WO2017031856A1 - 信息预测的方法和装置 - Google Patents

信息预测的方法和装置 Download PDF

Info

Publication number
WO2017031856A1
WO2017031856A1 PCT/CN2015/096131 CN2015096131W WO2017031856A1 WO 2017031856 A1 WO2017031856 A1 WO 2017031856A1 CN 2015096131 W CN2015096131 W CN 2015096131W WO 2017031856 A1 WO2017031856 A1 WO 2017031856A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
user
prediction model
prediction
point
Prior art date
Application number
PCT/CN2015/096131
Other languages
English (en)
French (fr)
Inventor
吴海山
武政伟
张潼
Original Assignee
百度在线网络技术（北京）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 百度在线网络技术（北京）有限公司 filed Critical 百度在线网络技术（北京）有限公司
Publication of WO2017031856A1 publication Critical patent/WO2017031856A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Definitions

  • the present application relates to the field of computer technologies, and in particular, to the field of communications technologies, and in particular, to a method and apparatus for information prediction.
  • the smart terminal is the abbreviation of mobile intelligent terminal. It has the ability to access the Internet. It is usually equipped with various operating systems and can customize various functions according to user needs.
  • user data is collected by various sensors embedded in intelligent terminals, user data is transmitted through communication technologies, and user behavior is analyzed, which becomes a commonly used technical means.
  • some smart terminals can collect user's location data through Global Positioning System (GPS) sensors, locate users, and then push information to users according to regional characteristics.
  • GPS Global Positioning System
  • Some smart terminals can measure user's blood pressure data through blood pressure sensors. And transmitted to the background server or other predetermined terminal, thereby monitoring the user's blood pressure data, and so on.
  • the prior art in the analysis of the user's location data, the prior art often predicts the future location of the user through the historical motion track or historical stop location data of the terminal.
  • the existing technique for predicting the future position of the user only the historical stop position data of the user is considered, and in the prediction of the position that the user has not reached, it is difficult to accurately give the prediction result. Therefore, this information prediction method has a problem that the terminal data is underutilized, resulting in low effectiveness of information prediction.
  • the purpose of the present application is to propose an improved method and apparatus for information prediction to solve the technical problems mentioned in the background section above.
  • the present application provides a method for information prediction, the method comprising: acquiring personalized information of a user, wherein the personalized information includes at least a current motion trajectory; and determining a corresponding prediction according to the personalized information. a model; predicting travel information of the user based on the personalized information and the predictive model.
  • the personalized information further includes: a historical motion trajectory.
  • determining the corresponding prediction model according to the personalized information comprises: detecting whether the personalized information further includes a search behavior feature, wherein the search behavior feature includes at least the searched geographic information point a location feature; if included, determining that the prediction model is a first prediction model, wherein the first prediction model is obtained based on a sample set of historical search behavior characteristics and historical travel information; if not, determining the prediction model Is a second prediction model, wherein the second prediction model is obtained based on historical motion trajectory training.
  • the search behavior feature further includes at least one of: a temporal feature of searching for a geographic information point, a location at which the geographic information point is searched, and a search for the searched geographic information point. The number of times, the time interval for searching for the same geographic information point, whether to perform path planning based on the searched geographic information points, and the category of the geographic information points that are searched.
  • the predicting the travel information of the user based on the personalized information and the prediction model comprises: obtaining a first prediction of the travel information of the user based on the search behavior feature and the first prediction model a result; based on the current motion trajectory, adjusting the first prediction result to determine a time when the user arrives at the searched geographic information point and a probability of arrival.
  • the predicting the travel information of the user based on the personalized information and the prediction model comprises: acquiring a stay point transition probability matrix based on the historical motion trajectory; and according to the current motion trajectory and the staying The point transition probability matrix determines the probability and time that the user appears at any stay point.
  • the method further comprises: obtaining actual travel information of the user; updating the predictive model based on the actual travel information.
  • the present application provides an apparatus for information prediction, the apparatus comprising: an information acquisition module configured to acquire personalized information of a user, wherein the personalized information includes at least a current motion track; and a model determination module And configured to determine a corresponding prediction model according to the personalized information; and the information prediction module is configured to predict the travel information of the user based on the personalized information and the prediction model.
  • the personalized information further includes: a historical motion trajectory.
  • the model determining module includes: a detecting unit configured to detect whether the personalized information further includes a search behavior feature, wherein the search behavior feature includes at least a location feature of the searched geographic information point a determining unit configured to determine that the predictive model is a first predictive model if the personalized information includes a search behavior feature, wherein the first predictive model is based on a sample set training of historical search behavior characteristics and historical travel information Obtaining; and, if the personalized information does not include a search behavior feature, determining that the prediction model is a second prediction model, wherein the second prediction model is obtained based on historical motion trajectory training.
  • the search behavior feature further includes at least one of: a temporal feature of searching for a geographic information point, a location at which the geographic information point is searched, and a search for the searched geographic information point. The number of times, the time interval for searching for the same geographic information point, whether to perform path planning based on the searched geographic information points, and the category of the geographic information points that are searched.
  • the information prediction module includes: a first prediction unit configured to acquire a first prediction result for user travel information based on the search behavior feature and the first prediction model; an adjustment unit, configured And for adjusting the first prediction result based on the current motion trajectory to determine a time when the user arrives at the searched geographic information point and a probability of arrival.
  • the information prediction module includes: a transition probability matrix acquisition module configured to acquire a stay point transition probability matrix based on the historical motion trajectory; and a second prediction unit configured to use the current motion trajectory and The stay point transition probability matrix determines the probability and time that the user appears at any stay point.
  • the apparatus further includes: an actual travel information acquisition module configured to acquire actual travel information of the user; and a model update module configured to be based on the The actual travel information is updated to update the predictive model.
  • the method and apparatus for information prediction provided by the present application, by acquiring personalized information of a user, wherein the personalized information includes at least a current motion trajectory, and then determining a corresponding prediction model according to the personalized information, and then based on the personalized information. And the prediction model predicts the user's travel information, and the personalized information such as the current motion trajectory is introduced, thereby improving the effectiveness of the information prediction.
  • FIG. 1 illustrates an exemplary system architecture to which embodiments of the present application may be applied
  • FIG. 2 is a flow chart of one embodiment of a method of information prediction in accordance with the present application.
  • FIG. 3 is a schematic diagram of motion trajectories and stay points of a method of information prediction according to the present application
  • FIG. 4 is a schematic diagram of an application scenario of a method for information prediction according to the present application.
  • FIG. 5 is a flow chart of still another embodiment of a method of information prediction according to the present application.
  • FIG. 6 is a schematic structural diagram of an embodiment of an apparatus for predicting information according to the present application.
  • FIG. 7 is a schematic structural diagram of still another embodiment of an apparatus for predicting information according to the present application.
  • FIG. 1 illustrates an exemplary system architecture 100 in which embodiments of the present application may be applied.
  • the system architecture 100 can include terminal devices 101, 102, a network 103, and a server 104.
  • the network 103 is used to provide a medium for communication links between the terminal devices 101, 102 and the server 104.
  • Network 103 may include various types of connections, such as wired, wireless communication links, fiber optic cables, and the like.
  • the first user can interact with the server 104 over the network 103 using the terminal devices 101, 102 to receive or send messages and the like.
  • the terminal device 101, 102 can be installed with monitoring or management tools for location information, clients for information prediction applications, map applications, search applications, and the like.
  • the server 104 may acquire the current motion trajectory of the first user according to the monitoring or management tool of the location information installed by the terminal device 101, 102, and predict the travel information of the first user, or may receive the information prediction installed by the terminal device 101, 102.
  • the client predicts the travel information of the first user.
  • the server 104 may also perform statistics and analysis on the travel information of the at least one first user predicted by the terminal device 101, 102 or the server 104.
  • the server 104 may push information to the first user based on statistics and analysis of travel information of the at least one first user.
  • system architecture 100 may also include terminal device 105.
  • the terminal device 105 interacts with the server 104 via the network 103 to receive or transmit a message or the like.
  • the server 104 may send the travel information of the at least one first user predicted by the terminal device 101, 102 or the server 104 to the terminal device 105, so that the second user can understand the statistical characteristics of the travel information of the at least one first user through the terminal device 105.
  • the terminal device 105 can also push information to the first user through the server 104.
  • the terminal devices 101, 102 and the terminal device 105 can be various electronic devices including, but not limited to, personal computers, smart phones, smart watches, tablets, personal digital assistants, and the like.
  • Server 104 may be a server that provides various services.
  • the server can store, analyze, and the like the received data, and feed back the processing result to the terminal device.
  • the method for predicting information provided by the embodiment of the present application may be performed by the terminal device 101, 102 or may be performed by the server 104.
  • the device for predicting information may be set in the terminal device 101, 102, or may be set in In the server 104.
  • terminal devices, networks, and servers in Figure 1 is merely illustrative. Depending on the implementation needs, there can be any number of terminal devices, networks, and servers.
