WO2017185462A1 - 位置推荐方法及系统 - Google Patents
位置推荐方法及系统 Download PDFInfo
- Publication number
- WO2017185462A1 WO2017185462A1 PCT/CN2016/084088 CN2016084088W WO2017185462A1 WO 2017185462 A1 WO2017185462 A1 WO 2017185462A1 CN 2016084088 W CN2016084088 W CN 2016084088W WO 2017185462 A1 WO2017185462 A1 WO 2017185462A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- location
- user
- data
- vector
- attribute
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
Definitions
- the present invention relates to the field of mobile internet application technologies, and in particular, to a location recommendation method and system.
- Internet-based personalized recommendation is basically based on user historical behavior data (for example) , including purchase behavior, click behavior, collection behavior, review behavior, etc.) + current content scenarios (for example, content being queried, books viewed, SNS friends being exchanged, etc.) for context recommendation, it can be said that Internet personalized recommendation Mainly focused on the online service itself, there is not much consideration for the user's location factors and other related factors (such as personal preferences).
- Location service application convergence results in location-based social networks, referred to as location social networks.
- location social networks users can establish social links to check-in their location, especially some interested locations, via mobile devices with location capabilities (such as mobile phones with GPS). Such as restaurants, shops, museums, etc.; write reviews to share experiences of visiting (check-in) locations.
- location social networks contain a large number of locations
- location-based recommendation techniques can make it easier for users to find locations that match their preferences. Therefore, location recommendation is conducive to people exploring new areas in the city and improving the quality of urban life. Especially when the user is in an unfamiliar environment. You can also use location recommendations to discover potential customers for physical stores and provide relevant ads to those customers, prompting customers to visit the store, thereby increasing the profitability of the business. Location recommendations can also recommend travel routes that match your preferences, helping users avoid information overload, save travel schedules, and increase travel willingness to drive tourism. Because location recommendations can bring convenience and benefits to the general public, businesses, and travelers, it is especially important to position users for location recommendations in location social networks.
- the location result obtained by the location recommendation method according to the traditional collaborative filtering idea may be far away from the current location of the user and cannot be immediately accessed, or is similar to the current location attribute of the user, and is unlikely to be accessed continuously, thereby affecting the enthusiasm of the user, that is, the traditional location.
- the recommendation system does not consider the impact of factors such as the user's current location information and location attributes on the user.
- a location recommendation method including:
- the first K recommended locations are selected and sent to the real-time recommendation engine according to the order.
- the method further comprises:
- the real-time recommendation engine performs location recommendation to a terminal device.
- the user track record includes user location data acquired in real time and access time data of the location.
- the statistical inference method is a maximum likelihood estimation idea for inferring the most likely inter-location transfer relationship data, wherein:
- the probability function representing the probability value of user u moving from position l to position i is:
- the symbol ⁇ ,> represents the inner product of two vectors; the logic function Normalization; vector v u represents the preference vector of user u; vector v l represents the attribute vector of the current position l; vector v iu represents the attribute vector of the next position i interacting with user u; and vector v il represents the next The attribute vector of position i interacting with the current position l.
- the statistical reasoning method comprises:
- a maximum likelihood estimation is performed on the target data set Ds to obtain an optimization function:
- a location recommendation system that includes:
- a data processing module configured to acquire a location data set in a geographic attribute range from one or more user track records, and extract location transfer relationship data according to the location data set to form a data transfer relationship between locations a set of target data sets;
- a data modeling and solving module configured to calculate a probability of occurrence of transfer relationship data between each location from the target data set by using a statistical inference method
- the recommended location analysis module is configured to select the top K recommended locations based on the occurrence probability of the calculated transfer relationship data and the current location of the user, and send them to the real-time recommendation engine according to the order.
- the system further comprises:
- a data acquisition module configured to acquire the user track record from one or more terminal devices.
- the user track record includes user location data acquired in real time and access time data of the location.
- the statistical inference method is a maximum likelihood estimation idea for inferring the most likely inter-location transfer relationship data, wherein:
- the probability function representing the probability value of user u moving from position l to position i is:
- the symbol ⁇ ,> represents the inner product of two vectors; the logic function Normalization; vector v u represents the preference vector of user u; vector v l represents the attribute vector of the current position l; vector v iu represents the attribute vector of the next position i interacting with user u; and vector v il represents the next The attribute vector of position i interacting with the current position l.
- the statistical reasoning method comprises:
- a maximum likelihood estimation is performed on the target data set Ds to obtain an optimization function:
- the integrated user analyzes and solves the trajectory data of the location access and the transfer relationship data between the generated locations, so as to implement the next location recommendation, and the recommendation result is more personalized and more suitable for the user. It is easy to accept because of the demand.
- Figure 1 is a flow chart showing a method of a preferred embodiment of the position recommending method of the present invention.
- Figure 2 is a schematic diagram of a user track.
- FIG. 3 is a schematic diagram showing the hardware system architecture of the preferred embodiment of the location recommendation method of the present invention.
- FIG. 4 is a schematic diagram showing functional blocks of a preferred embodiment of the position recommendation system of the present invention.
- FIG. 1 is a flowchart of a method for a preferred embodiment of a location recommendation method provided by the present invention.
- the order of execution of the steps in the flowchart shown in the figure may be changed according to different requirements, and some steps may be omitted.
- the location recommendation method includes:
- the terminal device obtains the user location data and the access time data of the location in real time, and records the acquired data into a user track record set, and transmits the user track record set to the server.
- the terminal device may be a user's handheld electronic device, such as a user's smart phone, a notebook computer, a smart wearable device, and the like.
- the smart wearable device can implement full or partial functions without relying on the smart phone.
- the smart wearable device may be a smart watch or smart glasses or the like.
- the smart wearable device can also focus on only one type of application function, and needs to be used together with other devices, such as a smart phone, to perform various physical condition monitoring.
- the smart wearable device can be a smart bracelet, smart jewelry, or the like.
- the terminal device has a location acquisition function.
- the terminal device first needs to detect whether the user is allowed to acquire the geolocation information of the user. Obtaining geolocation information may infringe the privacy of the user, so the location acquisition function of the terminal device is not available unless the user agrees.
- the terminal device can pop up a dialog box on its user interface to indicate whether to allow the user to obtain geolocation information. If the user chooses to obtain the geolocation information, the location acquisition function of the terminal device is available. Otherwise, if the user chooses not to obtain the geolocation information, the location acquisition function of the terminal device is unavailable.
- the location acquisition function may acquire user location data based on a GPS (Global Positioning System) location.
- the GPS-based positioning is performed by measuring a distance between a satellite at a known location to a user receiver (eg, a GPS module installed on a user's terminal device), and then calculating the receiver by integrating data of a plurality of satellites. The specific location.
- a GPS Global Positioning System
- the location obtaining function may also be based on the base station positioning of the communication carrier to obtain user location data.
- the base station location is generally applied to a mobile phone user, and the mobile phone base station location service is also called a location based service (LBS).
- the communication carrier's network (such as GSM network, Global System for Mobile Communication) obtains the location information (eg, latitude and longitude coordinates) of the mobile terminal user's mobile phone SIM card (Subscriber Identity Module), on an electronic map platform, such as Google.
- the docking service is supported by the map service, and the location information is displayed on the electronic map to achieve the purpose of positioning.
- the location acquisition function may also be based on assisted global satellite positioning system (AGPS, AssistedGPS) positioning to obtain user location data.
- AGPS assisted global satellite positioning system
- AssistedGPS assisted global satellite positioning system
- the AGPS positioning utilizes the communication base station information to assist the GPS module to perform mobile phone positioning, so that the location of the base station is used to provide location information, reduce the positioning blind zone, and reduce the dependence of the GPS module on the satellite in a place where there is no GPS signal in the covered room.
- the location obtaining function may also be a positioning method using Wi-Fi in a small range or any other location acquiring technology that has been used or not yet developed.
- the terminal device has a timing function to record the access time of the user in a certain geographical location.
- the data record includes the user location data, such as location name (eg, Tiananmen Square, Forbidden City, etc.) and data such as latitude and longitude information, and access time data of the location. Further, in other embodiments of the present invention, the location data may also include categories to which the location belongs (eg, dining, travel, accommodation, etc.).
- the data record can be displayed on an electronic map, such as Google Maps, Baidu Map, Gaode Map, and the like. As shown in Figure 2, on February 23, 2016, a user visited the front door, Tiananmen Square, and the Palace Museum from about 9 o'clock in the morning to about 5 o'clock in the afternoon. Things, Jingshan Park and other locations.
- Google Maps Baidu Map
- Gaode Map Gaode Map
- the server acquires a location data set within a geographic attribute range from all received user track record sets.
- Each terminal device after obtaining the permission of the user, transmits the user track record set composed of the location data to the server through the communication unit of the terminal device after acquiring the location data of the user. Therefore, the server can acquire a set of user track records of a plurality of users.
- the communication unit may be a wireless communication module, including a Wi-Fi module, a WiMax (World Interoperability for Microwave Access) module, and a GSM (Global System for Mobile Communication) module.
