WO2020199662A1 - 用于推送信息的方法和装置 - Google Patents

用于推送信息的方法和装置 Download PDF

Info

Publication number
WO2020199662A1
WO2020199662A1 PCT/CN2019/126769 CN2019126769W WO2020199662A1 WO 2020199662 A1 WO2020199662 A1 WO 2020199662A1 CN 2019126769 W CN2019126769 W CN 2019126769W WO 2020199662 A1 WO2020199662 A1 WO 2020199662A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
location
candidate push
matching
push information
Prior art date
Application number
PCT/CN2019/126769
Other languages
English (en)
French (fr)
Inventor
李磊
Original Assignee
北京字节跳动网络技术有限公司
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 北京字节跳动网络技术有限公司 filed Critical 北京字节跳动网络技术有限公司
Publication of WO2020199662A1 publication Critical patent/WO2020199662A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/50—Network services
    • H04L67/52—Network services specially adapted for the location of the user terminal
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/50—Network services
    • H04L67/55—Push-based network services

Definitions

  • the embodiments of the present disclosure relate to the field of computer technology, in particular to methods and devices for pushing information.
  • the information that can be pushed to users includes some information related to regions. For example, information used to describe events that occurred in a certain city, used to introduce traffic conditions in a certain street, and so on.
  • some information push applications usually push some information related to the user's location to the user based on the user's location.
  • the embodiments of the present disclosure propose methods and devices for pushing information.
  • the embodiments of the present disclosure provide a method for pushing information.
  • the method includes: acquiring first location information indicating the location of a target user, and acquiring a set of candidate push information; For the candidate push information, perform the following determining step: determine the first matching information of the candidate push information, where the first matching information is used to characterize the target user’s preference for the candidate push information based on the first location information; The matching information determines the first operation information corresponding to the candidate push information, where the first operation information is used to characterize the probability of the target user performing the target operation on the candidate push information based on the first matching information; according to the corresponding first operation information , Select candidate push information from the candidate push information set; push the selected candidate push information to the terminal device corresponding to the target user.
  • the above method further includes: determining at least one piece of second location information according to the first location information; and the determining step further includes: determining the candidate push information for the second location information in the at least one piece of second location information According to the second matching information, the second matching information is used to represent the degree of preference of the target user for the candidate push information based on the second location information; the second operation information corresponding to the candidate push information is determined according to the second matching information , Where the second operation information is used to characterize the probability that the target user performs the target operation on the candidate push information based on the second matching information; and according to the corresponding first operation information, the candidate push information is selected from the candidate push information set, including: According to the corresponding first operation information and at least one second operation information, candidate push information is selected from the candidate push information set.
  • determining at least one piece of second location information according to the first location information includes: determining at least one piece of second location information according to the first location information and a preset distance threshold, where the distance threshold is used for Characterize the distance between the position indicated by the first position information and the position indicated by the second position information.
  • determining the first operation information corresponding to the candidate push information according to the first matching information includes: inputting the first matching information into a pre-trained information prediction model to obtain the first operation corresponding to the candidate push information Information, wherein the information prediction model is used to characterize the correspondence between the matching information and the operation information; and according to the second matching information, determining the second operation information corresponding to the candidate push information includes: inputting the second matching information into the information The prediction model obtains the second operation information corresponding to the candidate push information.
  • the first matching information includes at least one of the following: information used to characterize the degree of preference of the target user for information related to the location indicated by the first location information, and information used to characterize the candidate push information and the first location Information about the degree of relevance of the location indicated by the information, and information used to characterize the degree of attention of the location indicated by the first location information.
  • the second matching information includes at least one of the following: information used to characterize the target user's preference for information related to the location indicated by the second location information, and information used to characterize the candidate push information and the second location information. 2. Information about the degree of relevance of the position indicated by the position information, and information used to characterize the degree of attention of the position indicated by the second position information.
  • the information prediction model is trained through the following steps: obtaining a training sample set, where the training samples in the training sample set include sample matching information and sample indication information, where the sample matching information includes historical push information corresponding to the target user
  • the sample indication information is used to indicate whether the target user has performed the target operation on the corresponding historical push information
  • obtain the pre-built initial model use the machine learning method, based on the training sample set and the preset loss function, to The model is trained, and the initial model after training is determined as the information prediction model.