  • FIG. 2 illustrates a flow 200 of one embodiment of a method of information prediction.
  • the method for predicting information includes the following steps:
  • Step 201 Acquire personalized information of the user.
  • an electronic device (such as the terminal device 101, 102 or the server 104 shown in FIG. 1) can acquire personalized information of the user locally or remotely.
  • the electronic device is a terminal device (such as the terminal device 101, 102) that is installed with the information prediction application or at least the monitoring or management tool with the location information installed, the above-mentioned electronic device can directly obtain the above personalized information locally.
  • Information when the above electronic device is a monitoring or management tool for location information, or a back-end server (such as server 104) that provides support for the client of the information prediction application, it can be obtained from the terminal device through a wired connection or a wireless connection.
  • User behavior information include, but are not limited to, 3G/4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods now known or developed in the future.
  • the personalized information may be information having individual characteristics, such as the user's personal behavior habit information (work schedule, motion trajectory, search behavior characteristics) and the like.
  • the personalized information of the user mainly includes personal behavior habit information of the user that can be obtained by monitoring the terminal device or by monitoring the operation of the user on the terminal device.
  • the personalized information described above may at least include the current motion trajectory.
  • the motion trajectory may represent a spatial feature composed of a route through which the object passes from the starting position to the ending position.
  • the motion trajectory can be represented by a motion trajectory direction (such as a moving direction) and a motion trajectory (such as a linear trajectory).
  • the location of the terminal device held by the user at different time points is used as a track point, and the motion track of the user can be represented by a connection of the track points.
  • reference numeral 301 indicates a track point, and the line between the track points constitutes a motion track 302, and the lined arrows in Fig. 3 indicate the direction of the motion track.
  • each terminal device can have an identity code, such as a mobile device's International Mobile Equipment Identity (IMEI).
  • IMEI International Mobile Equipment Identity
  • the electronic device can use the identity code as the identity identification number of the terminal device to distinguish different terminal devices.
  • the geographic location information of the user can be obtained by using a positioning unit built in the terminal device.
  • the positioning unit can be implemented by hardware, for example, it can be a terminal device
  • the integrated locator (such as the global positioning system GPS terminal) can also be implemented by software, for example, it can be a positioning application (such as a 51 positioning terminal that can run on the Android operating system); it can also be realized by software and hardware.
  • the positioning application can automatically find the corresponding base station according to the IP address (Internet Protocol Address) of the terminal access network (such as a WIFI network, a mobile data cellular network, etc.) Obtain the current location of the terminal, and also obtain the current location of the terminal through GPS positioning data.
  • the personalized information of the user may also include historical motion trajectories, search behavior features, and the like.
  • the user's motion trajectory can include a motion route and a stay point.
  • a moving route is a connection of multiple locations.
  • the stop point is determined based on the concentration of the track points in the motion route. For example, a plurality of temporally adjacent track points whose speeds are below a speed threshold are merged into one stop position, if the distance between the centers of any two of the stop positions is less than a distance threshold, and a plurality of stop positions When the time interval between the start time of the dwell time of the first stop position and the end time of the last stop position is greater than the time threshold, the area composed of the plurality of stop positions is determined as the stay point.
  • the position of the stay point can be represented by the center position of the area composed of the plurality of stay positions. As shown in FIG. 3, the area composed of the respective track points in the circular areas 303, 304 is a stay point.
  • the velocity of the track point can be calculated by dividing the distance between the two track points by the time taken to pass the two track points.
  • Step 202 Determine a corresponding prediction model according to the personalized information.
  • the electronic device may then analyze the personalized information of the user acquired in step 201, and determine a prediction model for predicting the travel information of the user according to the personalized information of the user.
  • the electronic device can determine a prediction model for different information predictions. For example, a new terminal device, the user does not perform the search behavior through the terminal device, and cannot obtain the historical motion track of the user, and the personalized information obtained by the electronic device may only have the current motion track.
  • the electronic device may determine that the prediction model is a prediction model established based on a route between geographic information points on the map.
  • the electronic device can also determine that the predictive model is a model based on statistics of travel routes for a large number of users.
  • the personalized information acquired by the electronic device may include a historical motion track and a current motion track.
  • the electronic device may have a stay point transition probability model that is previously trained according to the user's historical motion trajectory. If the personalized information acquired by the electronic device includes the current motion trajectory, it may be detected whether there is a historical motion trajectory of the user using the terminal device, and if so, the prediction model is determined as the stay point transition probability model.
  • each stay point is regarded as a state of an infinite hidden Markov model, and an observation vector generated by a state sequence with a corresponding probability density distribution is used.
  • the observation vector sequence is used as a prediction model.
  • the process of the electronic device training the stop point transition probability model according to the historical movement track of the user may be: firstly, the historical stop point of the user is obtained according to the historical motion track of the user; and then the user is obtained at each historical stop point according to the motion route in the historical motion track.
  • ⁇ is used to indicate the probability that the user arrives at each historical stop point
  • ⁇ is used to indicate the probability of the user accessing a place other than the historical stop point—that is, a new stop point is generated.
  • Probability the expected value of the number of stay points is represented by ⁇
  • the probability of shifting from the stay point i to the stay point j is represented by n ij
  • the probability that the stay point j is a known stay point on the map, when the current position of the user is the stay point i, the probability that the user shifts to each stay point is:
  • the probability of staying at the current stop point is:
  • the probability of moving from the current stop point to another historical stop point is:
  • the probability of moving from the current stop point to the new stop point is:
  • the probability that the new stop point is a known place on the map is:
  • the probability that the new stop point is an unknown stop point on the map is:
  • Step 203 predict travel information of the user based on the personalized information and the prediction model.
  • the electronic device may then perform matching and/or calculation on the acquired personalized information through the determined prediction model, and predict the user's possible travel information, such as travel time, travel location, and/or travel location. Probability and so on.
  • the prediction model is iHMM as an example: in step 202, the principle of predicting the probability that the user of the iHMM reaches each stay point has been described in detail by the formula, and when the prediction is made, the electronic The device may establish a stay point transfer sequence as a sample to be predicted according to the current motion trajectory of the user, and use the above formula representing each of the formulas for predicting the probability of the user reaching each stay point to evaluate the sample to be tested, and obtain the user to stay in the original stay. The probability of a point, a transfer to another historical stop point, or a new stop point.
  • the other stop points to which the predicted user is transferred are the travel locations, and the probability of moving to other stop points is the travel probability.
  • the travel time can be calculated according to the speed in the user's current running track and the distance between the two stay points, or can be calculated according to the time required for the transfer between the stay points (such as the transfer between two historical stop points). Time is averaged).
  • the electronic device may continue to monitor the user's motion track to obtain the user's actual travel information, including the travel route. , staying point, etc., and adding the current motion trajectory to the historical motion trajectory, adding the new stay point to the historical stay point, and updating the corresponding prediction model, increasing the sample capacity of the prediction model with new travel data, so as to predict the prediction model The result is more and more accurate.
  • the electronic device acquires the current motion trajectory of the user through the terminal device used by the user; then, the electronic device detects whether the historical motion trajectory of the user is acquired, and if so, determines the prediction model as the user a stop point transition probability model for historical trajectory training, as indicated by reference numeral 402; then, as indicated by reference numeral 403, the electronic device will The sequence of stay points included in the current motion trajectory is used as a sample to be predicted, and is evaluated by a stay point transition probability model to predict travel information such as a travel location, a travel probability, and a travel time of the user; further, the electronic device can predict the travel probability The largest travel location is used as the target location, and information about the target location, such as the road condition information of the current location to the target location, the parking location information of the target location, etc., is pushed to the user.
  • the information pushing method of the embodiment can improve the effectiveness of predicting the travel information of the user by introducing personalized information such as the current motion track.
  • the flow of the map display method 500 includes the following steps:
  • Step 501 Acquire personalized information of the user.
  • an electronic device (such as the terminal device 101, 102 or the server 104 shown in FIG. 1) can acquire personalized information of the user locally or remotely.
  • the personalized information of the user may include, but is not limited to, a user's work schedule, a motion track, a search behavior feature, and the like.
  • the personalized information includes at least the current motion trajectory.
  • Step 502 Detect whether the personalized information further includes a search behavior feature.
  • the electronic device may then further detect the personalized information to determine which information is included in the personalization information in addition to the current motion trajectory, such as whether or not the search behavior feature is further included.
  • the search behavior feature is a behavioral characteristic when the user searches through the terminal device, for example, a search term used in the search.
  • the search term is a word related to a geographic information point (such as Xiangshan)