- CDMA Code Division Multiple Access
- CDMA2000 includes CDMA2000, CDMA, CDMA2000 1x evdo, WCDMA, TD-SCDMA, etc., ITE (Long Term Evolution), HiperLAN (high-performance radio local area network, High-performance wireless LAN) and more.
- the server records the location data of all the user track record sets, and classifies all the location data by using the same geographic attribute, for example, by the attribute of the city, and obtains a certain city, for example, a certain city.
- the size of the selected geographical range will affect the accuracy of the data size and location recommendation.
- the location data is collected by the city.
- other geographic attributes may also be used, such as a location (such as the current location of user A), and activities of 500 meters, 1000 meters, or 2000 meters. Within the geographic range of the radius.
- the server extracts the inter-location transfer relationship data according to the location data set, thereby constructing a target data set Ds composed of inter-location transfer relationship data.
- the acquired location data set includes user access location data (u, l, t), where u represents the user, l represents the location, and t represents the access time.
- the inter-location transfer relationship data is (u, l, i), where i is a transfer position, which represents a user's transfer from position l to position i, wherein the selection principle of position i is ti-tl ⁇ T, ie, from The time at which the position l reaches the position i is less than the preset time T.
- the time from the position 1 to the position i may be the time calculated by using the navigation function of the electronic map.
- the current Baidu map, Gaode map, Google map, etc. can calculate the approximate time from the starting place to the destination according to the determined starting place and destination, including walking time and public transportation time. And self-driving time.
- the preset time T may be walking time, riding public transportation time or self-driving time, and the like.
- the user can set the personalized requirement of the preset time T through a setting interface provided by the user's terminal device, such as within one hour of self-driving, within twenty minutes of walking, direct access by public transportation, and time is half an hour. Inside, at most once, and within an hour, and so on.
- user A stays at t1 for t1 minutes, l2 for t2 minutes, l3 for t3 minutes, and l4 for t1 minutes.
- User B stayed at t1 for t4 minutes, stayed at t5 for t2 minutes, stayed at l6 for t5 minutes, and stayed at l4 for t6 minutes;
- user C stayed at l7 for t7 minutes, staying at l2 T8 minutes, stayed at the l8 position for t9 minutes, Stayed at t9 for t1 minutes.
- the user access location data includes (uA, l1, t1), (uA, l2, t2), (uA, l3, t3), (uA, l4, t1), (uB, l1, T4), (uB, l5, t2), (uB, l6, t5), (uB, l4, t6), (uC, l7, t7), (uC, l2, t8), (uC, l8, t9) , (uC, l9, t1).
- the extracted inter-position transfer relationship data may include (uA, l1, l5), (uA, l1, l6), (uA, l2, l5), (uA, l2, l4), and the like.
- the server uses a statistical inference method to calculate an occurrence probability of the transfer relationship data between each location from the target data set Ds.
- the statistical reasoning method is a maximum likelihood estimation idea, that is, the most reasonable reasoning is to maximize the possibility of occurrence of existing inter-location transfer relationship data.
- the probability function representing the probability value of user u moving from position l to position i is:
- the symbol ⁇ ,> represents the inner product of two vectors; the logic function The normalization effect; the vector v u represents the preference vector of the user u, such as the location category that the user u likes; the vector v l represents the attribute vector of the current position l, such as the position name of the position l, the category, the latitude and longitude, etc.; next Representative iu a position i attribute vector u interacting with the user, such as user u position Favor i; in and the vector v il represents a position i and the current attribute vector position l interaction, such as the transfer from the current position l to the next The probability of a position i.
- U represents a matrix composed of all user preference vectors v u
- L1 represents a matrix composed of attribute vectors v l of all current positions 1
- Liu represents a matrix composed of attribute vectors v iu in which all next positions i interact with user u
- Lil represents a matrix consisting of attribute vectors v il that interact with the current position l by all next positions i.
- maximum likelihood estimation is performed on the target data set Ds to obtain an optimization function:
- calculation solution algorithm is as follows:
- the set is also the algorithm output.
- the process is to continuously select a (u, l, i) transfer relationship data from the set Ds, and modify the corresponding parameters according to the gradient rising idea. Until the function converges.
- the server selects multiple (for example, the first K) according to the occurrence probability of the calculated transfer relationship data (u, l, i) and the current position of the user according to the transfer position i in the transfer relationship data.
- the recommended locations are sorted according to their probability of occurrence, and are sent to the real-time recommendation engine, and the real-time recommendation engine performs location recommendation to the terminal device.
- FIG. 1 details the location recommendation method of the present invention.
- the hardware system architecture for implementing the above location recommendation method and the functional modules of the software system for implementing the location recommendation method are respectively described below in conjunction with FIGS. 3 to 4.
- FIG. 3 it is a hardware system architecture diagram of a preferred embodiment of the location recommendation method according to the present invention.
- the location recommendation method is implemented by two major Part of the composition:
- the terminal device 1 may be a user's handheld electronic device, such as a user's smart phone, a notebook computer, a smart wearable device, and the like.
- the smart wearable device can implement full or partial functions without relying on the smart phone.
- the smart wearable device may be a smart watch or smart glasses or the like.
- the smart wearable device can also focus on only one type of application function, and needs to be used together with other devices, such as a smart phone, to perform various physical condition monitoring.
- the smart wearable device can be a smart bracelet, smart jewelry, or the like.
- the terminal device 1 includes a processor 11, a location acquisition unit 12, an electronic map unit 13, a timing unit 14, and a communication unit 15. It should be understood that the terminal device 1 may also include other hardware or software, such as a storage device, a display screen, a camera, etc., and is not limited to the components listed above.
- the location obtaining unit 12 may be a GPS (Global Positioning System) module that acquires user location data based on GPS system positioning.
- the GPS positioning is to measure the distance between a satellite at a known location to a user receiver (such as a GPS module of a user terminal device), and then integrate the data of multiple satellites to calculate the specific location of the receiver.
- the location obtaining unit 12 may also be a base station positioning module that acquires user location data based on the base station location of the communication carrier.
- the base station positioning is based on the positioning mode of the communication operator's signal tower, and the latitude and longitude information of the SIM (Subscriber Identity Module) card is obtained through the signal tower, and the location is clicked through calculation.
- the electronic map unit 13, such as the Google Maps service, is docked and displayed on the electronic map unit 13 to achieve the purpose of positioning.
- the location obtaining unit 12 may also be a combination of a GPS module and a base station positioning module, which acquires user location data based on AGPS (Assisted GPS).
- AGPS Assisted GPS
- the AGPS uses the communication base station information to assist the GPS module to perform mobile phone positioning, so as to use the base station positioning to provide location information and narrow the positioning blind area in a place where there is no GPS signal indoors.
- the location obtaining unit 12 may also be a Wi-Fi module that implements positioning in a small range using Wi-Fi.
- the location obtaining unit 12 may be any other module capable of supporting location acquisition technology that is already in use or not yet developed, and is not limited to the above enumerated.
- the electronic map unit 13, ie, a digital map, is a map that is stored and viewed digitally using computer technology.
- the method of storing information by electronic map generally uses vector image storage, and the map scale can be enlarged, reduced or rotated without affecting the display effect.
- the electronic map unit 13 can acquire global geographic information in combination with satellite images, maps, and search techniques.
- the electronic map unit 13 may be, but is not limited to, a Google map, a Baidu map, a high German map, and the like.
- the timing unit 14 is configured to record the access time of the user in a certain geographical location. For example, as shown in FIG. 2, the timing unit 14 records a user on February 23, 2016 at the front door, Tiananmen Square, the Palace Museum, and Jingshan Park, from about 9 o'clock in the morning to about 2 o'clock in the afternoon. , for about 5 hours of continuous visits.
- the communication unit 15 may be a wireless communication module, including a Wi-Fi module, a WiMax (World Interoperability for Microwave Access) module, and a GSM (Global System for Mobile Communication) module.
- CDMA Code Division Multiple Access
- CDMA2000 includes CDMA2000, CDMA, CDMA2000 1x evdo, WCDMA, TD-SCDMA, etc., ITE (Long Term Evolution), HiperLAN (high-performance radio local area network) , high performance wireless LAN) and so on.
- the communication unit 15 can be used for information exchange between the terminal device 1 and other devices, such as other terminal devices 1 or servers.
- the processor 11 also known as a central processing unit (CPU), is a very large-scale integrated circuit, and is a computing core (Core) and a control unit of the terminal device 1.
- the function of the processor 11 is mainly to interpret program instructions and data in the processing software.
- the processor 11 is connected to the location obtaining unit 12, the electronic map unit 13, the timing unit 14, and the communication unit 15, for controlling the location obtaining unit 12 to acquire the location data of the user and controlling the timing unit 14 to record the user in a certain geographic location.
- the location access time, the location data and the access time data are processed, such as stored in a user track record set, and/or displayed on the electronic map unit 13 and the like.
- the terminal device 1 carried by the user when the user comes to a certain place, the terminal device 1 carried by the user, such as the user's mobile phone, obtains the current location of the user through the location obtaining unit 12, and The longitude and dimensions at this time are saved. Further, the name of the location, such as XX Park, can also be obtained through the electronic map unit 13, and the location of the location is obtained. A category, such as belonging to an attraction, and save the information together. Further, the timing unit 14 of the terminal device 1 also records the access time (ie, the dwell time) of the user at the location. Wherein, the user is denoted as u, the position is denoted as l, and the time is denoted as t.