  • selecting candidate push information from the candidate push information set according to the corresponding first operation information and at least one second operation information includes: for the candidate push information in the candidate push information set, in response to determining the candidate push information If the corresponding target probability is greater than the preset probability threshold, the candidate push information is selected, where the target probability is used to indicate the maximum value in the probability set corresponding to the candidate push information or the average value of each probability in the probability set, where the probability set is
  • the candidate push information corresponding to the probability represented by the first operation information and the probability represented by the corresponding at least one second operation information are composed of respectively.
  • acquiring the candidate push information set includes: acquiring the candidate push information set according to the first location information, wherein the candidate push information in the candidate push information set is related to the location indicated by the first location information.
  • pushing the selected candidate push information to the terminal device corresponding to the target user includes: in response to detecting a preset operation of the target user, pushing the selected candidate push information to the terminal device, and controlling the terminal device The selected candidate push information is displayed on the screen, where the preset operation is used for requesting push of information related to the location indicated by the first location information.
  • an embodiment of the present disclosure provides a device for pushing information, the device comprising: an acquiring unit configured to acquire first location information used to indicate the location of a target user, and acquiring a set of candidate push information
  • the operation information determining unit is configured to perform the following determining step for the candidate push information in the candidate push information set: determine the first matching information of the candidate push information, wherein the first matching information is used to characterize the information based on the first location, The target user’s preference for the candidate push information; according to the first matching information, determine the first operation information corresponding to the candidate push information, where the first operation information is used to indicate that the target user pushes the candidate based on the first matching information The probability that the information performs the target operation; the selection unit is configured to select candidate push information from the candidate push information set according to the corresponding first operation information; the push unit is configured to push the selected candidate push information to the terminal device corresponding to the target user information.
  • the foregoing apparatus for pushing information further includes: a location information determining unit configured to determine at least one piece of second location information according to the first location information; and the foregoing operation information determining unit is further configured to The second location information in at least one piece of second location information determines the second matching information of the candidate push information, where the second matching information is used to characterize the target user’s preference for the candidate push information based on the second location information According to the second matching information, determine the second operation information corresponding to the candidate push information, where the second operation information is used to characterize the probability of the target user performing the target operation on the candidate push information based on the second matching information; and the above selection The unit is further configured to select candidate push information from the candidate push information set according to the corresponding first operation information and at least one second operation information.
  • the above-mentioned location information determining unit is further configured to determine at least one piece of second location information according to the first location information, including: determining at least one piece of second location information according to the first location information and a preset distance threshold. Location information, where the distance threshold is used to characterize the distance between the location indicated by the first location information and the location indicated by the second location information.
  • the operation information determining unit is further configured to input the first matching information into a pre-trained information prediction model to obtain the first operation information corresponding to the candidate push information, wherein the information prediction model is used to characterize the matching The corresponding relationship between the information and the operation information; and the above-mentioned operation information determining unit is further configured to input the second matching information into the information prediction model to obtain the second operation information corresponding to the candidate push information.
  • the first matching information includes at least one of the following: information used to characterize the degree of preference of the target user for information related to the location indicated by the first location information, and information used to characterize the candidate push information and the first location Information about the degree of relevance of the location indicated by the information, and information used to characterize the degree of attention of the location indicated by the first location information.
  • the second matching information includes at least one of the following: information used to characterize the target user's preference for information related to the location indicated by the second location information, and information used to characterize the candidate push information and the second location information. 2. Information about the degree of relevance of the position indicated by the position information, and information used to characterize the degree of attention of the position indicated by the second position information.
  • the information prediction model is trained through the following steps: obtaining a training sample set, where the training samples in the training sample set include sample matching information and sample indication information, where the sample matching information includes historical push information corresponding to the target user
  • the sample indication information is used to indicate whether the target user has performed the target operation on the corresponding historical push information
  • obtain the pre-built initial model use the machine learning method, based on the training sample set and the preset loss function, to The model is trained, and the initial model after training is determined as the information prediction model.
  • the aforementioned selection unit is further configured to select the candidate push information in response to determining that the target probability corresponding to the candidate push information is greater than a preset probability threshold for the candidate push information in the candidate push information set, where the target The probability is used to indicate the maximum value in the probability set corresponding to the candidate push information or the average value of each probability in the probability set, where the probability set is represented by the probability of the first operation information corresponding to the candidate push information and the corresponding at least one second The probabilistic composition represented by the operating information.