  • the search behavior feature includes at least a location feature of the searched geographic information point, and the search feature may further include, but is not limited to, at least one of the following : the time characteristics of the search for geographic information points (such as weekdays or rest days), the location where the geographic information points were searched (such as geographic location coordinates, etc.), the number of searches for the geographic information points being searched, The time interval for searching for the same geographic information point, whether to perform path planning based on the searched geographic information points (such as searching for a route), the type of geographical information points searched (such as enterprises, attractions, restaurants, etc.), and the geographic information points Search for weather features (such as sunny or raining, etc.), and more.
  • the time characteristics of the search for geographic information points such as weekdays or rest days
  • Step 503 if included, determining that the prediction model is the first prediction model, and if not, determining that the prediction model is the second prediction model.
  • the electronic device may determine a corresponding prediction model according to the detection result of the personalized information: if it is detected that the personalized information includes the search behavior feature, the electronic device may determine that the prediction model is based on the historical search behavior feature and The sample set of historical travel information trains the obtained first prediction model; if it is detected that the search behavior feature is not included in the personalized information, the electronic device may determine that the prediction model is a second prediction model obtained based on the historical motion trajectory training.
  • the first prediction model may include a model for searching weights of items in the behavior feature, the weights being calculated according to the historical search behavior characteristics and historical travel information of the user.
  • the search behavior feature includes a time feature of searching for the geographic information point, and the electronic device separately obtains the probability of the user resting the sunrise line and the probability of the work sunrise line according to the historical travel information.
  • the probability of a rest day or a work sunrise line may be, for example, the ratio of the number of days of rest days or work sunrises to the total number of days of rest or workdays.
  • the probability that the user works the sunrise line can be used as the weight of the search for the geographic information point on the working day.
  • each of the features of the search behavior is given a weight.
  • the personalized information acquired by the electronic device includes the search behavior feature
  • the prediction weight of each item in the search behavior feature may be obtained by the first prediction model.
  • the electronic device may use the search behavior feature in the preset time period (for example, 3 months) as the search behavior feature acquired by the current prediction, and calculate the weight for the search time according to the time interval between the search time and the current time. For example, the larger the time interval between the search time and the current time, the smaller the corresponding weight.
  • the first prediction model may also be a model of a decision tree structure trained by machine learning according to a user's historical search behavior characteristics and historical travel information, such as a gradient boost decision tree.
  • the model of the decision tree structure can be obtained by training the search behavior features in the sample and the travel information within a certain period of time after the search (for example, 3 months) to establish different subtrees, each leaf in the subtree.
  • the node corresponds to a prediction score, which may be a probability that the search behavior feature reaches the node, thereby obtaining an initial prediction model; adding weights to the class of the faulty in the previously acquired prediction model, dividing again, and performing the step cyclically until The accuracy of the comparison between the travel information predicted by the model and the actual travel information in the sample reaches a preset threshold (such as 98%).
  • a prediction score which may be a probability that the search behavior feature reaches the node, thereby obtaining an initial prediction model; adding weights to the class of the faulty in the previously acquired prediction model, dividing again, and performing the step cyclically until The accuracy of the comparison between the travel information predicted by the model and the actual travel information in the sample reaches a preset threshold (such as 98%).
  • the electronic device may constrain the distance, that is, if the distance between the searched geographic information point and the actually arrived geographic location is less than the constraint distance, the user is considered to have arrived.
  • the constraint distance may be a product of a distance between the current location and the searched location and a constraint coefficient, or may be an actual distance between the searched geographic location and the actually arrived geographic location.
  • the constraint distance may be 200 meters.
  • the electronic device When the geographical location is within 200 meters, the electronic device considers that the user actually reached the searched geographical location.
  • the constraint distance may be a product of a distance between the geographic location searched by the user and the current location and a constraint coefficient (eg, 0.05).
  • the second preset distance eg, 500 kilometers
  • the electronic device When the distance reached by the user from the searched geographic location is less than the product of the distance between the geographic location searched by the user and the current location and the constraint coefficient, the electronic device considers that the user actually reached the searched geographic location. position.
  • the electronic device thinks that the user actually reached the searched Geographic location.
  • the second prediction model may be a prediction model established according to a route between geographic information points on the map, or may be a stop point transition probability model (such as IHMM), and will not be described herein.
  • Step 504 predict travel information of the user based on the personalized information and the prediction model.
  • the electronic device may then perform matching and/or calculation on the acquired personalized information through the determined prediction model, and predict possible travel information of the user, such as travel time, travel location, and/or travel probability.
  • the personalized prediction information includes a search behavior feature
  • the first prediction model is an example including a model for calculating a weight of each item in the search behavior feature according to the historical search behavior feature and the historical travel information of the user
  • the electronic device may perform the prediction.
  • the search behavior characteristics in a certain period of time such as 3 months
  • the weights of the items in the sum are added to obtain the prediction coefficient.
  • the electronic device may divide the prediction coefficient obtained by the search behavior feature for each search by a prediction coefficient of the search behavior feature within a certain time period (eg, 3 months) as the user arrives at the search behavior.
  • the probability of travel of geographic information points contained in the feature can be determined according to the user's travel habits (such as the probability of a rest day).
  • the electronic device may use the prediction result of the user travel information acquired based on the search behavior feature and the first prediction model as the first prediction result, and then determine the direction of the current motion track.
  • the first prediction result is adjusted according to the relationship between the direction of the current motion trajectory, the position of the previous stay point, and the position of the searched geographic information point. For example, when the direction of the current motion trajectory of the user is the same as the direction of the location of the previous stay point to the location of the searched geographic information point, the probability of the user reaching the searched geographic information point in the first prediction result increases, reaching The time can be calculated from the speed of motion in the current motion trajectory and the distance between the current location and the searched geographic information point.
  • the steps 501 and 504 in the foregoing implementation process are substantially the same as the steps 201 and 203 in the foregoing embodiment, and details are not described herein again.
  • the flow 500 of the method for information prediction in this embodiment further includes a step 502 of detecting whether the personalized information further includes a search behavior feature, and the step 503 is different from step 202 in that step 503 divides the prediction model into a first prediction model including search behavior features and a second prediction model not including search behavior features.
  • the solution described in this embodiment can introduce more personalized information data of the user, thereby realizing more effective prediction of the user travel information.
  • the electronic device may summarize the predicted results of the travel information of a large number of users, and then push them to the management personnel or merchants of the geographic information points.
  • the present application provides an embodiment of an apparatus for predicting information, the apparatus embodiment corresponding to the method embodiment shown in FIG. 2, the apparatus specific Can be applied to electronic devices.
  • the apparatus 600 for information prediction includes an information acquisition module 601, a model determination module 602, and an information prediction module 603.
  • the information obtaining module 601 is configured to acquire personalized information of the user, where the personalized information includes at least a current motion track;
  • the model determining module 602 is configured to determine a corresponding prediction model according to the personalized information, and the information prediction module 603
  • the configuration is used to predict the travel information of the user based on the personalized information and the prediction model described above.
  • the model determination module 602 includes a detection unit 6021 and a determination unit 6022.
  • the detecting unit 6021 is configured to detect whether the personalized information further includes a search behavior feature, wherein the search behavior feature includes at least a location feature of the searched geographic information point.
  • the determining unit 6022 is configured to: if the personalized information includes the search behavior feature, determine the prediction model as the first prediction model, wherein the first prediction model is obtained based on the historical search behavior feature and the sample set training of the historical travel information; and, if The personalized information does not include the search behavior feature, and the prediction model is determined to be a second prediction model, wherein the second prediction model is obtained based on the historical motion trajectory training.
  • the apparatus 600 for predicting information described above also includes other well-known structures, such as processors, memories, etc., which are not shown in FIGS. 6 and 7 in order to unnecessarily obscure the embodiments of the present disclosure. show.
  • the units involved in the embodiments of the present application may be implemented by software or by hardware.
  • the described modules may also be provided in the processor, for example, as a processor including an information acquisition module, a model determination module, and an information prediction module.
  • the name of these modules does not constitute a limitation on the module itself under certain circumstances.
  • the information acquisition module can also be described as "configured for acquisition.
  • the module of the user's personalized information is not constitute a limitation on the module itself under certain circumstances.
  • the present application further provides a computer readable storage medium, which may be a computer readable storage medium included in the apparatus described in the foregoing embodiment, or may exist separately, not A computer readable storage medium that is assembled into a terminal.
  • the computer readable storage medium stores one or more programs that are used by one or more processors to perform the method of information prediction described herein.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Navigation (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请公开了一种信息预测的方法和装置。所述方法的一具体实施方式包括：获取用户的个性化信息，其中，所述个性化信息至少包括当前运动轨迹；根据所述个性化信息，确定对应的预测模型；基于所述个性化信息及所述预测模型，预测用户的出行信息。该实施方式可以提高信息预测的有效性。