- the terminal device 1 transmits a user track record set composed of user access location data (including the user u, location l, and time t) to a server system, and the server system performs data processing to obtain a recommendation.
- the location is pushed to the terminal device 1 for reference by the user.
- the server 2 can be a cloud server. Similar to the terminal device 1, the server 2 includes a processor 21, an electronic map unit 23, and a communication unit 25. Furthermore, the server 2 is also equipped with a location recommendation system 10.
- the location recommendation system 10 may include a plurality of functional modules consisting of program segments (see FIG. 4 for details).
- the program code of each program segment in the location recommendation system 10 may be stored in the storage device 26 of the server 2 and executed by the processor 21 of the server 2 to transmit the user track record set to the terminal device 1 Data processing is performed to obtain one or more recommended locations, and the recommended locations are pushed to the terminal device 1 for reference by the user (described in detail in FIG. 4).
- the location recommendation system 10 can be divided into multiple functional modules according to the functions performed by the location recommendation system 10.
- the function module includes: a data acquisition module 100, a data processing module 101, a data modeling solution module 102, and a recommended location analysis module 103.
- the data acquisition module 100 is configured to acquire, from each terminal device 1, a set of user track records collected by each terminal device 1.
- each terminal device 1 acquires user location data and access time data of the location in real time, and records the acquired data into a user track record set, and records the acquired data. Go to a user track record set.
- the terminal device 1 first detects whether the user is allowed to acquire geolocation information. Obtaining geolocation information may violate the privacy of the user, so the location acquisition feature is not available unless the user agrees.
- the terminal device 1 can pop up a dialog box on its user interface to indicate whether or not to obtain the user's geolocation information. If the user chooses to obtain the geolocation information, the location acquisition function of the terminal device 1 is available. Otherwise, if the user chooses not to obtain the geolocation information, the location acquisition function of the terminal device 1 is unavailable.
- the data record includes the user location data, such as location name (eg, Tiananmen Square, Forbidden City, etc.) and data such as latitude and longitude information, and access time data of the location.
- location data may further include a category to which the location belongs (eg, dining, travel, accommodation, etc.).
- the data record can be displayed on an electronic map, such as Google Maps, Baidu Map, Gaode Map, and the like. As shown in Figure 2, on February 23, 2016, a user visited the front door, Tiananmen Square, the Palace Museum, and Jingshan Park from about 9 o'clock in the morning to about 5 o'clock in the afternoon.
- Google Maps Google Maps
- Baidu Map Gaode Map
- Figure 2 As shown in Figure 2, on February 23, 2016, a user visited the front door, Tiananmen Square, the Palace Museum, and Jingshan Park from about 9 o'clock in the morning to about 5 o'clock in the afternoon.
- the data processing module 101 is configured to obtain a location data set within a geographic attribute range from all user track record sets.
- Each terminal device 1 transmits the user track record set composed of the location data to the server by wire or wirelessly after obtaining the location data of the user, when the user's permission is obtained. Therefore, the server can acquire a set of user track records of a plurality of users.
- the server records the location data of all the user track record sets, and classifies all the location data by using the same geographic attribute, for example, by the attribute of the city, to obtain a certain city, for example, a certain user A.
- the size of the selected geographic range affects the accuracy of the data size and location recommendation.
- the location data is collected in units of cities.
- other geographic attributes may also be used, such as a geographic location within a certain location (such as the current location of user A), with an active radius of 500 meters, 1000 meters, or 2000 meters. .
- the data processing module 101 is further configured to extract the inter-location transfer relationship data according to the location data set, thereby constructing a target data set Ds composed of the inter-location transfer relationship data.
- the acquired location data set includes user access location data (u, l, t), where u represents a user, l represents a location, and t represents an access time.
- the inter-location transfer relationship data is (u, l, i), where i is a transfer position, which represents a user's transfer from position l to position i, wherein the selection principle of position i is ti-tl ⁇ T, ie, from The time at which the position l reaches the position i is less than the preset time T.
- the time from the position 1 to the position i may be a navigation function using an electronic map.
- the calculated time For example, the current Baidu map, Gaode map, Google map, etc. can calculate the approximate time from the starting place to the destination according to the determined starting place and destination, including walking time and public transportation time. And self-driving time.
- the preset time T may be walking time, riding public transportation time or self-driving time, and the like.
- the user can set the personalized requirement of the preset time T through a setting interface provided by the user's terminal device, such as within one hour of self-driving, within twenty minutes of walking, direct access by public transportation, and time is half an hour. Inside, at most once, and within an hour, and so on.
- user A stays at t1 for t1 minutes, l2 for t2 minutes, l3 for t3 minutes, and l4 for t1 minutes.
- User B stayed at t1 for t4 minutes, stayed at t5 for t2 minutes, stayed at l6 for t5 minutes, and stayed at l4 for t6 minutes;
- user C stayed at l7 for t7 minutes, staying at l2 At t8 minutes, it stayed at t8 for t9 minutes and stayed at l9 for t1 minutes.
- the user access location data includes (uA, l1, t1), (uA, l2, t2), (uA, l3, t3), (uA, l4, t1), (uB, l1, T4), (uB, l5, t2), (uB, l6, t5), (uB, l4, t6), (uC, l7, t7), (uC, l2, t8), (uC, l8, t9) , (uC, l9, t1).
- the extracted inter-position transfer relationship data may include (uA, l1, l5), (uA, l1, l6), (uA, l2, l5), (uA, l2, l4), and the like.
- the data modeling and solving module 102 is configured to calculate a probability of occurrence of each of the transfer relationship data from the target data set Ds by using a statistical inference method.
- the statistical reasoning method is a maximum likelihood estimation idea, that is, the most reasonable reasoning is to maximize the possibility of occurrence of existing inter-location transfer relationship data.
- the probability function representing the probability value of user u moving from position l to position i is:
- the symbol ⁇ ,> represents the inner product of two vectors; the logic function The normalization effect; the vector v u represents the preference vector of the user u, such as the location category that the user u likes; the vector v l represents the attribute vector of the current position l, such as the position name of the position l, the category, the latitude and longitude, etc.; next Representative iu a position i attribute vector u interacting with the user, such as user u position Favor i; in and the vector v il represents a position i and the current attribute vector position l interaction, such as the transfer from the current position l to the next The probability of a position i.
- U represents a matrix composed of all user preference vectors v u
- L1 represents a matrix composed of attribute vectors v l of all current positions 1
- Liu represents a matrix composed of attribute vectors v iu in which all next positions i interact with user u
- Lil represents a matrix consisting of attribute vectors v il that interact with the current position l by all next positions i.
- maximum likelihood estimation is performed on the target data set Ds to obtain an optimization function:
- calculation solution algorithm is as follows:
- the set is also the algorithm output.
- the process is to continuously select a (u, l, i) transfer relationship data from the set Ds, and modify the corresponding parameters according to the gradient rising idea. Until the function converges.
- the recommended location analysis module 103 is configured to select the first K recommendations according to the occurrence probability of the calculated transfer relationship data (u, l, i) and the current location of the user according to the transfer location i in the transfer relationship data.
- the locations are sorted according to their probability of occurrence, and are sent to the real-time recommendation engine in the server 2, and the real-time recommendation engine performs location recommendation to the terminal device 1.
- the above described embodiment adopts a server-client mode to perform location recommendation on the user. That is, the location recommendation system is installed in the server, and the server performs data processing and location recommendation actions.
- the application scenario of the mode may be an active location recommendation by using a server through a network. As long as the user has networked and an application is opened, such as when a browser, a group purchase website, etc. is opened, the server performs a location recommendation operation.
- Another application scenario of the mode may be to take the form of the popular WeChat public account.
- the server will perform a location recommendation operation.
- the user may also input a specific keyword, such as “catering”, on the public number platform, and the server may also filter the top K with the keyword characteristics from the analyzed locations. Recommended location for recommendation.
- the location recommendation system 10 may also be installed in any terminal device, and the terminal device performs location recommendation on the user.
- the server can Timing or implementation sends the collected user track record set of other users to the terminal device.
- the location recommendation operation can be implemented when the location recommendation system 10 on the terminal device is turned on. At this time, even if the terminal device cannot connect to the network, the location recommendation operation can be performed based on the user trajectory previously obtained from the server.
- each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
- the above integrated unit can be implemented in the form of hardware or in the form of hardware plus software function modules.
- the above-described integrated unit implemented in the form of a software function module can be stored in a computer readable storage medium.
- the software function modules described above are stored in a storage medium and include instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to perform the methods of the various embodiments of the present invention. Part of the steps.