  • the aforementioned acquiring unit is further configured to acquire a candidate push information set according to the first location information, wherein the candidate push information in the candidate push information set is related to the location indicated by the first location information.
  • the aforementioned pushing unit is further configured to, in response to detecting a preset operation of the target user, push the selected candidate push information to the terminal device, and control to display the selected candidate push information on the terminal device, wherein ,
  • the preset operation is used for requesting to push information related to the location indicated by the first location information.
  • the embodiments of the present disclosure provide a server, which includes: one or more processors; a storage device for storing one or more programs; when one or more programs are processed by one or more The processor executes, so that one or more processors implement the method described in any implementation manner of the first aspect.
  • an embodiment of the present disclosure provides a computer-readable medium on which a computer program is stored, and when the computer program is executed by a processor, the method as described in any implementation manner in the first aspect is implemented.
  • the method and apparatus for pushing information provided by the embodiments of the present disclosure predict the probability of the user performing a target operation on the candidate push information based on the user's preference for the candidate push information at his location.
  • the predicted probability can reflect the user's degree of interest in the information related to their location, and further, can selectively push information to the user according to the predicted probability, in order to as far as possible to the information related to their location Users push corresponding information, and avoid pushing corresponding information to users who are not interested in information related to their location, thereby reducing the unnecessary push of corresponding information to users who are not interested in information related to their location during the information push process Resource consumption.
  • FIG. 1 is an exemplary system architecture diagram in which an embodiment of the present disclosure can be applied
  • Fig. 2 is a flowchart of an embodiment of a method for pushing information according to the present disclosure
  • FIG. 3 is a flowchart of another embodiment of a method for pushing information according to the present disclosure.
  • Fig. 4 is a schematic diagram of an application scenario of a method for pushing information according to an embodiment of the present disclosure
  • Fig. 5 is a schematic structural diagram of an embodiment of an apparatus for pushing information according to the present disclosure
  • Fig. 6 is a schematic structural diagram of an electronic device suitable for implementing embodiments of the present disclosure.
  • FIG. 1 shows an exemplary architecture 100 to which an embodiment of the method for pushing information or the apparatus for pushing information of the present disclosure can be applied.
  • the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105.
  • the network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105.
  • the network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables.
  • the terminal devices 101, 102, 103 interact with the server 105 through the network 104 to receive or send messages and so on.
  • Various client applications may be installed on the terminal devices 101, 102, 103. For example, browser applications, search applications, instant messaging tools, reading applications, information sharing applications, etc.
  • the terminal devices 101, 102, and 103 may be hardware or software.
  • the terminal devices 101, 102, and 103 may be various electronic devices with display screens, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and so on.
  • the terminal devices 101, 102, and 103 are software, they can be installed in the electronic devices listed above. It can be implemented as multiple software or software modules (for example, multiple software or software modules for providing distributed services), or as a single software or software module. There is no specific limitation here.
  • the server 105 may be a server that provides various services, for example, a back-end server that provides support for some applications (such as information sharing applications) installed on the terminal devices 101, 102, and 103.
  • the server 105 may predict the user's operation information on each candidate push information based on the location of the user corresponding to the terminal device 101, 102, 103 and the candidate push information set.
  • the candidate push information may be selected from the candidate push information set according to the corresponding operation information, and the selected candidate push information may be pushed to the terminal devices 101, 102, 103.
  • the method for pushing information provided by the embodiments of the present disclosure is generally executed by the server 105, and accordingly, the device for pushing information is generally set in the server 105.
  • the terminal devices 101, 102, and 103 may also be installed with information processing applications, and the terminal devices 101, 102, and 103 may also process the candidate push information in the candidate push information set based on the information processing application.
  • the method for pushing information can also be executed by the terminal devices 101, 102, 103, and accordingly, the device for pushing information can also be provided in the terminal devices 101, 102, 103.
  • the exemplary system architecture 100 may not include the server 105 and the network 104.
  • the server 105 may be hardware or software.
  • the server 105 can be implemented as a distributed server cluster composed of multiple servers, or as a single server.