Description

信息预测的方法和装置
相关申请的交叉引用
本申请要求于2015年08月25日提交的中国专利申请号为“201510527644.0”的优先权，其全部内容作为整体并入本申请中。
技术领域
本申请涉及计算机技术领域，具体涉及通信技术领域，尤其涉及一种信息预测的方法和装置。
背景技术
智能终端即移动智能终端的简称，拥有接入互联网能力，通常搭载各种操作系统，可根据用户需求定制各种功能。随着互联网技术和智能移动终端的发展，通过植入智能终端的各种传感器采集用户数据，通过通信技术传输用户数据，进而对用户行为进行分析，成为普遍使用的技术手段。例如，有些智能终端可以通过全球定位系统(Global Positioning System，GPS)传感器采集用户的位置数据，对用户进行定位，进而根据区域特点向用户推送信息，有些智能终端可以通过血压传感器测量用户的血压数据，并传送给后台服务器或者预定的其他终端，进而对用户的血压数据进行监测，等等。
其中，现有技术在对用户的位置数据的分析中，往往通过终端的历史运动轨迹或历史停留位置数据对用户的未来位置进行预测。现有的对用户的未来位置进行预测的技术中，只考虑用户的历史停留位置数据，在对用户未曾到达过的位置的预测中，难以准确给出预测结果。因此，这种信息预测方法存在着终端数据利用不足，导致信息预测的有效性较低的问题。
发明内容
本申请的目的在于提出一种改进的信息预测的方法和装置，来解决以上背景技术部分提到的技术问题。
一方面，本申请提供了一种信息预测的方法，所述方法包括：获取用户的个性化信息，其中，所述个性化信息至少包括当前运动轨迹；根据所述个性化信息，确定对应的预测模型；基于所述个性化信息及所述预测模型，预测用户的出行信息。
在一些实施例中，所述个性化信息还包括：历史运动轨迹。
在一些实施例中，所述根据所述个性化信息，确定对应的预测模型包括：检测所述个性化信息是否还包括搜索行为特征，其中，所述搜索行为特征至少包括所搜索的地理信息点的位置特征；若包括，确定所述预测模型为第一预测模型，其中，所述第一预测模型基于历史搜索行为特征和历史出行信息的样本集训练获得；若不包括，确定所述预测模型为第二预测模型，其中，所述第二预测模型基于历史运动轨迹训练获得。
在一些实施例中，所述搜索行为特征还包括以下至少一项：对地理信息点进行搜索的时间特征、对地理信息点进行搜索时所处的位置、对所搜索的地理信息点进行搜索的次数、对同一地理信息点搜索的时间间隔、是否基于所搜索的地理信息点进行路径规划、所搜索的地理信息点的类别。
在一些实施例中，所述基于所述个性化信息及所述预测模型，预测用户的出行信息包括：基于所述搜索行为特征及所述第一预测模型，获取对用户出行信息的第一预测结果；基于所述当前运动轨迹，对所述第一预测结果进行调整，确定用户到达所搜索的地理信息点的时间和到达的概率。
在一些实施例中，所述基于所述个性化信息及所述预测模型，预测用户的出行信息包括：基于所述历史运动轨迹获取停留点转移概率矩阵；根据所述当前运动轨迹和所述停留点转移概率矩阵确定用户出现在任意停留点的概率和时间。
在一些实施例中，所述方法还包括：获取用户的实际出行信息；基于所述实际出行信息，更新所述预测模型。
第二方面，本申请提供了一种信息预测的装置，所述装置包括：信息获取模块，配置用于获取用户的个性化信息，其中，所述个性化信息至少包括当前运动轨迹；模型确定模块，配置用于根据所述个性化信息，确定对应的预测模型；信息预测模块，配置用于基于所述个性化信息及所述预测模型，预测用户的出行信息。
在一些实施例中，所述个性化信息还包括：历史运动轨迹。
在一些实施例中，所述模型确定模块包括：检测单元，配置用于检测所述个性化信息是否还包括搜索行为特征，其中，所述搜索行为特征至少包括所搜索的地理信息点的位置特征；确定单元，配置用于若所述个性化信息包括搜索行为特征，确定所述预测模型为第一预测模型，其中，所述第一预测模型基于历史搜索行为特征和历史出行信息的样本集训练获得；以及，若所述个性化信息不包括搜索行为特征，确定所述预测模型为第二预测模型，其中，所述第二预测模型基于历史运动轨迹训练获得。
在一些实施例中，所述搜索行为特征还包括以下至少一项：对地理信息点进行搜索的时间特征、对地理信息点进行搜索时所处的位置、对所搜索的地理信息点进行搜索的次数、对同一地理信息点搜索的时间间隔、是否基于所搜索的地理信息点进行路径规划、所搜索的地理信息点的类别。
在一些实施例中，所述信息预测模块包括：第一预测单元，配置用于基于所述搜索行为特征及所述第一预测模型，获取对用户出行信息的第一预测结果；调整单元，配置用于基于所述当前运动轨迹，对所述第一预测结果进行调整，确定用户到达所搜索的地理信息点的时间和到达的概率。
在一些实施例中，所述信息预测模块包括：转移概率矩阵获取模块，配置用于基于所述历史运动轨迹获取停留点转移概率矩阵；第二预测单元，配置用于根据所述当前运动轨迹和所述停留点转移概率矩阵确定用户出现在任意停留点的概率和时间。
在一些实施例中，所述装置还包括：实际出行信息获取模块，配置用于获取用户的实际出行信息；模型更新模块，配置用于基于所述 实际出行信息，更新所述预测模型。
本申请提供的信息预测的方法和装置，通过获取用户的个性化信息，其中，该个性化信息至少包括当前运动轨迹，接着根据上述个性化信息，确定对应的预测模型，然后基于上述个性化信息及所述预测模型，预测用户的出行信息，由于引入了当前运动轨迹等个性化信息，从而提高了信息预测的有效性。
附图说明
通过阅读参照以下附图所作的对非限制性实施例的详细描述，本申请的其它特征、目的和优点将会变得更明显：
图1示出了可以应用本申请实施例的示例性系统架构；
图2是根据本申请的信息预测的方法的一个实施例的流程图；
图3是根据本申请的信息预测的方法的运动轨迹及停留点的示意图；
图4是根据本申请的信息预测的方法的一个应用场景的示意图；
图5是根据本申请的信息预测的方法的又一个实施例的流程图；
图6是根据本申请的信息预测的装置的一个实施例的结构示意图；
图7是根据本申请的信息预测的装置的又一个实施例的结构示意图。
具体实施方式
下面结合附图和实施例对本申请作进一步的详细说明。可以理解的是，此处所描述的具体实施例仅仅用于解释相关发明，而非对该发明的限定。另外还需要说明的是，为了便于描述，附图中仅示出了与有关发明相关的部分。
需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本申请。
图1示出了可以应用本申请实施例的示例性系统架构100。
如图1所示，系统架构100可以包括终端设备101、102、网络103、服务器104。网络103用以在终端设备101、102和服务器104之间提供通信链路的介质。网络103可以包括各种连接类型，例如有线、无线通信链路或者光纤电缆等等。
第一用户可以使用终端设备101、102通过网络103与服务器104交互，以接收或发送消息等。终端设备101、102上可以安装位置信息的监控或管理工具、信息预测应用的客户端、地图类应用、搜索类应用等。服务器104可以根据终端设备101、102安装的位置信息的监控或管理工具获取第一用户的当前运动轨迹，并对第一用户的出行信息进行预测，或者可以接收终端设备101、102安装的信息预测的客户端预测的第一用户的出行信息。服务器104还可以对终端设备101、102或服务器104预测的至少一个第一用户的出行信息进行统计、分析。可选地，服务器104可以基于对至少一个第一用户的出行信息的统计、分析向第一用户推送信息。
可选地，系统架构100还可以包括终端设备105。终端设备105通过网络103与服务器104交互，以接收或发送消息等。服务器104可以将终端设备101、102或服务器104预测的至少一个第一用户的出行信息发送给终端设备105，以供第二用户通过终端设备105了解至少一个第一用户的出行信息的统计特征。终端设备105也可以通过服务器104向第一用户推送信息。
终端设备101、102和终端设备105可以是各种电子设备，包括但不限于个人电脑、智能手机、智能手表、平板电脑、个人数字助理等等。
服务器104可以是提供各种服务的服务器。服务器可以对接收到的数据进行存储、分析等处理，并将处理结果反馈给终端设备。
需要说明的是，本申请实施例所提供的信息预测的方法可以由终端设备101、102执行，也可以由服务器104执行，信息预测的装置可以设置于终端设备101、102中，也可以设置于服务器104中。
应当理解，图1中的终端设备、网络和服务器的数目仅仅是示意性的。根据实现需要，可以具有任意数目的终端设备、网络和服务器。
请参考图2，其示出了信息预测的方法的一个实施例的流程200。该信息预测的方法，包括以下步骤：
步骤201，获取用户的个性化信息。
在本实施例中，电子设备(例如图1所示的终端设备101、102或服务器104)可以从本地或远程地获取用户的个性化信息。具体而言，当上述电子设备就是安装有信息预测应用的客户端或者至少安装有位置信息的监控或管理工具的终端设备(如终端设备101、102)时，其可以直接从本地获取上述个性化信息；而当上述电子设备是为位置信息的监控或管理工具，或者信息预测应用的客户端提供支持的后台服务器(如服务器104)时，其可以通过有线连接方式或者无线连接方式从终端设备获取用户的行为信息。上述无线连接方式包括但不限于3G/4G连接、WiFi连接、蓝牙连接、WiMAX连接、Zigbee连接、UWB(ultra wideband)连接、以及其他现在已知或将来开发的无线连接方式。