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Navigation (AREA)
Abstract
一种位置推荐方法,包括:从一个或者多个用户轨迹记录中,获取一个地理属性范围内的位置数据集合(S11);根据所述位置数据集合提取出位置间转移关系数据,构成一个由位置间转移关系数据组成的目标数据集合(S12);采用统计推理方法,从所述目标数据集合中计算出每一个位置间转移关系数据的发生概率(S13);及基于上述计算出来的转移关系数据的发生概率以及用户的当前位置,选择前K个推荐位置,并按其排序发送给实时推荐引擎(S14)。还提供一种位置推荐系统。利用该方法和系统能够使位置推荐结果更加个性化,更符合用户需求。
Description
本申请要求于2016年4月26日提交中国专利局,申请号为201610266263.6、发明名称为“位置推荐方法及系统”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本发明涉及移动互联网应用技术领域,尤其涉及一种位置推荐方法以及系统。
对于移动互联网应用而言,除了传统互联网所积累的海量信息、各种丰富应用可供使用外,移动互联网本身也产生了海量的内容和应用,怎样基于所述海量信息,准确识别地出用户的喜好并在此基础上向用户推荐最相关的产品、服务、信息是移动互联网的主流应用。
目前使用的无论是基于内容过滤(Content-Based filtering)的个性化推荐技术,还是基于协同过滤(Collaborative filtering)的个性化推荐技术,基于互联网的个性化推荐基本上还是基于用户历史行为数据(例如,包括购买行为、点击行为、收藏行为、点评行为等)+当前内容场景(例如,正在查询的内容、浏览的图书、正在交流的SNS好友等)来进行上下文推荐的,可以说互联网个性化推荐主要还是偏重于线上服务本身的,对于用户的位置因素以及其他相关因素(如个人喜好等)并没有太多的考虑。
与基于互联网的个性化推荐相比,基于位置服务的个性化推荐更有意义。位置服务应用融合产生了基于位置的社交网络(location-based social networks),简称位置社交网络。在位置社交网络中,用户可以建立社交链接(social links),通过具有定位功能的移动设备(如带GPS的手机),签到(check-in)自己所在的位置,特别是某些感兴趣的位置,如餐馆、商店、博物馆等;撰写评论以分享访问(签到)位置的经历。
由于位置社交网络包含大量的位置,基于位置服务的推荐技术可以使用户更容易找到符合自己偏好的位置。因此,位置推荐有利于人们探索城市中的新地带,提高城市生活质量。特别是当用户身处在一个陌生环境的时候。也可以利用位置推荐为实体商店发现潜在客户,并为这些客户提供相关广告,促使客户访问商店,从而提高商家的利润。位置推荐还可以为用户推荐符合个人偏好的旅行路线,帮助用户避免信息过载、节约行程安排时间,以及提高旅行意愿,推动旅游业的发展。因为位置推荐能够为大众、商家和旅行者带来便利和利益,所以,在位置社交网络中,为用户进行位置推荐是尤为重要的。
目前,有许多关于位置推荐的技术,但其只是简单地利用空间特征,忽略了时间特征。依据传统协同过滤思想的位置推荐方法得到的位置结果,可能距离用户当前位置非常远而不能马上前往,或者与用户当前位置属性相似而不大可能连续被访问,从而影响用户的积极性,即传统位置推荐系统未考虑用户当前位置信息、位置间属性等因素对用户的影响。
发明内容
鉴于以上内容,有必要提出一种位置推荐方法,能够使位置推荐结果更加个性化,更符合用户需求。
一种位置推荐方法,包括:
从一个或者多个用户轨迹记录中,获取一地理属性范围内的位置数据集合;
根据所述位置数据集合提取出位置间转移关系数据,构成一个由位置间转移关系数据组成的目标数据集合;
采用统计推理方法,从所述目标数据集合中计算出每一个位置间转移关系数据的发生概率;及
基于上述计算出来的转移关系数据的发生概率以及用户的当前位置,选择前K个推荐位置,并按其排序发送给实时推荐引擎。
优选地,该方法还包括:
所述实时推荐引擎向一终端设备进行位置推荐。
优选地,所述用户轨迹记录包括实时获取的用户位置数据以及该位置的访问时间数据。
优选地,所述统计推理方法为最大似然估计思想,用于推理出发生的可能性最大的位置间转移关系数据,其中:
代表用户u从位置l转移到位置i的概率值的概率函数为:
ful(i)=σ(<vu,viu>+<vl,vil>);
其中:
符号<,>代表两向量内积;逻辑函数起归一化作用;向量vu代表用户u的偏好向量;向量vl代表当前位置l的属性向量;向量viu代表下一个位置i与用户u交互的属性向量;以及向量vil代表下一个位置i与当前位置l交互的属性向量。
优选地,所述统计推理方法包括:
将所有可能的vu、vl、viu、vil组成一个参数矩阵Θ={U,Ll,Liu,Lil},其中,U代表由所有用户偏好向量vu组成的矩阵,Ll代表由所有当前位置l的属性向量vl组成的矩阵,Liu代表由所有下一个位置i与用户u交互的属性向量viu组成的矩阵,以及Lil代表由所有下一个位置i与当前位置l交互的属
性向量vil组成的矩阵;
对所述目标数据集合Ds做最大似然估计,得到优化函数:
L(Θ)=argmaxΠσ(<vu,viu>+<vl,vil>);及
对上述优化函数进行计算求解,以不断更新其属性向量值vu,vl,viu及vil,使得函数L(Θ)结果不断增大直至收敛,得到的收敛时的向量值vu,vl,viu及vil。
鉴于以上内容,还有必要提出一种位置推荐方法,能够使位置推荐结果更加个性化,更符合用户需求。
一种位置推荐系统,包括:
数据处理模块,用于从一个或者多个用户轨迹记录中,获取一地理属性范围内的位置数据集合,并根据所述位置数据集合提取出位置间转移关系数据,构成一个由位置间转移关系数据组成的目标数据集合;
数据建模求解模块,用于采用统计推理方法,从所述目标数据集合中计算出每一个位置间转移关系数据的发生概率;及
推荐位置分析模块,用于基于上述计算出来的转移关系数据的发生概率以及用户的当前位置,选择前K个推荐位置,并按其排序发送给实时推荐引擎。
优选地,该系统进一步包括:
数据获取模块,用于从一个或者多个终端设备获取所述用户轨迹记录。
优选地,所述用户轨迹记录包括实时获取的用户位置数据以及该位置的访问时间数据。
优选地,所述统计推理方法为最大似然估计思想,用于推理出发生的可能性最大的位置间转移关系数据,其中:
代表用户u从位置l转移到位置i的概率值的概率函数为:
ful(i)=σ(<vu,viu>+<vl,vil>);
其中:
符号<,>代表两向量内积;逻辑函数起归一化作用;向量vu代表用户u的偏好向量;向量vl代表当前位置l的属性向量;向量viu代表下一个位置i与用户u交互的属性向量;以及向量vil代表下一个位置i与当前位置l交互的属性向量。
优选地,所述统计推理方法包括:
将所有可能的vu、vl、viu、vil组成一个参数矩阵Θ={U,Ll,Liu,Lil},其中,U代表由所有用户偏好向量vu组成的矩阵,Ll代表由所有当前位置l的属性向量vl组成的矩阵,Liu代表由所有下一个位置i与用户u交互的属性向量viu组成的矩阵,以及Lil代表由所有下一个位置i与当前位置l交互的属性向量vil组成的矩阵;
对所述目标数据集合Ds做最大似然估计,得到优化函数:
L(Θ)=argmaxΠσ(<vu,viu>+<vl,vil>);及
对上述优化函数进行计算求解,以不断更新其属性向量值vu,vl,viu及
vil,使得函数L(Θ)结果不断增大直至收敛,得到的收敛时的向量值vu,vl,viu及vil。
相较于现有技术,本发明综合用户对位置访问的轨迹数据及其产生的位置间的转移关系数据,进行建模、求解,以实现下一个位置推荐,推荐结果更加个性化,更符合用户需求而容易被接受。
图1是一个示意图演示了本发明位置推荐方法较佳实施例的方法流程图。
图2为一个用户轨迹示意图。
图3是一个示意图演示了本发明实现所述位置推荐方法较佳实施例的硬件系统架构图。
图4是一个示意图演示了本发明位置推荐系统的较佳实施例的功能模块图。
【主要元件符号说明】
终端设备 1
服务器 2
位置推荐系统 10
处理器 11、21
位置获取单元 12
电子地图单元 13、23
计时单元 14
通讯单元 15、25
存储设备 26
数据获取模块 100
数据处理模块 101
数据建模求解模块 102
推荐位置分析模块 103
如下具体实施方式将结合上述附图进一步说明本发明。
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清除、完整地描述,显然,所描述的实施例仅仅是本发明的一部分实施例,而不是全部的实施例。
基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例,都属于本发明保护的范围。
请参考图1,是本发明提供的位置推荐方法较佳实施例的方法流程图。根据不同的需求,该图所示流程图中步骤的执行顺序可以改变,某些步骤可以省略。
所述位置推荐方法包括:
S10,终端设备在得到用户允许的情况下,实时获取用户位置数据以及该位置的访问时间数据,将获取的数据记录到一个用户轨迹记录集中,并将该用户轨迹记录集传送给服务器。
其中,所述终端设备可以是用户的手持式电子设备,如用户的智能手机,笔记本电脑,智能式穿戴式设备等。
所述智能式穿戴设备可以不依赖智能手机实现完整或者部分的功能。例如,所述的智能式穿戴设备可以是智能手表或智能眼镜等。
所述智能式穿戴设备也可以只专注于某一类应用功能,需要和其它设备,如智能手机配合使用,以进行各种体征监测。例如,所述的智能式穿戴设备可以是智能手环、智能首饰等。
所述终端设备具有位置获取功能。所述终端设备首先需要检测用户是否允许获取该用户的地理定位信息。获取地理定位信息可能侵犯用户的隐私,因此,除非用户同意,否则所述终端设备的位置获取功能是不可用的。