  • the server 105 is software, it can be implemented as multiple software or software modules (for example, multiple software or software modules for providing distributed services), or as a single software or software module. There is no specific limitation here.
  • terminal devices, networks, and servers in FIG. 1 are merely illustrative. According to implementation needs, there can be any number of terminal devices, networks and servers.
  • FIG. 2 shows a flow 200 of an embodiment of the method for pushing information according to the present disclosure.
  • the method for pushing information includes the following steps:
  • Step 201 Obtain first location information indicating the location of the target user, and obtain a candidate push information set.
  • the target user may refer to terminal devices (terminal devices 101, 102, 103 shown in FIG. 1) that are in communication with the execution subject of the method for pushing information (the server 105 shown in FIG. 1).
  • the first location information may be various information capable of indicating a location.
  • the first location information may be latitude and longitude information, communication address information, and so on.
  • the above-mentioned execution subject may determine the location of the target user based on various existing positioning technologies.
  • the location of the target user can be determined based on GPS (Global Positioning System, Global Positioning System) technology.
  • GPS Global Positioning System, Global Positioning System
  • the determined location of the target user may be directly used as the first location information, or preset information indicating the determined location of the target user may be obtained from a local or other storage device as the first location information.
  • the candidate push information set may be a set consisting of some information pre-designated by a technician, or a set of some information determined according to preset screening conditions.
  • the candidate push information in the candidate push information set may be various types of information.
  • the candidate push information may be images, text, video, audio, and so on.
  • the above-mentioned execution subject may obtain the candidate push information set from a local or other storage device (such as a connected database, a third-party data platform, etc.).
  • the candidate push information set may be obtained according to the first location information.
  • the candidate push information in the candidate push information set is related to the position indicated by the first position information.
  • Whether the candidate push information is related to the location indicated by the first location information may be determined according to a preset related condition.
  • the related condition may be that the content described by the candidate push information is related to the location indicated by the first location information.
  • the candidate push information is an article describing a certain kind of food. If the food described is a specialty food in a certain area where the location indicated by the first location information is located, then it can be considered that the candidate push information and the first location information indicate Location related.
  • the relevant condition may be that a preset keyword in a preset keyword set corresponding to a location indicated by the first location information appears in the candidate push information.
  • the preset keyword set may include various description methods of the location indicated by the first location information.
  • the preset keyword set may also include various description methods corresponding to all locations covered in a certain area where the location indicated by the first location information is located.
  • the location of the target user may be a very specific location.
  • the location of the target user can be accurate to a specific latitude and longitude. In this case, there may be less information related to the location of the target user. Therefore, according to actual application requirements, the location of the target user can be generalized to a certain area where the location of the target user is located. At this time, all information related to a certain area where the location of the target user is located can be regarded as information related to the location of the target user.
  • the above-mentioned execution subject may obtain some information related to the location indicated by the first location information from a local or other storage device (such as a connected database, a third-party data platform, etc.) according to the first location information to form a candidate push information set.
  • a local or other storage device such as a connected database, a third-party data platform, etc.
  • Step 202 Perform the following determining steps for the candidate push information in the candidate push information set:
  • Step 2021 Determine the first matching information of the candidate push information.
  • the first matching information may be used to characterize the degree of preference of the target user for the candidate push information based on the first location information.
  • the degree of preference can be measured according to various related information of the user and the candidate push information.
  • one or more pieces of information that can characterize the target user's preference for the candidate push information based on the first location information can be flexibly selected as the first matching information.
  • the first matching information may include the similarity between some historical push information that the user has a higher preference within a preset time period and the candidate push information.
  • some historical push information that the user has a higher preference within the preset time period may be determined according to the historical behavior data of the user at the location indicated by the first location information within the preset time period.
  • the historical push information that the user has clicked on or commented on can be selected from the historical push information pushed to the user when the user is at the position indicated by the first location information within a preset time period as the user has a higher preference.
  • Historical push information can be selected from the historical push information pushed to the user when the user is at the position indicated by the first location information within a preset time period as the user has a higher preference.
  • the first matching information may include at least one of the following: information used to characterize the degree of preference of the target user for information related to the location indicated by the first location information, and information used to characterize the candidate push information and the first location information Information about the degree of relevance of the indicated position, and information used to characterize the degree of attention of the position indicated by the first position information.