个性化信息，可以是具有个体特性的信息，例如用户的个人行为习惯信息(作息规律、运动轨迹、搜索行为特征)等等。在这里，用户的个性化信息主要包括可以通过对终端设备的监测或者通过对用户在终端设备上的操作的监测获取的用户的个人行为习惯信息。
上述个性化信息至少可以包括当前运动轨迹。运动轨迹可以表示物体从开始位置到结束位置所经过的路线组成的空间特征。运动轨迹可以由运动轨迹方向(如移动方向)、运动轨迹形式(如直线轨迹)表示。在本实施例中，将用户持有的终端设备在不同时间点所在的位置作为轨迹点，用户的运动轨迹可以通过轨迹点的连线来表示。如图3所示，标号301指示的是一个轨迹点，各轨迹点之间的连线构成了运动轨迹302，图3中连线的箭头表示运动轨迹的方向。
实践中，每个终端设备都可以具有一个身份码，例如智能手机的移动设备国际身份码(International Mobile Equipment Identity，IMEI)。电子设备可以将该身份码作为终端设备的身份标识号码，用以区分不同的终端设备。其中，用户的地理位置信息可以通过终端设备内置的定位单元获取。该定位单元可以通过硬件实现，例如可以是终端设备 所集成的定位器(如全球卫星定位系统GPS终端)；也可以通过软件实现，例如可以是定位应用(如可以运行在安卓操作系统的51定位终端)；还可以通过软件、硬件结合实现。当终端的位置信息的获取通过定位应用实现时，定位应用可以根据终端接入网络(如WIFI网络、移动数据蜂窝网络等)的IP地址(Internet Protocol Address，网际协议地址)自动查找相应的基站从而获取终端的当前位置，也可以通过GPS定位数据获取终端的当前位置。可选地，用户的个性化信息还可以包括历史运动轨迹、搜索行为特征等。
用户的运动轨迹可以包括运动路线和停留点。运动路线即多个位置的连线。停留点是根据运动路线中轨迹点的集中情况来确定的。例如，将多个时间相邻的速度均在速度阈值以下的轨迹点合并为一个停留位置，如果这些停留位置中任两个停留位置的中心之间的距离小于距离阈值，并且，多个停留位置中第一个停留位置的停留时间的开始时刻到最后一个停留位置的结束时刻之间的时间间隔大于时间阈值时，则将这多个停留位置组成的区域判定为停留点。其中，停留点的位置可以用这多个停留位置组成的区域的中心位置来表示。如图3所示，圆形区域303、304内的各轨迹点组成的区域为一个停留点。可选地，轨迹点的速度可以通过两个轨迹点之间的距离除以经过两个轨迹点所用的时间来计算。
步骤202，根据上述个性化信息，确定对应的预测模型。
在本实施例中，电子设备接着可以对步骤201中获取的用户的个性化信息进行分析，确定根据用户的个性化信息对用户的出行信息进行预测的预测模型。
本领域技术人员可以理解，如果用户的个性化信息不同，电子设备可以确定不同的信息预测的预测模型。比如一个新的终端设备，用户没有通过该终端设备进行过搜索行为，也无法获取用户的历史运动轨迹，则电子设备获得的个性化信息可能只有当前运动轨迹。此时，电子设备可以确定预测模型为根据地图上的地理信息点之间的路线建立的预测模型。电子设备也可以确定预测模型为基于对大量用户的出行路线进行统计而建立的模型。
如果用户对终端设备持有一段时间，而没有进行过搜索行为，则电子设备所获取的个性化信息可能包含历史运动轨迹和当前运动轨迹。在这种情况下，电子设备可以具有预先根据用户的历史运动轨迹训练的停留点转移概率模型。如果电子设备获取的个性化信息包括当前运动轨迹，则可以检测是否有使用该终端设备的用户的历史运动轨迹，若有，则将预测模型确定为该停留点转移概率模型。
以无限的隐马尔科夫模型(infinite Hidden Markov Model，iHMM)为例，将每个停留点作为无限的隐马尔科夫模型的一个状态，由具有相应概率密度分布的状态序列产生的观测向量的观测向量序列作为预测模型。电子设备根据用户的历史运动轨迹训练停留点转移概率模型的过程可以为：首先根据用户的历史运动轨迹得到用户的历史停留点；接着根据历史运动轨迹中的运动路线得到用户在每个历史停留点向其它各历史停留点转移的顺序和次数，从而得到各历史停留点之间的转移序列，进而得到各历史停留点之间的转移概率矩阵；进一步地，对用户处于每个历史停留点时，用α、β、γ三个超参数控制产生三种预测结果，得到预测用户在当前运动轨迹基础上的停留点预测模型。这三种预测结果即停留在原停留点、转移至其它历史停留点、到达新的停留点。
具体地，作为一种可选的实现方式的示例，用α表示用户到达在每个历史停留点的先验概率，用β表示用户访问历史停留点以外的地点的概率—即产生新的停留点的概率，用γ表示停留点数量的期望值，用nij表示从停留点i转移到停留点j的概率，用
Figure PCTCN2015096131-appb-000001
表示停留点j为地图上已知的停留点的概率，则当用户当前位置为停留点i时，用户向各停留点转移的概率为：
停留在当前停留点的概率为：
Figure PCTCN2015096131-appb-000002
从当前停留点转移到其他历史停留点转移的概率为：
Figure PCTCN2015096131-appb-000003
从当前停留点转移到新的停留点的概率为：
Figure PCTCN2015096131-appb-000004
其中，从当前停留点转移到新的停留点时，新的停留点为地图上 已知的地点的概率为：
Figure PCTCN2015096131-appb-000005
新的停留点为地图上未知的停留点的概率为：
Figure PCTCN2015096131-appb-000006
步骤203，基于上述个性化信息及上述预测模型，预测用户的出行信息。
在本实施例中，电子设备接着可以通过确定的预测模型，对获取的个性化信息进行匹配和/或计算，预测出用户可能的出行信息，例如可以是出行时间、出行地点和/或出行地概率等等。
以用户的个性化信息包括历史运动轨迹和当前运动轨迹，预测模型为iHMM为例：在步骤202中已经通过公式详细介绍了iHMM的预测用户到达各停留点的概率的原理，进行预测时，电子设备可以根据用户的当前运动轨迹建立停留点转移序列作为待预测样本，并使用上述的表示预测用户到达各停留点的概率的原理的各公式组成的模型对待检测样本进行评估，获得用户停留在原停留点、转移到其他历史停留点或到达新的停留点的概率。预测的用户转移到的其他停留点即为出行地点，转移到其他停留点的概率即为出行概率。出行时间可以根据用户当前运行轨迹中的速度和两个停留点之间的距离计算，也可以根据停留点之间的转移所需的时间进行计算(如对在两个历史停留点之间的转移时间取平均值)。
在本实施例的一些实现方式中，电子设备在根据相应的个性化信息和预测模型对用户出行信息进行预测后，还可以继续监测用户的运动轨迹，以获取用户的实际出行信息，包括出行路线、停留点等，并将当前运动轨迹加入历史运动轨迹，将新的停留点加入历史停留点，同时更新相应的预测模型，以新的出行数据增加预测模型的样本容量，以使预测模型的预测结果越来越准确。
如图4所示，给出了本实施例的一个应用场景的示意图。如图4所示，在标号401中，电子设备通过用户所使用的终端设备获取用户的当前运动轨迹；接着，电子设备检测是否获取过用户的历史运动轨迹，若是，确定预测模型为根据用户的历史运动轨迹训练的停留点转移概率模型，如标号402所示；然后，如标号403所示，电子设备将 当前运动轨迹中包含的停留点序列作为待预测样本，通过停留点转移概率模型进行评估，预测用户的出行地点、出行概率、出行时间等出行信息；进一步地，电子设备可以将所预测的出行概率最大的出行地点作为目标地点，并获取目标地点的相关信息，例如当前位置到目标地点的路况信息、目标地点的停车场位置信息等，推送给用户。
本实施例的信息推送方法，通过引入当前运动轨迹等个性化信息，可以提高对用户出行信息预测的有效性。
进一步参考图5，其示出了信息预测的方法的又一个实施例的流程500。该地图显示方法的流程500，包括以下步骤：
步骤501，获取用户的个性化信息。
在本实施例中，电子设备(例如图1所示的终端设备101、102或服务器104)可以从本地或远程地获取用户的个性化信息。其中，用户的个性化信息可以包括但不限于用户的作息规律、运动轨迹、搜索行为特征等。这里，个性化信息至少包括当前运动轨迹。
步骤502，检测上述个性化信息是否还包括搜索行为特征。
在本实施例中，电子设备接着可以对上述个性化信息进一步检测，以确定上述个性化信息中除了包括当前运动轨迹外还包括哪些信息，例如是否还包括搜索行为特征。
搜索行为特征是用户通过终端设备进行搜索时的行为特征，例如，搜索时使用的搜索词。在本实施例中，如果该搜索词为与地理信息点(如香山)相关的词，搜索行为特征至少包括所搜索的地理信息点的位置特征，搜索特征还可以包括但不限于以下至少一项：对地理信息点进行搜索的时间特征(例如工作日或休息日)、对地理信息点进行搜索时所处的位置(例如地理位置坐标等)、对所搜索的地理信息点进行搜索的次数、对同一地理信息点搜索的时间间隔、是否基于所搜索的地理信息点进行路径规划(例如搜索路线)、所搜索的地理信息点的类别(例如企业、景点、饭店等)、对地理信息点进行搜索的天气特征(如晴天或下雨等)，等等。