终端设备可以在其用户界面上弹出一个对话框,表明是否允许获取用户的地理定位信息。若用户选择允许获取地理定位信息,则所述终端设备的位置获取功能可用,否则,若用户选择不允许获取地理定位信息,则所述终端设备的位置获取功能不可用。
本发明的一个实施例中,所述位置获取功能可以基于GPS(Global Positioning System,全球定位系统)定位获取用户位置数据。所述基于GPS定位是通过测量出已知位置的卫星到用户接收机(例如,用户的终端设备上安装的GPS模块)之间的距离,然后,综合多颗卫星的数据计算出所述接收机的具体位置。
本发明其他实施例中,所述位置获取功能也可以是基于通讯运营商的基站定位获取用户位置数据。所述基站定位一般应用于手机用户,手机基站定位服务又叫做移动位置服务(LBS,Location Based Service),是通过
通讯运营商的网络(如GSM网,Global System for Mobile Communication)获取移动终端用户的手机SIM卡(Subscriber Identity Module,客户识别模块)的位置信息(如,经纬度坐标),在电子地图平台,如Google地图服务,的支持下进行对接,并将所述位置信息显示到所述电子地图上面,从而达到定位的目的。
本发明其他实施例中,所述位置获取功能也可以是基于辅助全球卫星定位系统(AGPS,AssistedGPS)定位获取用户位置数据。所述AGPS定位是利用通讯基站信息来辅助GPS模块进行手机定位,以在受遮盖的室内没有GPS信号的地方,利用基站定位来提供位置信息,缩小定位盲区,减轻GPS模块对卫星的依赖度。
此外,在本发明其他实施例中,所述位置获取功能也可以是利用Wi-Fi在小范围内的定位方式或者其他任何已经在使用或者尚未开发使用的任何位置获取技术。
进一步地,所述终端设备同时具有计时功能,以记录用户在某个地理位置的访问时间。
其中,所述数据记录包括所述用户位置数据,例如位置名称(例如天安门、故宫等)以及经纬度信息等数据,以及该位置的访问时间数据。进一步地,在本发明的其他实施例中,所述位置数据还可以包括位置所属类别(例如餐饮、旅游、住宿等)。
所述数据记录可以显示在电子地图上,如Google地图、百度地图、高德地图等。如图2所示,某一用户在2016年2月23日,从上午9点钟左右到下午2点钟左右的大概5个小时,连续访问了前门、天安门、故宫博
物院、景山公园等位置。
S11,所述服务器从所接收的所有用户轨迹记录集中获取一个地理属性范围内的位置数据集合。
每一个终端设备在得到了用户的允许的情况下,在获取了该用户的位置数据后,都会通过该终端设备的通讯单元将由所述位置数据组成的用户轨迹记录集传送给所述服务器。因此,所述服务器可以获取多个用户的用户轨迹记录集。
所述通讯单元可以是无线通讯模块,包括Wi-Fi模块,WiMax(World Interoperability for Microwave Access,即全球微波接入互操作性)模块,GSM(Global System for Mobile Communication,全球移动通信系统)模块,CDMA(Code Division Multiple Access,码分多址)包括CDMA2000,CDMA,CDMA2000 1x evdo,WCDMA,TD-SCDMA等等),ITE(Long Term Evolution,长期演进),HiperLAN(high-performance radio local area network,高性能无线局域网)等等。
在本实施例中,所述服务器对所有用户轨迹记录集中的位置数据记录,以一个相同的地理属性,例如,以城市这个属性,对所有位置数据进行分类,并得到某一个城市,例如某一用户A当前所在城市这个地理范围内的所有位置数据集合。
应该了解,在收集位置数据时,所选取的地理范围的大小会影响数据规模和位置推荐的准确度,本实施例以城市为单位进行位置数据收集。在本发明的其他实施例中,也可以采用其他的地理属性,如以某一位置(如用户A当前所在位置)为中心,以500米、1000米或者2000米等为活动
半径的地理范围内。
S12,所述服务器根据所述位置数据集合提取出位置间转移关系数据,从而构成一个由位置间转移关系数据组成的目标数据集合Ds。
在本实施例中,上述获取的位置数据集合中包括用户访问位置数据(u,l,t),其中,u代表用户,l代表位置,t代表访问时间。所述位置间转移关系数据为(u,l,i),其中,i是转移位置,其代表用户从位置l转移到位置i,其中,位置i的选择原则为ti-tl<T,即从位置l到达位置i的时间小于预设时间T。
本实施例中,从位置l到位置i的时间可以是利用电子地图的导航功能所计算出来的时间。例如,现在的百度地图、高德地图、Google地图等都能够根据确定的起始地及目的地计算从所述起始地到所述目的地的大概时间,包括步行时间、乘坐公共交通工具时间及自驾时间等。
根据不同的系统设置或者用户设置,所述预设时间T可以是步行时间、乘坐公共交通工具时间或者自驾时间等。例如,用户可以通过该用户的终端设备提供的一个设置界面设置预设时间T的个性化需求,如自驾一个小时之内、步行二十分钟之内、公共交通工具直达且时间在半个小时之内、最多转车一次且时间在一个小时之内,等等。
在其中一个示意性的例子中,根据所述用户轨迹记录集,用户A在l1位置停留了t1分钟,在l2位置停留了t2分钟,在l3位置停留了t3分钟,在l4位置停留了t1分钟;用户B在l1位置停留了t4分钟,在l5位置停留了t2分钟,在l6位置停留了t5分钟,在l4位置停留了t6分钟;用户C在l7位置停留了t7分钟,在l2位置停留了t8分钟,在l8位置停留了t9分钟,
在l9位置停留了t1分钟。
因此,根据上述例子,所述用户访问位置数据包括(uA,l1,t1)、(uA,l2,t2)、(uA,l3,t3)、(uA,l4,t1)、(uB,l1,t4)、(uB,l5,t2)、(uB,l6,t5)、(uB,l4,t6)、(uC,l7,t7)、(uC,l2,t8)、(uC,l8,t9)、(uC,l9,t1)。所提取出的位置间转移关系数据可以包括(uA,l1,l5)、(uA,l1,l6)、(uA,l2,l5)、(uA,l2,l4),等等。
S13,所述服务器采用统计推理方法,从所述目标数据集合Ds中计算出每一个位置间转移关系数据的发生概率。
在本实施例中,所述统计推理方法为最大似然估计思想,即最合理的推理是使已有的位置间转移关系数据发生的可能性最大。其中,代表用户u从位置l转移到位置i的概率值的概率函数为:
ful(i)=σ(<vu,viu>+<vl,vil>)。
其中:
符号<,>代表两向量内积;逻辑函数起归一化作用;向量vu代表用户u的偏好向量,如用户u喜欢的位置类别;向量vl代表当前位置l的属性向量,如位置l的位置名称、所属类别、经纬度等;向量viu代表下一个位置i与用户u交互的属性向量,如用户u对位置i的喜欢程度;以及向量vil代表下一个位置i与当前位置l交互的属性向量,如从当前位置l转移到下一位置i的概率。
所有可能的vu、vl、viu、vil可以组成一个参数矩阵Θ={U,Ll,Liu,Lil}。
其中:
U代表由所有用户偏好向量vu组成的矩阵,Ll代表由所有当前位置l的属性向量vl组成的矩阵,Liu代表由所有下一个位置i与用户u交互的属性向量viu组成的矩阵,以及Lil代表由所有下一个位置i与当前位置l交互的属性向量vil组成的矩阵。
在本实施例中,对所述目标数据集合Ds做最大似然估计,得到优化函数:
L(Θ)=argmaxΠσ(<vu,viu>+<vl,vil>)。
进一步对上述优化函数进行计算求解,以不断更新其属性向量值vu,vl,viu及vil,使得函数L(Θ)结果不断增大直至收敛,得到的收敛时的向量值vu,vl,viu及vil即为最优结果。
详细地,所述计算求解算法如下:
在上述求解算法中,所有位置间转移关系数据(u,l,i)组成集合Ds作为算法输入,其对应参数集合为Θ={U,Ll,Liu,Lil},当优化函数收敛时,参数集合亦是算法输出。过程则是不断从集合Ds中随机选取一个(u,l,i)转移关系数据,并依据梯度上升思想对相应参数做修改,既直至函数收敛。
利用每一个位置间转移关系数据(u,l,i)对应的参数θ,以及上述的概率函数ful(i)=σ(<vu,viu>+<vl,vil>),可以计算出每一个位置间转移关系数据(u,l,i)的发生概率。
S14,所述服务器基于上述计算出来的转移关系数据(u,l,i)的发生概率以及用户的当前位置,根据所述转移关系数据中的转移位置i,选择多个(例如前K个)推荐位置,并按其发生概率从大到小进行排序后,发送给实时推荐引擎,由所述实时推荐引擎向终端设备进行位置推荐。
至此,完成一次位置推荐过程。
上述图1详细介绍了本发明的位置推荐方法,下面结合第3~4图,分别对实现上述位置推荐方法的硬件系统架构以及实现该位置推荐方法的软件系统的功能模块进行介绍。
应该了解,该实施例仅为说明之用,在专利申请范围上并不受此结构的限制。
如图3所示,为本发明实现所述位置推荐方法较佳实施例的硬件系统架构图。
在本发明的其中一个较佳实施例中,所述位置推荐方法的实现由两大
部分构成:
一、收集用户位置数据的一台或者多台终端设备1
其中,所述终端设备1可以是用户的手持式电子设备,如用户的智能手机,笔记本电脑,智能式穿戴式设备等。
所述智能式穿戴设备可以不依赖智能手机实现完整或者部分的功能。例如,所述智能式穿戴设备可以是智能手表或智能眼镜等。其他实施例中,所述智能式穿戴设备也可以只专注于某一类应用功能,需要和其它设备,如智能手机配合使用,以进行各种体征监测。例如,所述的智能式穿戴设备可以是智能手环、智能首饰等。
所述终端设备1包括处理器11、位置获取单元12、电子地图单元13、计时单元14以及通讯单元15。应该了解,所述终端设备1也可以包括其他硬件或者软件,例如,存储设备、显示屏幕、摄像头等,而并不限制于上述列举的部件。