  • the degree of preference of the target user for the information related to the location indicated by the first location information can be determined in various ways according to actual application scenarios.
  • the target user's preference for information related to the location indicated by the first location information may be analyzed based on historical behavior data of the target user for information related to the location indicated by the first location information.
  • the user will have some interactive operations on the preferred push information. For example, click operation, comment operation, bookmark or share the connection address of push information, etc.
  • Some attribute information of the interactive operations performed by the user for different push information will also be different. For example, the number of clicks, the length of browsing, etc. Therefore, the user's preference for the pushed information can be analyzed based on the user's historical behavior data. Under normal circumstances, the more or more interactive the user performs, the higher the user's preference for the corresponding push information.
  • the total number of information related to the location indicated by the first location information that has been pushed to the target user during the target historical time period can be counted as the first number, and among the pushed information, the target user has executed
  • the number of information of the preset operation is used as the second number. Thereafter, the quotient of the second number and the first number can be used to characterize the degree of preference of the target user for the information related to the location indicated by the first location information.
  • the target historical time period can be any time period pre-designated by a technician.
  • the target time period may be a time period corresponding to the current time as a starting point and three months before the current time.
  • the preset operations may be various user operations designated in advance by the technician.
  • some existing open source predictive models can be used to predict the target user's operation information (such as click-through rate) of information related to the location indicated by the first location information. Based on this, the predicted operation information can be used to characterize the degree of preference of the target user for the information related to the location indicated by the first location information.
  • the model used for prediction can also be trained by a technician using a large amount of user historical behavior data in advance.
  • the information prediction model can predict the user's operation information (such as click-through rate, etc.) of the candidate push information based on the feature vector used to characterize the user and the feature vector used to characterize the candidate push information.
  • the degree of correlation between the candidate push information and the position indicated by the first position information may refer to the degree of association between the candidate push information and the position indicated by the first position information.
  • the candidate push information "A” is an article describing a scenic spot in the area where the location indicated by the first location information is located.
  • the candidate push information "B” is an article in which the name indicating the location indicated by the first location information appears, but the content has nothing to do with the location indicated by the first location information. Then, it can generally be considered that the correlation degree between the candidate push information "A” and the position indicated by the first position information is higher than the correlation degree between the candidate push information "B" and the position indicated by the first position information.
  • the method for determining the correlation degree can be selected according to different application scenarios.
  • the comment information set corresponding to the candidate push information may be obtained first.
  • the comment information set may include some comment information of the candidate push information.
  • the location information of the user corresponding to each piece of comment information in the comment information set can be obtained to obtain the location information set.
  • the user corresponding to the comment information may refer to the user corresponding to the terminal device that publishes the comment information. After that, a certain area where the location indicated by the first location information is located may be determined as the target area. Then, it is possible to count the number of location information in the location information set whose indicated location belongs to the target area as the first number, and determine the total number of location information included in the location information set as the second number.
  • the quotient of the first number and the second number may be used to characterize the correlation between the candidate push information and the location indicated by the first location information.
  • the larger the quotient of the first number and the second number the higher the correlation between the candidate push information and the location indicated by the first location information.
  • users who are interested in information related to the location indicated by the first location information are usually users who have been in a certain area of the location indicated by the first location information for a long time. Then, for information related to the location indicated by the first location information, users who comment on the information are also users who have been in a certain area of the location indicated by the first location information for a long time. Therefore, the correlation between the candidate push information and the location indicated by the first position information can be analyzed according to the location information of the user corresponding to each piece of comment information of the candidate push information.
  • various existing keyword extraction methods may be used to extract the keywords of the candidate push information to obtain a keyword set corresponding to the candidate push information. Then, a preset keyword set corresponding to the location indicated by the first location information can be acquired. After that, the similarity between the keyword set corresponding to the candidate push information and the keyword set corresponding to the location indicated by the first location information may be used to characterize the degree of relevance between the candidate push information and the location indicated by the first location information. Generally, the higher the similarity between the keyword set corresponding to the candidate push information and the keyword set corresponding to the location indicated by the first location information, the higher the correlation between the candidate push information and the location indicated by the first location information.