步骤503，若包括，确定预测模型为第一预测模型，若不包括，确定预测模型为第二预测模型。
在本实施例中，电子设备可以根据对个性化信息的检测结果，确定相应的预测模型：如果检测到个性化信息中包括搜索行为特征，则电子设备可以确定预测模型为基于历史搜索行为特征和历史出行信息的样本集训练获得的第一预测模型；如果检测到个性化信息中不包括搜索行为特征，则电子设备可以确定预测模型为基于历史运动轨迹训练获得的第二预测模型。
其中，作为第一预测模型的一个示例，第一预测模型可以包括搜索行为特征中各项的权重的模型，上述权重根据用户的历史搜索行为特征和历史出行信息计算。例如，搜索行为特征包括对地理信息点进行搜索的时间特征，则电子设备根据历史出行信息分别获取用户休息日出行的概率和工作日出行的概率。休息日或工作日出行的概率例如可以是一个时间段内休息日或工作日出行的天数与总休息日或工作日的天数的比值。当对地理信息点进行搜索的时间特征为工作日时，可以将用户工作日出行的概率作为工作日的对地理信息点进行搜索的权重。同样，对搜索行为特征的各项分别获得一个权重。当电子设备获取的个性化信息包括搜索行为特征时，可以通过第一预测模型得到搜索行为特征中各项的预测权重。可选地，电子设备可以将预设时间段(如3个月)内的搜索行为特征作为本次预测所获取的搜索行为特征，并按照搜索时间与当前时间的时间间隔为搜索时间计算权重，例如搜索时间与当前时间的时间间隔越大，则对应的权重越小。
作为第一预测模型的另一个示例，第一预测模型还可以是根据用户的历史搜索行为特征和历史出行信息通过机器学习训练的决策树结构的模型，例如梯度提升(gradient boost)决策树。决策树结构的模型可以通过以下方法训练获得：将样本中的搜索行为特征结合搜索后一定时间段(如3个月)内的出行信息进行分类，建立不同的子树，子树中每个叶子节点对应一个预测分数，该预测分数可以是搜索行为特征到达该节点的概率，从而获得初始预测模型；对前一次获取的预测模型中分错的类增加权重，再次划分，循环执行该步骤，直到模型对样本中的搜索行为特征预测的出行信息与实际的出行信息对比准确率达到预设阈值(如98％)。
在一些可选的实现方式中，进行模型训练时，电子设备可以对距离进行约束，即：如果所搜索的地理信息点与实际到达的地理位置之间的距离小于约束距离，则认为用户到达了所搜索的地理位置。可选地，约束距离可以是当前位置与所搜索的位置之间的距离与约束系数的乘积，也可以是所搜索的地理位置与实际到达的地理位置之间实际距离。例如，约束距离可以为200米，当用户所搜索的地理位置与当前位置的距离小于第一预设距离(如100公里)，或者同属于一个城市时，用户实际到达了与所搜索的地理位置相距200米以内的地理位置时，则电子设备认为用户实际到达了所搜索的地理位置。再例如，约束距离可以为用户所搜索的地理位置与当前位置的距离与约束系数(如0.05)的乘积，当用户所搜索的地理位置与当前位置的距离大于第二预设距离(如500公里)时，如果用户到达的地理位置与所搜索的地理位置的距离小于用户所搜索的地理位置与当前位置之间的距离与约束系数的乘积时，则电子设备认为用户实际到达了所搜索的地理位置。如：当前位置为广州的用户搜索了“北京机场”，则如果用户所到达的地理位置与北京机场的距离小于广州与北京的距离乘以约束系数时，电子设备认为用户实际到达了所搜索的地理位置。
关于个性化信息中不包括搜索行为特征时，前述实施例已经给出了一些实现方式。第二预测模型可以是根据地图上的地理信息点之间的路线建立的预测模型，也可以是停留点转移概率模型(如IHMM)，在此不再赘述。
步骤504，基于上述个性化信息及上述预测模型，预测用户的出行信息。
在本实施例中，电子设备接着可以通过确定的预测模型，对获取的个性化信息进行匹配和/或计算，预测出用户可能的出行信息，例如出行时间、出行地点和/或出行概率。
以个性化信息中包括搜索行为特征，第一预测模型为包括根据用户的历史搜索行为特征和历史出行信息计算搜索行为特征中各项的权重的模型为例，电子设备进行本次预测时，可以将一定时间段(如3个月)内的搜索行为特征进行统计，对每次进行搜索的搜索行为特征 中各项的权重相加，得到预测系数。可选地，电子设备可以将对每次进行搜索的搜索行为特征获得的预测系数与一定时间段(如3个月)内的搜索行为特征的预测系数相除得到的商，作为用户到达搜索行为特征包含的地理信息点的出行概率。其中，出行时间可以根据用户的出行习惯而定(如休息日的概率较大)。
在本实施例的可选实现方式中，电子设备可以将基于搜索行为特征及上述第一预测模型获取的对用户出行信息的预测结果作为第一预测结果，接着对当前运动轨迹的方向进行判断，根据当前运动轨迹的方向、前一停留点的位置和所搜索的地理信息点的位置之间的关系对第一预测结果进行调整。例如，当用户的当前运动轨迹的方向与前一停留点的位置到所搜索的地理信息点的位置的方向相同时，第一预测结果中用户到达所搜索的地理信息点的概率增大，到达时间可由当前运动轨迹中的运动速度和当前位置与所搜索的地理信息点之间的距离进行计算。
在本实施例中，上述实现流程中的步骤501、步骤504分别与前述实施例中的步骤201、步骤203基本相同，在此不再赘述。
从图5中可以看出，与图2对应的实施例不同的是，本实施例中的信息预测的方法的流程500多出了检测个性化信息是否还包括搜索行为特征的步骤502，同时步骤503与步骤202不同的是，步骤503将预测模型分为包括搜索行为特征的第一预测模型和不包括搜索行为特征的第二预测模型。通过增加的步骤502和与步骤202不同的步骤503，本实施例描述的方案可以引入更多的用户的个性化信息数据，从而实现更有效的用户出行信息的预测。该实施例中，电子设备可以将对大量用户的出行信息的预测结果进行汇总，进而推送给各地理信息点的管理人员或商户。
应当注意，尽管在附图中以特定顺序描述了本申请方法的操作，但是，这并非要求或者暗示必须按照该特定顺序来执行这些操作，或是必须执行全部所示的操作才能实现期望的结果。相反，流程图中描绘的步骤可以改变执行顺序。附加地或备选地，可以省略某些步骤，将多个步骤合并为一个步骤执行，和/或将一个步骤分解为多个步骤执 行。
进一步参考图6，作为对上述各图所示方法的实现，本申请提供了一种信息预测的装置的一个实施例，该装置实施例与图2所示的方法实施例相对应，该装置具体可以应用于电子设备中。
如图6所示，本实施例所述的信息预测的装置600包括：信息获取模块601、模型确定模块602和信息预测模块603。其中，信息获取模块601配置用于获取用户的个性化信息，其中，个性化信息至少包括当前运动轨迹；模型确定模块602配置用于根据上述个性化信息，确定对应的预测模型；信息预测模块603配置用于基于上述个性化信息及上述预测模型，预测用户的出行信息。
进一步地，如图7所示，模型确定模块602包括检测单元6021和确定单元6022。检测单元6021配置用于检测个性化信息是否还包括搜索行为特征，其中，搜索行为特征至少包括所搜索的地理信息点的位置特征。确定单元6022配置用于若上述个性化信息包括搜索行为特征，确定预测模型为第一预测模型，其中，第一预测模型基于历史搜索行为特征和历史出行信息的样本集训练获得；以及，若上述个性化信息不包括搜索行为特征，确定预测模型为第二预测模型，其中，第二预测模型基于历史运动轨迹训练获得。
应当理解，图6、图7中记载的诸模块或单元分别与参考图2、图4描述的方法中的各个步骤相对应。由此，上文针对方法描述的操作和特征同样适用于图6、图7中的装置及其中包含的单元或模块，在此不再赘述。
本领域技术人员可以理解，上述信息预测的装置600还包括一些其他公知结构，例如处理器、存储器等，为了不必要地模糊本公开的实施例，这些公知的结构在图6、图7中未示出。
本申请实施例中所涉及到的单元可以通过软件的方式实现，也可以通过硬件的方式来实现。所描述的模块也可以设置在处理器中，例如，可以描述为：一种处理器包括信息获取模块，模型确定模块和信息预测模块。其中，这些模块的名称在某种情况下并不构成对该模块本身的限定，例如，信息获取模块还可以被描述为“配置用于获取用 户的个性化信息的模块”。
作为另一方面，本申请还提供了一种计算机可读存储介质，该计算机可读存储介质可以是上述实施例中所述装置中所包含的计算机可读存储介质；也可以是单独存在，未装配入终端中的计算机可读存储介质。所述计算机可读存储介质存储有一个或者一个以上程序，所述程序被一个或者一个以上的处理器用来执行描述于本申请的信息预测的方法。
以上描述仅为本申请的较佳实施例以及对所运用技术原理的说明。本领域技术人员应当理解，本申请中所涉及的发明范围，并不限于上述技术特征的特定组合而成的技术方案，同时也应涵盖在不脱离所述发明构思的情况下，由上述技术特征或其等同特征进行任意组合而形成的其它技术方案。例如上述特征与本申请中公开的(但不限于)具有类似功能的技术特征进行互相替换而形成的技术方案。