在本发明的其中一个实施例中,所述位置获取单元12可以是GPS(Global Positioning System,全球定位系统)模块,其基于GPS系统定位获取用户位置数据。所述GPS定位是测量出已知位置的卫星到用户接收机(如用户终端设备的GPS模块)之间的距离,然后综合多颗卫星的数据计算出接收机的具体位置。
在本发明的其他实施例中,所述位置获取单元12也可以是基站定位模块,其基于通讯运营商的基站定位获取用户位置数据。所述基站定位是基于通讯运营商信号塔的定位方式,通过信号塔获取到手机SIM(Subscriber Identity Module客户识别模块)卡的经纬度信息,通过计算将该位置点通
过与电子地图单元13,如Google地图服务,进行对接显示到电子地图单元13上,达到定位的目的。
在本发明的其他实施例中,所述位置获取单元12也可以是GPS模块与基站定位模块的组合,其基于AGPS(AssistedGPS:辅助全球卫星定位系统)定位获取用户位置数据。所述AGPS利用通讯基站信息来辅助GPS模块进行手机定位,以在室内没有GPS信号的地方利用基站定位来提供位置信息,缩小定位盲区。
此外,在本发明的其他实施例中,所述位置获取单元12也可以是Wi-Fi模块,其利用Wi-Fi实现在小范围内的定位。或者所述位置获取单元12也可以是其他任何已经在使用或者尚未开发使用的能够支持位置获取技术的任何模块,而不限于上述所列举的。
所述电子地图单元13即数字地图,是利用计算机技术,以数字方式存储和查阅的地图。电子地图储存资讯的方法,一般使用向量式图像储存,地图比例可放大、缩小或旋转而不影响显示效果。
所述电子地图单元13结合卫星图片、地图,以及搜索技术可以获取全球地理信息。
在本发明的其中一个实施例中,所述电子地图单元13可以是,但并不限制于,Google地图、百度地图、高德地图等。
所述计时单元14用于记录用户在某个地理位置的访问时间。例如,图2所示,所述计时单元14记录了某个用户于2016年2月23日,在前门、天安门、故宫博物院、景山公园等位置,从上午9点钟左右到下午2点钟左右,连续访问了大概5个小时。
所述通讯单元15可以是无线通讯模块,包括Wi-Fi模块,WiMax(World Interoperability for Microwave Access,即全球微波接入互操作性)模块,GSM(Global System for Mobile Communication,全球移动通信系统)模块,CDMA(Code Division Multiple Access,码分多址)包括CDMA2000,CDMA,CDMA2000 1x evdo,WCDMA,TD-SCDMA等等),ITE(Long Term Evolution,长期演进),HiperLAN(high-performance radio local area network,高性能无线局域网)等等。
所述通讯单元15可以用于终端设备1与其他设备,如其他终端设备1或者服务器之间的信息交换。
所述处理器11又称中央处理器(CPU,Central Processing Unit),是一块超大规模的集成电路,是终端设备1的运算核心(Core)和控制核心(Control Unit)。处理器11的功能主要是解释程序指令以及处理软件中的数据。
所述处理器11连接于所述位置获取单元12、电子地图单元13、计时单元14以及通讯单元15,用于控制位置获取单元12获取用户的位置数据以及控制计时单元14记录用户在某个地理位置的访问时间,将所述位置数据以及访问时间数据进行处理,如存储在用户轨迹记录集,及/或显示在所述电子地图单元13上等等。
在本实施例的其中一个应用实例中,当用户来到某个地方时,该用户随身携带的终端设备1,如用户的手机,通过其位置获取单元12,得到用户当前所述的位置,将此时的经度和维度保存起来。进一步地,还可以透过电子地图单元13获取所述位置的名称,如XX公园,并得到该位置的所
属类别,如属于景点,并将该些信息一并保存起来。进一步地,终端设备1的计时单元14还会记录用户在该位置的访问时间(即停留时间)。其中,所述用户记为u,所述位置记为l,以及所述时间记为t。
所述终端设备1会将由用户访问位置数据(包括所述用户u,位置l,以及时间t)组成的用户轨迹记录集传送至一服务端系统中,待该服务端系统进行数据处理,得到推荐位置,并将所述推荐位置推送给所述终端设备1,供用户参考使用。
二、数据处理的服务器2
所述服务器2可以是云服务器。与所述终端设备1相似地,所述服务器2包括处理器21、电子地图单元23以及通讯单元25。此外,所述服务器2还安装有位置推荐系统10。
本实施例中,所述位置推荐系统10可以包括多个由程序段所组成的功能模块(详见图4)。所述位置推荐系统10中的各个程序段的程序代码可以存储于服务器2的存储设备26中,并由服务器2的处理器21所执行,以对所述终端设备1传送过来的用户轨迹记录集进行数据处理,得到一个或者多个推荐位置,并将所述推荐位置推送给所述终端设备1,供用户参考使用(详见图4中描述)。
参阅图4所示,为本发明位置推荐系统较佳实施例的功能模块图。本实施例中,所述位置推荐系统10根据其所执行的功能,可以被划分为多个功能模块。本实施例中,所述功能模块包括:数据获取模块100、数据处理模块101、数据建模求解模块102以及推荐位置分析模块103。
所述数据获取模块100用于从各个终端设备1获取每个终端设备1所收集的用户轨迹记录集。
本实施例中,每一台终端设备1在得到用户允许的情况下,实时获取用户位置数据以及该位置的访问时间数据,将获取的数据记录到一个用户轨迹记录集中,并将获取的数据记录到一个用户轨迹记录集中。
所述终端设备1首先要检测用户是否允许获取地理定位信息。获取地理定位信息可能侵犯用户的隐私,因此除非用户同意,否则位置获取功能是不可用的。
终端设备1可以在其用户界面上弹出一个对话框,表明是否允许获取用户的地理定位信息。若用户选择允许获取地理定位信息,则所述终端设备1的位置获取功能可用,否则,若用户选择不允许获取地理定位信息,则所述终端设备1的位置获取功能不可用。
其中,其中,所述数据记录包括所述用户位置数据,例如位置名称(例如天安门、故宫等)以及经纬度信息等数据,以及该位置的访问时间数据。进一步地,本发明其他实施例中,所述位置数据还可以包括位置所属类别(例如餐饮、旅游、住宿等)。
所述数据记录可以显示在电子地图上,如Google地图、百度地图、高德地图等。如图2所示,某个用户在2016年2月23日,从上午9点钟左右到下午2点钟左右的大概5个小时,连续访问了前门、天安门、故宫博物院、景山公园等位置。
所述数据处理模块101用于从所有用户轨迹记录集中获取一个地理属性范围内的位置数据集合。
每一个终端设备1在得到了用户的允许的情况下,在获取了该用户的位置数据后,都会通过有线或者无线的方式将由所述位置数据组成的用户轨迹记录集传送给所述服务器。因此,所述服务器可以获取多个用户的用户轨迹记录集。
本实施例中,所述服务器对所有用户轨迹记录集中的位置数据记录,以一个相同的地理属性,例如,以城市这个属性,对所有位置数据进行分类,得到某一个城市,例如某个用户A当前所在城市这个地理范围内的所有位置数据集合。
应该了解,在收集位置数据时,所选取的地理范围的大小会影响数据规模和位置推荐的准确度,本实施例中以城市为单位进行位置数据收集。在本发明的其他实施例中,也可以采用其他的地理属性,如以某个位置(如用户A当前所在位置)为中心,以500米、1000米或者2000米等为活动半径的地理范围内。
进一步地,所述数据处理模块101还用于根据所述位置数据集合提取出位置间转移关系数据,从而构成一个由出位置间转移关系数据组成的目标数据集合Ds。
本实施例中,上述获取的位置数据集合中包括用户访问位置数据(u,l,t),其中,u代表用户,l代表位置,t代表访问时间。所述位置间转移关系数据为(u,l,i),其中,i是转移位置,其代表用户从位置l转移到位置i,其中,位置i的选择原则为ti-tl<T,即从位置l到达位置i的时间小于预设时间T。
本实施例中,从位置l到位置i的时间可以是利用电子地图的导航功能
所计算出来的时间。例如,现在的百度地图、高德地图、Google地图等都能够根据确定的起始地及目的地计算从所述起始地到所述目的地的大概时间,包括步行时间、乘坐公共交通工具时间及自驾时间等。
根据不同的系统设置或者用户设置,所述预设时间T可以是步行时间、乘坐公共交通工具时间或者自驾时间等。例如,用户可以通过该用户的终端设备提供的一个设置界面设置预设时间T的个性化需求,如自驾一个小时之内、步行二十分钟之内、公共交通工具直达且时间在半个小时之内、最多转车一次且时间在一个小时之内,等等。
在其中一个示意性的例子中,根据所述用户轨迹记录集,用户A在l1位置停留了t1分钟,在l2位置停留了t2分钟,在l3位置停留了t3分钟,在l4位置停留了t1分钟;用户B在l1位置停留了t4分钟,在l5位置停留了t2分钟,在l6位置停留了t5分钟,在l4位置停留了t6分钟;用户C在l7位置停留了t7分钟,在l2位置停留了t8分钟,在l8位置停留了t9分钟,在l9位置停留了t1分钟。
因此,根据上述例子,所述用户访问位置数据包括(uA,l1,t1)、(uA,l2,t2)、(uA,l3,t3)、(uA,l4,t1)、(uB,l1,t4)、(uB,l5,t2)、(uB,l6,t5)、(uB,l4,t6)、(uC,l7,t7)、(uC,l2,t8)、(uC,l8,t9)、(uC,l9,t1)。所提取出的位置间转移关系数据可以包括(uA,l1,l5)、(uA,l1,l6)、(uA,l2,l5)、(uA,l2,l4),等等。
所述数据建模求解模块102用于采用统计推理方法,从所述目标数据集合Ds中计算出每一个转移关系数据的发生概率。
本实施例中,所述统计推理方法为最大似然估计思想,即最合理的推理是使已有的位置间转移关系数据发生的可能性最大。其中,代表用户u从位置l转移到位置i的概率值的概率函数为:
ful(i)=σ(<vu,viu>+<vl,vil>)。
其中:
符号<,>代表两向量内积;逻辑函数起归一化作用;向量vu代表用户u的偏好向量,如用户u喜欢的位置类别;向量vl代表当前位置l的属性向量,如位置l的位置名称、所属类别、经纬度等;向量viu代表下一个位置i与用户u交互的属性向量,如用户u对位置i的喜欢程度;以及向量vil代表下一个位置i与当前位置l交互的属性向量,如从当前位置l转移到下一位置i的概率。
所有可能的vu、vl、viu、vil可以组成一个参数矩阵Θ={U,Ll,Liu,Lil}。
其中:
U代表由所有用户偏好向量vu组成的矩阵,Ll代表由所有当前位置l的属性向量vl组成的矩阵,Liu代表由所有下一个位置i与用户u交互的属性向量viu组成的矩阵,以及Lil代表由所有下一个位置i与当前位置l交互的属性向量vil组成的矩阵。
本实施例中,对所述目标数据集合Ds做最大似然估计,得到优化函数:
L(Θ)=argmaxΠσ(<vu,viu>+<vl,vil>)。
进一步对上述优化函数进行计算求解,以不断更新其属性向量值vu,vl,viu及vil,使得函数L(Θ)结果不断增大直至收敛,得到的收敛时的向量值vu,vl,viu及vil即为最优结果。
详细地,所述计算求解算法如下:
在上述求解算法中,所有位置间转移关系数据(u,l,i)组成集合Ds作为算法输入,其对应参数集合为Θ={U,Ll,Liu,Lil},当优化函数收敛时,参数集合亦是算法输出。过程则是不断从集合Ds中随机选取一个(u,l,i)转移关系数据,并依据梯度上升思想对相应参数做修改,既直至函数收敛。
利用每一个位置间转移关系数据(u,l,i)对应的参数θ,以及上述的概率函数ful(i)=σ(<vu,viu>+<vl,vil>),可以计算出每一个位置间转移关系数据(u,l,i)的发生概率。
所述推荐位置分析模块103用于基于上述计算出来的转移关系数据(u,l,i)的发生概率以及用户的当前位置,根据所述转移关系数据中的转移位置i,选择前K个推荐位置,并按其发生概率从大到小进行排序后,发送给所述服务器2中的实时推荐引擎,由所述实时推荐引擎向终端设备1进行位置推荐。
需要说明的是,上述描述的实施例采用的是服务器-客户端的模式对用户进行位置推荐。即所述位置推荐系统安装于服务器中,由服务器执行数据处理与位置推荐的动作。
该种模式的应用场景可以是利用服务器通过网络进行主动的位置推荐。只要用户有联网,并开启了某个应用时,如开启了浏览器、团购网站等时,所述服务器就会执行位置推荐操作。
该种模式的另一个应用场景可以是利用目前流行的微信公众号的形式,当用户关注了某个特定公众号,并在该公众号平台执行了刷新或者其他任何预先设定的动作时,所述服务器就会执行位置推荐操作。在其他实施例中,用户也可以在所述公众号平台输入一个特定关键字,如“餐饮”,则所述服务器也可以从所分析出来的位置中筛选出具有该关键字特性的前K个推荐位置进行推荐。
应该了解,本发明所述应用场景不限于上述所列举的情况。本领域技术人员在了解本发明方案的基础上,可以将本发明所述方案应用到任何适合的场景中。
在本发明的其他实施例中,所述位置推荐系统10也可以安装于任何的终端设备中,由终端设备对用户进行位置推荐。本实施例中,服务器可以
定时或者实施将收集到的其他用户的用户轨迹记录集发送给终端设备。当终端设备上的位置推荐系统10被开启后即可实现位置推荐操作。此时,即使终端设备不能连上网络,也可以根据其之前从服务器中得到的用户轨迹记录执行位置推荐操作。
在本发明所提供的几个实施例中,应该理解到,所揭露的系统,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述模块的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式。
另外,在本发明各个实施例中的各功能模块可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能模块的形式实现。
上述以软件功能模块的形式实现的集成的单元,可以存储在一个计算机可读取存储介质中。上述软件功能模块存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)或处理器(processor)执行本发明各个实施例所述方法的部分步骤。
对于本领域技术人员而言,显然本发明不限于上述示范性实施例的细节,而且在不背离本发明的精神或基本特征的情况下,能够以其他的具体形式实现本发明。因此,无论从哪一点来看,均应将实施例看作是示范性的,而且是非限制性的,本发明的范围由所附权利要求而不是上述说明限定,因此旨在将落在权利要求的等同要件的含义和范围内的所有变化涵括
在本发明内。不应将权利要求中的任何附图标记视为限制所涉及的权利要求。此外,显然“包括”一词不排除其他单元或步骤,单数不排除复数。系统权利要求中陈述的多个单元或装置也可以由一个单元或装置通过软件或者硬件来实现。第一,第二等词语用来表示名称,而并不表示任何特定的顺序。
最后应说明的是,以上实施例仅用以说明本发明的技术方案而非限制,尽管参照较佳实施例对本发明进行了详细说明,本领域的普通技术人员应当理解,可以对本发明的技术方案进行修改或等同替换,而不脱离本发明技术方案的精神和范围。
Claims (10)
- 一种位置推荐方法,其特征在于,该方法包括:从一个或者多个用户轨迹记录中,获取一个地理属性范围内的位置数据集合;根据所述位置数据集合提取出位置间转移关系数据,构成一个由位置间转移关系数据组成的目标数据集合;采用统计推理方法,从所述目标数据集合中计算出每一个位置间转移关系数据的发生概率;及基于所述转移关系数据的发生概率以及用户的当前位置,选择前K个推荐位置,并按推荐位置的排序发送给实时推荐引擎。
- 如权利要求1所述的位置推荐方法,其特征在于,该方法进一步包括:所述实时推荐引擎向终端设备发送位置推荐。
- 如权利要求1所述的位置推荐方法,其特征在于,所述用户轨迹记录包括实时获取的用户位置数据以及该位置数据对应的访问时间数据。
- 如权利要求4所述的位置推荐方法,其特征在于,所述统计推理方法包括:将所有可能的vu、vl、viu、vil组成一个参数矩阵Θ={U,Ll,Liu,Lil},其中,U代表由所有用户偏好向量vu组成的矩阵,Ll代表由所有当前位置l的属性向量vl组成的矩阵,Liu代表由所有下一个位置i与用户u交互的属性向量viu组成的矩阵,以及Lil代表由所有下一个位置i与当前位置l交互的属性向量vil组成的矩阵;对所述目标数据集合Ds做最大似然估计,得到优化函数:L(Θ)=arg maxΠσ(<vu,viu>+<vl,vil>);及对上述优化函数进行计算求解,以不断更新其属性向量值vu,vl,viu及vil,使得函数L(Θ)结果不断增大直至收敛,得到的收敛时的向量值vu,vl,viu及vil。
- 一种位置推荐系统,其特征在于,该系统包括:数据处理模块,用于从一个或者多个用户轨迹记录中,获取一个地理属性范围内的位置数据集合,并根据所述位置数据集合提取出位置间转移关系数据,构成一个由位置间转移关系数据组成的目标数据集合;数据建模求解模块,用于采用统计推理方法,从所述目标数据集合中计算出每一个位置间转移关系数据的发生概率;及推荐位置分析模块,用于基于所述转移关系数据的发生概率以及用户的当前位置,选择前K个推荐位置,并按推荐位置的排序发送给实时推荐引擎。
- 如权利要求6所述的位置推荐系统,其特征在于,该系统进一步包括:数据获取模块,用于从一个或者多个终端设备获取所述用户轨迹记录。
- 如权利要求7所述的位置推荐系统,其特征在于,所述用户轨迹记录包括实时获取的用户位置数据以及该位置数据对应的访问时间数据。
- 如权利要求9所述的位置推荐系统,其特征在于,所述统计推理方法包括:将所有可能的vu、vl、viu、vil组成一个参数矩阵Θ={U,Ll,Liu,Lil},其中,U代表由所有用户偏好向量vu组成的矩阵,Ll代表由所有当前位置l的属 性向量vl组成的矩阵,Liu代表由所有下一个位置i与用户u交互的属性向量viu组成的矩阵,以及Lil代表由所有下一个位置i与当前位置l交互的属性向量vil组成的矩阵;对所述目标数据集合Ds做最大似然估计,得到优化函数:L(Θ)=arg maxΠσ(<vu,viu>+<vl,vil>);及对上述优化函数进行计算求解,以不断更新其属性向量值vu,vl,viu及vil,使得函数L(Θ)结果不断增大直至收敛,得到的收敛时的向量值vu,vl,viu及vil。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610266263.6A CN105912702A (zh) | 2016-04-26 | 2016-04-26 | 位置推荐方法及系统 |
| CN201610266263.6 | 2016-04-26 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2017185462A1 true WO2017185462A1 (zh) | 2017-11-02 |
Family
ID=56752783
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2016/084088 Ceased WO2017185462A1 (zh) | 2016-04-26 | 2016-05-31 | 位置推荐方法及系统 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105912702A (zh) |
| WO (1) | WO2017185462A1 (zh) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738525A (zh) * | 2019-10-16 | 2020-01-31 | 李栋 | 一种基于商家定位的广告推广方法及系统 |