  • the algorithm for determining the similarity of two keyword sets can use various existing related algorithms. For example, algorithms based on TF-IDF (Term Frequency-Inverse Document Frequency, word frequency-inverse text frequency index), algorithms based on SimRank (a model that measures the similarity of any two objects), and algorithms based on WMD (Word Mover's Distance, one A method to measure the distance between two objects) algorithms.
  • TF-IDF Term Frequency-Inverse Document Frequency, word frequency-inverse text frequency index
  • SimRank a model that measures the similarity of any two objects
  • WMD Wide Mover's Distance, one A method to measure the distance between two objects
  • a pre-trained correlation determination model can be used to analyze the correlation between the candidate push information and the location indicated by the first location information.
  • the correlation determination model may be used to determine the correlation between the candidate push information and the first location information according to the candidate push information and the first location information.
  • the correlation determination model can be trained based on a large number of training samples. Each training sample may include historical push information, location information, and the correlation between historical push information and location information. Wherein, the correlation between the historical push information and the location information in the training sample can be determined using some of the aforementioned correlation determination methods, or can be pre-marked by a technician.
  • the degree of attention of the location indicated by the first location information may be used to characterize the degree of attention of the location indicated by the first location information.
  • the attention degree of the location of the scenic spot is usually higher than the attention degree of the non-scenic spot.
  • the attention of each location is relatively stable. Therefore, the attention degree corresponding to each position can be set in advance.
  • the attention of certain locations needs to be dynamically adjusted. For example, a relatively well-known activity has recently been carried out in the area where a certain location is located. In this case, the attention of this location can be appropriately increased.
  • the information used to characterize the degree of preference of the target user for the information related to the location indicated by the location information may be information in various forms.
  • the click-through rate when used to characterize the target user's preference for information related to a location indicated by location information, the click-through rate may be information that characterizes the target user's preference for information related to a location indicated by the location information.
  • the display information corresponding to different click rates can also be preset, and then the display information corresponding to the click rates can be obtained as information that characterizes the target user's preference for information related to the location indicated by the location information.
  • the information used to characterize the correlation between the candidate push information and the location indicated by the first location information may also be information in various forms.
  • the information used to characterize the degree of attention of the location indicated by the first location information may also be information in various forms.
  • the specific representation method can be flexibly set by the technicians according to actual application requirements.
  • Step 2022 Determine first operation information corresponding to the candidate push information according to the first matching information.
  • the first operation information may be used to characterize the probability of the target user performing the target operation on the candidate push information based on the first matching information.
  • the target operation can be a pre-designated user operation (such as a user operation used to indicate a click operation, a user operation used to indicate a comment operation, etc.), or a user operation determined based on the candidate push information (for example, the target operation can be Any user operation supported by candidate push information, etc.).
  • different methods for determining the first operation information corresponding to the candidate push information can be selected.
  • the maximum value indicated by each information or the average value of the values indicated by each information included in the first matching information may be selected as the first operation information.
  • the first matching information may be input to a pre-trained information prediction model to obtain the first operation information corresponding to the candidate push information.
  • the information prediction model can be used to characterize the correspondence between matching information and operation information.
  • the information prediction model can be obtained through training in the following steps:
  • Step 1 Obtain a training sample set.
  • the training samples in the training sample set may include sample matching information and sample indication information.
  • the sample matching information may include matching information of historical push information corresponding to the target user.
  • the sample indication information may be used to indicate whether the target user has performed a target operation on the corresponding historical push information.
  • the sample matching information may include at least one of the following according to different application requirements: information used to characterize the degree of preference of the target user for information related to the location of the target user, and information used to characterize historical push information and target user Information about the degree of relevance of the corresponding location, and information used to characterize the degree of attention of the location corresponding to the target user.
  • the location corresponding to the target user may point to the location of the target user determined when the user pushes historical push information.
  • historical push information may refer to information related to a location corresponding to the target user.
  • Step 2 Obtain the pre-established initial model.
  • the initial model can be various artificial neural networks that have not been trained or completed, or a model obtained by combining various artificial neural networks.
  • the initial model can also be built by technical personnel according to actual application requirements (such as which network layers are needed, the parameters of each network layer, the size of the convolution kernel, etc.) using some deep learning frameworks (such as TensorFlow, Caffe, etc.).