Claims (16)

  1. 一种信息预测的方法，其特征在于，所述方法包括：
    获取用户的个性化信息，其中，所述个性化信息至少包括当前运动轨迹；
    根据所述个性化信息，确定对应的预测模型；
    基于所述个性化信息及所述预测模型，预测用户的出行信息。
  2. 根据权利要求1所述的方法，其特征在于，所述个性化信息还包括：历史运动轨迹。
  3. 根据权利要求2所述的方法，其特征在于，所述根据所述个性化信息，确定对应的预测模型包括：
    检测所述个性化信息是否还包括搜索行为特征，其中，所述搜索行为特征至少包括所搜索的地理信息点的位置特征；
    若包括，确定所述预测模型为第一预测模型，其中，所述第一预测模型基于历史搜索行为特征和历史出行信息的样本集训练获得；
    若不包括，确定所述预测模型为第二预测模型，其中，所述第二预测模型基于历史运动轨迹训练获得。
  4. 根据权利要求3所述的方法，其特征在于，所述搜索行为特征还包括以下至少一项：对地理信息点进行搜索的时间特征、对地理信息点进行搜索时所处的位置、对所搜索的地理信息点进行搜索的次数、对同一地理信息点搜索的时间间隔、是否基于所搜索的地理信息点进行路径规划、所搜索的地理信息点的类别。
  5. 根据权利要求4所述的方法，其特征在于，所述基于所述个性化信息及所述预测模型，预测用户的出行信息包括：
    基于所述搜索行为特征及所述第一预测模型，获取对用户出行信息的第一预测结果；
    基于所述当前运动轨迹，对所述第一预测结果进行调整，确定用户到达所搜索的地理信息点的时间和到达的概率。
  6. 根据权利要求4所述的方法，其特征在于，所述基于所述个性化信息及所述预测模型，预测用户的出行信息包括：
    基于所述历史运动轨迹获取停留点转移概率矩阵；
    根据所述当前运动轨迹和所述停留点转移概率矩阵确定用户出现在各停留点的概率和时间。
  7. 根据权利要求1-6中任一所述的方法，其特征在于，所述方法还包括：
    获取用户的实际出行信息；
    基于所述实际出行信息，更新所述预测模型。
  8. 一种信息预测的装置，其特征在于，所述装置包括：
    信息获取模块，配置用于获取用户的个性化信息，其中，所述个性化信息至少包括当前运动轨迹；
    模型确定模块，配置用于根据所述个性化信息，确定对应的预测模型；
    信息预测模块，配置用于基于所述个性化信息及所述预测模型，预测用户的出行信息。
  9. 根据权利要求8所述的装置，其特征在于，所述个性化信息还包括：历史运动轨迹。
  10. 根据权利要求9所述的装置，其特征在于，所述模型确定模块包括：
    检测单元，配置用于检测所述个性化信息是否还包括搜索行为特征，其中，所述搜索行为特征至少包括所搜索的地理信息点的位置特征；
    确定单元，配置用于若所述个性化信息包括搜索行为特征，确定所述预测模型为第一预测模型，其中，所述第一预测模型基于历史搜索行为特征和历史出行信息的样本集训练获得；以及，若所述个性化信息不包括搜索行为特征，确定所述预测模型为第二预测模型，其中，所述第二预测模型基于历史运动轨迹训练获得。
  11. 根据权利要求10所述的装置，其特征在于，所述搜索行为特征还包括以下至少一项：对地理信息点进行搜索的时间特征、对地理信息点进行搜索时所处的位置、对所搜索的地理信息点进行搜索的次数、对同一地理信息点搜索的时间间隔、是否基于所搜索的地理信息点进行路径规划、所搜索的地理信息点的类别。
  12. 根据权利要求11所述的装置，其特征在于，所述信息预测模块包括：
    第一预测单元，配置用于基于所述搜索行为特征及所述第一预测模型，获取对用户出行信息的第一预测结果；
    调整单元，配置用于基于所述当前运动轨迹，对所述第一预测结果进行调整，确定用户到达所搜索的地理信息点的时间和到达的概率。
  13. 根据权利要求11所述的装置，其特征在于，所述信息预测模块包括：
    转移概率矩阵获取模块，配置用于基于所述历史运动轨迹获取停留点转移概率矩阵；
    第二预测单元，配置用于根据所述当前运动轨迹和所述停留点转移概率矩阵确定用户出现在任意停留点的概率和时间。
  14. 根据权利要求8-13中任一所述的装置，其特征在于，所述装置还包括：
    实际出行信息获取模块，配置用于获取用户的实际出行信息；
    模型更新模块，配置用于基于所述实际出行信息，更新所述预测 模型。
  15. 一种设备，包括：
    处理器；和
    存储器，
    所述存储器中存储有能够被所述处理器执行的计算机可读指令，在所述计算机可读指令被执行时，所述处理器执行权利要求1至7中任一项所述的方法。
  16. 一种非易失性计算机存储介质，所述计算机存储介质存储有能够被处理器执行的计算机可读指令，当所述计算机可读指令被处理器执行时，所述处理器执行权利要求1至7中任一项所述的方法。
PCT/CN2015/096131 2015-08-25 2015-12-01 信息预测的方法和装置 WO2017031856A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510527644.0 2015-08-25
CN201510527644.0A CN105183800A (zh) 2015-08-25 2015-08-25 信息预测的方法和装置

Publications (1)

Publication Number Publication Date
WO2017031856A1 true WO2017031856A1 (zh) 2017-03-02

Family

ID=54905883

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/096131 WO2017031856A1 (zh) 2015-08-25 2015-12-01 信息预测的方法和装置

Country Status (2)

Country Link
CN (1) CN105183800A (zh)
WO (1) WO2017031856A1 (zh)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109242043A (zh) * 2018-09-29 2019-01-18 北京京东金融科技控股有限公司 用于生成信息预测模型的方法和装置
CN110704469A (zh) * 2019-10-22 2020-01-17 北京明略软件系统有限公司 一种预警等级的更新方法、更新装置及可读存储介质
CN110766154A (zh) * 2019-09-18 2020-02-07 北京邮电大学 一种行人航迹推断方法、装置、设备及存储介质
CN111427932A (zh) * 2020-04-02 2020-07-17 南方科技大学 出行预测方法、装置、设备和存储介质
CN111723123A (zh) * 2019-03-20 2020-09-29 杭州海康威视数字技术股份有限公司 轨迹预测方法、装置、电子设备及存储介质
CN111736481A (zh) * 2020-07-14 2020-10-02 北京自如信息科技有限公司 基于用户行为特征的模型训练、智能家居控制方法及装置
CN111861538A (zh) * 2020-06-02 2020-10-30 北京嘀嘀无限科技发展有限公司 信息推送方法、装置、电子设备和存储介质
CN112070288A (zh) * 2020-08-26 2020-12-11 北京百度网讯科技有限公司 出发时间预估方法、装置、设备以及存储介质
CN112419078A (zh) * 2020-12-03 2021-02-26 中国人寿保险股份有限公司 产品推荐方法、装置和电子设备
CN112508592A (zh) * 2019-09-13 2021-03-16 东芝泰格有限公司 区域转移预测装置及存储介质
CN113487163A (zh) * 2021-06-30 2021-10-08 支付宝(杭州)信息技术有限公司 基于地理位置信息进行业务预测的方法及装置
CN113498010A (zh) * 2020-03-19 2021-10-12 北京地平线机器人技术研发有限公司 基于行驶目的地信息的通信建立方法及装置
CN113496167A (zh) * 2020-04-02 2021-10-12 北京京东乾石科技有限公司 确定目标运动轨迹的方法和装置
CN113743845A (zh) * 2020-05-29 2021-12-03 京东城市(北京)数字科技有限公司 配送区域划分方法及装置、电子设备、存储介质
CN114265527A (zh) * 2021-12-20 2022-04-01 中国农业银行股份有限公司 一种预测鼠标点击位置的方法、装置、介质及电子设备
CN115690163A (zh) * 2023-01-04 2023-02-03 中译文娱科技(青岛)有限公司 基于影像内容的目标跟踪方法、系统及存储介质
CN116205376A (zh) * 2023-04-27 2023-06-02 北京阿帕科蓝科技有限公司 行为预测方法、行为预测模型的训练方法和装置
CN116680346A (zh) * 2022-09-23 2023-09-01 荣耀终端有限公司 运动轨迹分析方法、设备及介质

Families Citing this family (36)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105893537B (zh) * 2016-03-31 2019-06-07 百度在线网络技术（北京）有限公司 地理信息点的确定方法和装置
CN105912702A (zh) * 2016-04-26 2016-08-31 南京酷派软件技术有限公司 位置推荐方法及系统
CN106055607B (zh) * 2016-05-25 2020-05-19 百度在线网络技术（北京）有限公司 用户到访预测模型建立、用户到访预测方法和装置
SG10201912862TA (en) * 2016-06-21 2020-02-27 Amadeus Sas Data warehouse for mining search query logs
CN107665375A (zh) * 2016-07-29 2018-02-06 滴滴（中国）科技有限公司 代驾司机到达代驾乘客位置的时间预估方法及装置
CN106227889A (zh) * 2016-08-15 2016-12-14 华云科技有限公司 一种轨迹停留点分析提取方法
CN106469205B (zh) * 2016-08-31 2020-06-05 百度在线网络技术（北京）有限公司 一种确定用户的地理位置信息的方法与装置
CN107861957B (zh) * 2016-09-22 2020-10-02 杭州海康威视数字技术股份有限公司 一种数据分析方法及装置
CN106446157B (zh) * 2016-09-22 2020-01-21 北京百度网讯科技有限公司 行程目的地推荐方法和装置
CN106384022B (zh) * 2016-11-23 2019-03-15 东软集团股份有限公司 出行行为预测方法及装置
CN106709606B (zh) * 2016-12-29 2020-10-30 平安科技（深圳）有限公司 个性化场景预测方法及装置
WO2018141093A1 (zh) * 2017-02-04 2018-08-09 华为技术有限公司 移动网络优化处理方法、设备及系统
CN106767835B (zh) * 2017-02-08 2020-09-25 百度在线网络技术（北京）有限公司 定位方法和装置
CN106875066B (zh) * 2017-02-28 2021-06-11 百度在线网络技术（北京）有限公司 用车出行行为的预测方法、装置、服务器以及存储介质
CN107688865A (zh) * 2017-07-31 2018-02-13 上海恺英网络科技有限公司 识别网络游戏中潜在高消费用户的方法及设备
CN107622083A (zh) * 2017-08-10 2018-01-23 深圳市悦动天下科技有限公司 一种基于大数据的计步方法、装置
CN109558961B (zh) * 2017-09-25 2023-05-02 阿里巴巴集团控股有限公司 确定位置信息的方法和系统、存储介质、处理器以及装置
CN109752011A (zh) * 2017-11-06 2019-05-14 上海擎感智能科技有限公司 一种用于车载的信息交互方法与系统
CN108280160A (zh) * 2018-01-17 2018-07-13 北京小度信息科技有限公司 一种消息推送方法、装置、服务器和计算机可读存储介质
CN108749596B (zh) * 2018-04-11 2020-12-04 蔚来(安徽)控股有限公司 车机端启动方法、系统及装置
CN112236787B (zh) * 2018-06-08 2024-06-11 北京嘀嘀无限科技发展有限公司 用于生成个性化目的地推荐的系统和方法
CN108898520B (zh) * 2018-06-12 2021-11-02 山东众云教育科技有限公司 基于轨迹数据的学生安全监控方法和系统
CN109189693B (zh) * 2018-07-18 2020-10-30 深圳大普微电子科技有限公司 一种对lba信息进行预测的方法及ssd
CN108966148A (zh) * 2018-08-10 2018-12-07 深圳北斗应用技术研究院有限公司 出行信息的预测方法、服务器及存储介质
CN109466479B (zh) * 2018-10-26 2021-10-12 广州小鹏汽车科技有限公司 车辆控制方法、装置、终端设备和介质
CN109491284B (zh) * 2018-10-26 2021-02-19 广州小鹏汽车科技有限公司 基于用户出行习惯的车辆控制方法、装置和终端设备
CN110321343B (zh) * 2019-07-11 2023-11-14 广东工业大学 一种可穿戴设备轨迹预测方法、装置和系统
CN110426050B (zh) * 2019-08-07 2021-08-03 北京百度网讯科技有限公司 地图匹配纠正方法、装置、设备和存储介质
CN110929932A (zh) * 2019-11-20 2020-03-27 满江(上海)软件科技有限公司 基于区块链的非特定人群流动临时聚集点的预警系统
CN112866482B (zh) * 2019-11-27 2022-04-15 青岛海信移动通信技术股份有限公司 一种对象行为习惯预测的方法和终端
CN111209318B (zh) * 2020-01-02 2020-09-08 平安科技（深圳）有限公司 基于地理位置数据分析方法、装置、电子设备及存储介质
CN111329486B (zh) * 2020-03-03 2021-02-05 和宇健康科技股份有限公司 一种基于大数据的室内外安全监测系统及方法
CN111563759B (zh) * 2020-06-24 2023-09-01 浙江天演维真网络科技股份有限公司 基于ai技术的农产品溯源过程的识别与分析系统
CN113885608B (zh) * 2021-10-21 2022-06-28 奇瑞汽车股份有限公司 车辆的控制方法和装置
CN115564134B (zh) * 2022-10-21 2024-02-09 中禹规划设计有限公司 一种防汛防涝预警系统及方法
CN116321007B (zh) * 2023-03-13 2024-04-02 深圳市交投科技有限公司 出行目的预测方法、装置、设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101247392A (zh) * 2008-02-28 2008-08-20 中兴通讯股份有限公司 目标活动预测装置及方法
US20120010996A1 (en) * 2010-07-07 2012-01-12 Microsoft Corporation Recommendations and targeted advertising based upon directions requests activity and data
CN104680250A (zh) * 2015-02-11 2015-06-03 北京邮电大学 一种位置预测系统

