cards (SMC), Secure Digital (SD) cards, Flash Cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0104] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, servers, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0105] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), servers, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0106] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A location prediction method based on multi-resolution context association, characterized in that, Includes the following steps: Step S1: Obtain multi-resolution time information based on the user's historical trajectory data; Step S2: Obtain multi-resolution spatial information based on the user's historical trajectory data; Step S3: Based on the acquired multi-resolution temporal and spatial information, obtain the multi-resolution temporal and spatial contextual patterns of the sequence; Step S4: Obtain location prediction information based on the multi-resolution temporal and spatial context patterns of the acquired sequence; Step S1 specifically includes the following steps: Step S10: Based on the user's historical trajectory data, obtain the time features represented by multi-level resolution time windows; Step S2 specifically includes the following steps: Step S20: Based on the user's historical trajectory data, obtain spatial features represented by geographic coordinates and multi-level resolution regions; Step S3 specifically includes the following steps: Step S31: Obtain the fusion hidden layer state value based on the temporal features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions. Step S32: Based on the obtained spatial features represented by geographic coordinates, obtain multi-resolution spatial context rules; Step S31 specifically includes the following steps: Step S311: Input the temporal features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions into the embedding expression layer to obtain the corresponding embedding expression; Step S312: Input the corresponding embedding expression into the multi-resolution spatiotemporal feature connection layer to obtain the concatenated vector of the corresponding embedding expression; Step S313: Input the concatenated vector into the RNN network for sequence pattern learning to obtain the updated state values ​​of the hidden layers of the RNN network; wherein, the formula for updating the hidden layer state values ​​in the RNN network is: ； In the formula, This represents the activation function. , and These are the parameters to be learned. In RNN networks The hidden layer state value at time t; Step S314: Obtain the fused hidden layer state value based on the obtained update state value of the RNN network hidden layer; Step S32 specifically includes the following steps: The spatial correlation coefficient between two adjacent points in a spatial feature represented by geographic coordinates can be obtained using the following formula: ； in, It is the absolute spatial distance between two adjacent points calculated from their geographic coordinates. Spatial correlation coefficient; Step S314 is specifically implemented as follows: The fused hidden layer state value is obtained by weighted averaging the hidden layer state values ​​of all points representing spatial features in geographic coordinates according to the following formula. : ； in, The weighting coefficient, i.e., the spatial correlation coefficient. , In RNN networks The hidden layer state value at time j, where j is the number of hidden layers in the sequence. A moment before that moment.

2. The location prediction method based on multi-resolution context association as described in claim 1, characterized in that, Before step S1, the following steps are also included: Step S0: Obtain user's historical trajectory data.

3. The location prediction method based on multi-resolution context association as described in claim 1, characterized in that, Step S4 specifically includes the following steps: Step S40: Obtain location prediction information based on the following formula, the obtained fusion hidden layer state values, and the spatial correlation coefficient between two adjacent points: ； in, This represents the embedded expression of user information. These are parameters that need to be learned. This represents the fused hidden layer state values, and the softmax() function is used for normalization.

4. A location prediction system based on multi-resolution context association, employing the location prediction method based on multi-resolution context association as described in any one of claims 1-3, characterized in that, include: The multi-resolution time information acquisition module is used to acquire multi-resolution time information based on the user's historical trajectory data; The multi-resolution spatial information acquisition module is used to acquire multi-resolution spatial information based on the user's historical trajectory data. The time and space context pattern acquisition module is communicatively connected to the multi-resolution time information acquisition module and the multi-resolution space information acquisition module, and is used to acquire the multi-resolution time and space context patterns of the sequence based on the acquired multi-resolution time habits and multi-resolution space information; The location prediction information acquisition module is communicatively connected to the time and space context pattern acquisition module, and is used to acquire location prediction information based on the acquired multi-resolution time and space context patterns.

5. The location prediction system based on multi-resolution context association as described in claim 4, characterized in that, The time and space context pattern acquisition module includes: The fusion hidden layer state value acquisition module is communicatively connected to the multi-resolution time information acquisition module and the multi-resolution spatial information acquisition module, and is used to acquire the fusion hidden layer state value based on the time features represented by multi-level resolution time windows and the spatial features represented by geographic coordinates and multi-level resolution regions. The spatial context pattern acquisition unit is communicatively connected to the multi-resolution spatial information acquisition module and is used to acquire multi-resolution spatial context patterns based on the acquired spatial features represented by geographic coordinates.

Images