  • an initial model can be built based on some existing regression models (such as a model based on logistic regression, a model based on stepwise regression, etc.).
  • Step three is to use machine learning methods to train the initial model based on the training sample set and preset loss function, and determine the trained initial model as an information prediction model.
  • the loss function can be pre-designed by technicians.
  • the loss function is usually used according to the difference between the output of the initial model and the corresponding expected output. Generally, it is desirable that the value of the loss function is as small as possible.
  • the choice of loss function can also be determined according to the initial model. Different initial models can choose different loss functions. For example, for the initial model based on logistic regression, an existing loss function derived from maximum likelihood estimation can be selected.
  • At least one training sample can be selected from the training sample set each time. Then, the sample matching information in the selected training samples is input to the initial model, and the output operation information corresponding to the selected training samples are obtained. Afterwards, it can be determined according to the preset probability threshold that the output operation information represents that the target user will perform the target operation on the historical push information corresponding to the input sample matching information or the target user will not perform the historical push information corresponding to the input sample matching information Target operation.
  • the probability threshold is 0.5. Then, if the probability of the output operation information characterization is greater than 0.5, it can be considered that the target user will perform the target operation on the historical push information corresponding to the input sample matching information. Correspondingly, if the probability of the output operation information characterization is less than 0.5, it can be considered that the target user will not perform the target operation on the historical push information corresponding to the input sample matching information.
  • the value of the preset loss function can be determined according to the output operation information and the sample indication information. After that, it is determined whether the training of the initial model is completed according to the value of the loss function. If it is determined that the initial model training is completed according to the value of the loss function, the completed initial model can be determined as the information prediction model. Wherein, it can be preset to determine that the initial model training is completed when the value of the loss function meets the preset condition.
  • the parameters of the initial model can be adjusted based on the value of the loss function by using back propagation and gradient descent algorithms.
  • at least one training sample can be re-selected from the training sample set, and the adjusted initial model can be determined as the initial model to continue training.
  • the training of the initial model requires repeated iterations to complete the training.
  • the parameters of the initial model can be adjusted based on the average value of the loss function value corresponding to each training sample selected each time.
  • the information prediction model can be obtained by training through the following steps: obtaining a training sample set.
  • the training samples in the training sample set may include sample matching information and sample operation information.
  • the sample matching information may include matching information of historical push information corresponding to the target user.
  • the sample operation information may be used to indicate the probability of the target user performing the target operation on the corresponding historical push information.
  • the sample operation information can be pre-marked by the technicians.
  • a curve fitting method can be used to perform curve fitting based on the training sample set, and the obtained fitting curve can be used as an information prediction model.
  • various existing methods such as a fitting algorithm based on the least square method, etc.
  • applications that can be used for curve fitting can be used to perform curve fitting based on a training sample set.
  • Step 203 Select candidate push information from the candidate push information set according to the corresponding first operation information.
  • the first operation information can be used to characterize the probability of the target user performing the target operation on the candidate push information based on the first matching information. Therefore, the number of candidate push information targets can be selected from the candidate push information set in descending order of the corresponding first operation information.
  • the target number can be preset by a technician, or can be determined according to a set determination condition (for example, ten percent of the total number of candidate push information included in the candidate push information set is determined as the target number, etc.).
  • the candidate push information whose corresponding probability of the first operation information is greater than the probability threshold may be selected from the candidate push information set to obtain the candidate push subset.
  • a preset number of candidate push messages can be randomly selected from the candidate push subset.
  • Step 204 Push the selected candidate push information to the terminal device corresponding to the target user.
  • the preset operation in response to detecting the preset operation of the target user, push the selected candidate push information to the terminal device, and control the display of the selected candidate push information on the terminal device.
  • the preset operation may be used to request to push information related to the location indicated by the first location information.
  • the preset operation may include any user operation preset by a technician.
  • the preset operation may include a refresh operation, an operation of clicking a preset control on a page displayed by the terminal device, and so on.
  • candidate push information can be displayed on the terminal device in a unified manner, so that the user can quickly browse related information.
  • two buttons can be set on the user browsing information page to switch between different display pages.
  • a button can be used to display information related to the location corresponding to the user.
  • the other button can be used to display other push information.