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013192585A2 (en) * 2012-06-22 2013-12-27 Google Inc. Weather forecasts based on expected location
CN104581622B (zh) * 2013-10-28 2018-09-07 华为技术有限公司 移动用户位置预测方法及设备
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

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101247392A (zh) * 2008-02-28 2008-08-20 中兴通讯股份有限公司 目标活动预测装置及方法
US20120010996A1 (en) * 2010-07-07 2012-01-12 Microsoft Corporation Recommendations and targeted advertising based upon directions requests activity and data
CN104680250A (zh) * 2015-02-11 2015-06-03 北京邮电大学 一种位置预测系统

Cited By (26)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109242043A (zh) * 2018-09-29 2019-01-18 北京京东金融科技控股有限公司 用于生成信息预测模型的方法和装置
CN111723123A (zh) * 2019-03-20 2020-09-29 杭州海康威视数字技术股份有限公司 轨迹预测方法、装置、电子设备及存储介质
CN112508592A (zh) * 2019-09-13 2021-03-16 东芝泰格有限公司 区域转移预测装置及存储介质
CN110766154A (zh) * 2019-09-18 2020-02-07 北京邮电大学 一种行人航迹推断方法、装置、设备及存储介质
CN110704469A (zh) * 2019-10-22 2020-01-17 北京明略软件系统有限公司 一种预警等级的更新方法、更新装置及可读存储介质
CN110704469B (zh) * 2019-10-22 2022-11-11 北京明智和术科技有限公司 一种预警等级的更新方法、更新装置及可读存储介质
CN113498010A (zh) * 2020-03-19 2021-10-12 北京地平线机器人技术研发有限公司 基于行驶目的地信息的通信建立方法及装置
CN113496167A (zh) * 2020-04-02 2021-10-12 北京京东乾石科技有限公司 确定目标运动轨迹的方法和装置
CN111427932A (zh) * 2020-04-02 2020-07-17 南方科技大学 出行预测方法、装置、设备和存储介质
US11809970B2 (en) 2020-04-02 2023-11-07 South University Of Science And Technology Of China Travel prediction method and apparatus, device, and storage medium
CN111427932B (zh) * 2020-04-02 2022-10-04 南方科技大学 出行预测方法、装置、设备和存储介质
CN113743845A (zh) * 2020-05-29 2021-12-03 京东城市(北京)数字科技有限公司 配送区域划分方法及装置、电子设备、存储介质
CN111861538A (zh) * 2020-06-02 2020-10-30 北京嘀嘀无限科技发展有限公司 信息推送方法、装置、电子设备和存储介质
CN111736481A (zh) * 2020-07-14 2020-10-02 北京自如信息科技有限公司 基于用户行为特征的模型训练、智能家居控制方法及装置
CN111736481B (zh) * 2020-07-14 2023-06-23 北京自如信息科技有限公司 基于用户行为特征的模型训练、智能家居控制方法及装置
CN112070288A (zh) * 2020-08-26 2020-12-11 北京百度网讯科技有限公司 出发时间预估方法、装置、设备以及存储介质
CN112070288B (zh) * 2020-08-26 2024-04-05 北京百度网讯科技有限公司 出发时间预估方法、装置、设备以及存储介质
CN112419078A (zh) * 2020-12-03 2021-02-26 中国人寿保险股份有限公司 产品推荐方法、装置和电子设备
CN113487163A (zh) * 2021-06-30 2021-10-08 支付宝(杭州)信息技术有限公司 基于地理位置信息进行业务预测的方法及装置
CN114265527A (zh) * 2021-12-20 2022-04-01 中国农业银行股份有限公司 一种预测鼠标点击位置的方法、装置、介质及电子设备
CN114265527B (zh) * 2021-12-20 2024-02-20 中国农业银行股份有限公司 一种预测鼠标点击位置的方法、装置、介质及电子设备
CN116680346A (zh) * 2022-09-23 2023-09-01 荣耀终端有限公司 运动轨迹分析方法、设备及介质
CN116680346B (zh) * 2022-09-23 2024-04-16 荣耀终端有限公司 运动轨迹分析方法、设备及介质
CN115690163A (zh) * 2023-01-04 2023-02-03 中译文娱科技(青岛)有限公司 基于影像内容的目标跟踪方法、系统及存储介质
CN116205376A (zh) * 2023-04-27 2023-06-02 北京阿帕科蓝科技有限公司 行为预测方法、行为预测模型的训练方法和装置
CN116205376B (zh) * 2023-04-27 2023-10-17 北京阿帕科蓝科技有限公司 行为预测方法、行为预测模型的训练方法和装置

Also Published As

Publication number Publication date
CN105183800A (zh) 2015-12-23

Similar Documents

Publication Publication Date Title
WO2017031856A1 (zh) 信息预测的方法和装置
CN106462627B (zh) 根据多个位置数据报告分析语义地点和相关数据
CN108875007B (zh) 兴趣点的确定方法和装置、存储介质、电子装置
US11305780B2 (en) Road condition status prediction method, device, and server, and storage medium
Rong et al. Du-parking: Spatio-temporal big data tells you realtime parking availability
Mohamed et al. Accurate real-time map matching for challenging environments
Tang et al. Estimating the most likely space–time paths, dwell times and path uncertainties from vehicle trajectory data: A time geographic method
Karimi et al. A predictive location model for location-based services
CN105532030B (zh) 用于分析目标实体的移动的装置、系统和方法
US8612134B2 (en) Mining correlation between locations using location history
US20160227367A1 (en) Method of providing positioning data to mobile device
WO2015120746A1 (zh) 一种用户目的地点的预测方法及装置
CN105532056B (zh) 用于管理基于预测的定位方法的方法、无线设备和网络节点
KR101365993B1 (ko) 데이터처리방법, 데이터처리장치, 데이터수집방법, 및 정보제공방법
CN113906716A (zh) 雾节点资源的分配
JP2012003494A (ja) 情報処理装置、情報処理方法及びプログラム
CN102087709A (zh) 学习设备、学习方法以及程序
Shang et al. Modeling of traffic-aware travel time in spatial networks
CN111194005A (zh) 一种室内行人语义位置提取方法及预测方法
US9752888B2 (en) Method and apparatus of computing location of safe exit for moving range query in road network
WO2023147892A1 (en) Long-term accurate crowd estimation in smart cities
Amirat et al. Nextroute: a lossless model for accurate mobility prediction
GB2570853A (en) Identifying sites visited by a user device
CN108924763B (zh) 面向概率型位置流数据的室内区域访问量连续监控方法
Konstantinidis et al. Radio map prefetching for indoor navigation in intermittently connected wi-fi networks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15902125

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15902125

Country of ref document: EP

Kind code of ref document: A1