| CN113254570A (zh) * | 2021-07-15 | 2021-08-13 | 浙江网商银行股份有限公司 | 数据识别方法以及装置 |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110659920B (zh) * | 2018-06-28 | 2023-04-18 | 阿里巴巴集团控股有限公司 | 一种业务对象的推荐方法和装置 |
| CN109525664B (zh) * | 2018-11-15 | 2021-06-01 | 中国联合网络通信集团有限公司 | 位置获取方法、装置和存储介质 |
| CN109522491B (zh) * | 2018-11-29 | 2020-07-31 | 杭州飞弛网络科技有限公司 | 一种基于位置属性的陌生人社交活动推荐方法与系统 |
| CN113065064B (zh) * | 2021-03-24 | 2023-09-29 | 支付宝(杭州)信息技术有限公司 | 信息推荐处理方法、装置、设备及存储介质 |
| CN116595259A (zh) * | 2021-03-25 | 2023-08-15 | 支付宝(杭州)信息技术有限公司 | 位置推荐处理方法及装置 |
| CN114328794B (zh) * | 2022-01-06 | 2025-02-28 | 迪爱斯信息技术股份有限公司 | 一种警情位置推理方法、系统、计算机设备及存储介质 |
| CN119312895A (zh) * | 2024-10-18 | 2025-01-14 | 南京航空航天大学 | 一种基于组合赋权和用户反馈的地点实时推荐方法及系统 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559252A (zh) * | 2013-11-01 | 2014-02-05 | 桂林电子科技大学 | 给游客推荐其很可能会浏览的景点的方法 |
| CN103995837A (zh) * | 2014-04-25 | 2014-08-20 | 西北工业大学 | 一种基于群体足迹的个性化旅游路线规划方法 |
| CN104765873A (zh) * | 2015-04-24 | 2015-07-08 | 百度在线网络技术(北京)有限公司 | 用户相似度确定方法和装置 |
| CN105183800A (zh) * | 2015-08-25 | 2015-12-23 | 百度在线网络技术(北京)有限公司 | 信息预测的方法和装置 |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1661604A (zh) * | 2004-02-25 | 2005-08-31 | 松下电器产业株式会社 | 移动终端的活动记录分析和自动信息推荐系统及方法 |
| CN103731916B (zh) * | 2014-01-14 | 2017-07-07 | 上海河广信息科技有限公司 | 一种基于无线网络的用户位置预测系统和方法 |
| CN103747523B (zh) * | 2014-01-14 | 2017-09-22 | 上海河广信息科技有限公司 | 一种基于无线网络的用户位置预测系统和方法 |
| CN104833365B (zh) * | 2014-02-12 | 2017-12-08 | 华为技术有限公司 | 一种用户目的地点的预测方法及装置 |
-
2016
- 2016-04-26 CN CN201610266263.6A patent/CN105912702A/zh active Pending
- 2016-05-31 WO PCT/CN2016/084088 patent/WO2017185462A1/zh not_active Ceased
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559252A (zh) * | 2013-11-01 | 2014-02-05 | 桂林电子科技大学 | 给游客推荐其很可能会浏览的景点的方法 |
| CN103995837A (zh) * | 2014-04-25 | 2014-08-20 | 西北工业大学 | 一种基于群体足迹的个性化旅游路线规划方法 |
| CN104765873A (zh) * | 2015-04-24 | 2015-07-08 | 百度在线网络技术(北京)有限公司 | 用户相似度确定方法和装置 |
| CN105183800A (zh) * | 2015-08-25 | 2015-12-23 | 百度在线网络技术(北京)有限公司 | 信息预测的方法和装置 |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738525A (zh) * | 2019-10-16 | 2020-01-31 | 李栋 | 一种基于商家定位的广告推广方法及系统 |
| CN110738525B (zh) * | 2019-10-16 | 2023-09-29 | 李栋 | 一种基于商家定位的广告推广方法及系统 |
| CN113254570A (zh) * | 2021-07-15 | 2021-08-13 | 浙江网商银行股份有限公司 | 数据识别方法以及装置 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105912702A (zh) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2017185462A1 (zh) | 位置推荐方法及系统 | |
| US12579135B2 (en) | System and method for data collection to validate location data | |
| CN109376313B (zh) | 信息推荐方法、装置、终端及存储介质 | |
| JP6784308B2 (ja) | 施設特性を更新するプログラム、施設をプロファイリングするプログラム、コンピュータ・システム、及び施設特性を更新する方法 | |
| US8457653B2 (en) | Method and apparatus for pre-fetching location-based data while maintaining user privacy | |
| KR101502043B1 (ko) | 지오-소셜 인자에 기반한 컨텐츠 서피싱 | |
| CN106570722B (zh) | 一种智能推荐系统及智能推荐方法 | |
| US9715233B1 (en) | System and method for inputting a second taxi-start location parameter for an autonomous vehicle to navigate to whilst reducing distraction | |
| US20110087685A1 (en) | Location-based service middleware | |
| CN106209935B (zh) | 一种信息处理方法、终端及服务器 | |
| US9652548B2 (en) | Venue prediction based on ranking | |
| US11449530B2 (en) | Determining attribute information of geographical locations | |
| WO2019005333A1 (en) | GEOGRAPHIC RESEARCH OFFLINE | |
| CN108419205A (zh) | 提供热点给服务器控制区域内的用户装置 | |
| US12277180B2 (en) | Information processing apparatus and method | |
| CN107710246A (zh) | 用以促进描述位置的用户图像的提交的系统和方法 | |
| US20180253777A1 (en) | Method for analyzing and ranking venues | |
| US20180302739A1 (en) | Signal sharing between trusted groups of devices | |
| CN111859143A (zh) | 信息推送方法、装置、电子设备以及存储介质 | |
| EP3682396A1 (en) | Status streaming pop-up map | |
| US20110310793A1 (en) | On-demand information retrieval using wireless communication devices | |
| CN103139906A (zh) | 对移动终端的定位进行校正的方法和设备 | |
| CN102360373A (zh) | 一种移动搜索方法 | |
| Jakkhupan et al. | A context-aware personalized venue recommender system on smart device | |
| CN120091389A (zh) | 一种数据处理方法及相关设备 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16899972 Country of ref document: EP Kind code of ref document: A1 |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 16899972 Country of ref document: EP Kind code of ref document: A1 |