  • the method provided by the above-mentioned embodiments of the present disclosure predicts the probability that the user performs a target operation on the candidate push information based on the user's preference for the candidate push information at the location where the user is located. Furthermore, according to the predicted probability, the push information can be selected to push to the user, which helps to improve the accuracy of the push information, avoid pushing relevant information to users who are not interested in the information related to the corresponding location, and reduce the information push process Resource consumption. On this basis, it can also push information related to the user's corresponding location to the user according to the user's push request and control the display on the user's corresponding terminal device, so that the user wants to browse the location-related information corresponding to the user When information, you can view relevant information in time.
  • FIG. 3 shows a flow 300 of another embodiment of a method for pushing information.
  • the process 300 of the method for pushing information includes the following steps:
  • Step 301 Obtain first location information indicating the location of the target user, and obtain a candidate push information set.
  • step 201 For the specific execution process of this step, reference may be made to the related description of step 201 in the embodiment corresponding to FIG. 2, which will not be repeated here.
  • Step 302 Determine at least one piece of second location information according to the first location information.
  • a correspondence relationship table storing a correspondence relationship between the first location information and at least one second location information may be preset. At this time, after determining the first location information, at least one piece of second location information corresponding to the first location information can be searched in the correspondence table.
  • At least one piece of second location information may be determined according to the first location information and at least one preset distance threshold.
  • the distance threshold may be used to characterize the distance between the position indicated by the first position information and the position indicated by the second position information.
  • the distance threshold can be preset by a technician.
  • the distance threshold may be ten kilometers, then a position within ten kilometers from the position indicated by the first position information and containing the position indicated by the first position information may be determined as a second position, so that the second position information may be obtained.
  • the location information has a certain hierarchical relationship. For example, XX Street, XX District, XX City, XX province. Among them, XX province can be a piece of location information, and XX city can also be a piece of location information. Similarly, XX district and XX street are also location information. That is, location information can have different granularities. Therefore, in practical applications, how to choose the representation granularity of location information is also an aspect that needs to be considered.
  • the location of the acquiring user is in the XX area, but the user may not be interested in the information related to the XX area, but is more interested in the information related to the XX city, or may also be more interested in the information related to the XX street.
  • a push message can be information related to XX district, and it can also be information related to XX street.
  • the decibels are named the first position, the second position, etc., and those skilled in the art should understand that the first or the second does not constitute a special position. limited.
  • Step 303 For the candidate push information in the candidate push information set, perform the following determining steps:
  • Step 3031 Determine the first matching information of the candidate push information.
  • Step 3032 Determine the first operation information corresponding to the candidate push information according to the first matching information.
  • Step 3033 For the second location information in the at least one second location information, perform the following steps:
  • Step 30331 Determine second matching information of the candidate push information.
  • the second matching information may be used to characterize the degree of preference of the target user for the candidate push information based on the second location information.
  • the degree of preference can be measured according to various related information of the user and the candidate push information.
  • the second Matching information Similar to the first matching information, different ones can be flexibly selected according to actual application scenarios and application requirements, and one or more information that can characterize the target user’s preference for the candidate push information based on the second location information is used as the second Matching information.
  • the second matching information may include the similarity between some historical push information that the user has a higher preference within a preset time period and the candidate push information.
  • some historical push information that the user has a higher preference during the preset time period may be determined according to the historical behavior data of the user at the location indicated by the second location information during the preset time period.
  • the historical push information that the user has clicked on or commented on can be selected from the historical push information pushed to the user when the user is at the position indicated by the second position information within a preset time period as the user has a higher preference.
  • Historical push information can be selected from the historical push information pushed to the user when the user is at the position indicated by the second position information within a preset time period as the user has a higher preference.
  • the second matching information includes at least one of the following: information used to characterize the target user’s preference for information related to the location indicated by the second location information, and information used to characterize the candidate push information and the second location Information about the degree of relevance of the position indicated by the information, and information used to characterize the degree of attention of the position indicated by the second position information.
  • Step 30332 Determine the second operation information corresponding to the candidate push information according to the second matching information.
  • the second operation information may be used to characterize the probability of the target user performing the target operation on the candidate push information based on the second matching information.
  • the second matching information may be input to the information prediction model to obtain the second operation information corresponding to the candidate push information.
  • step 2022 For the specific process of determining the second operation information and the information prediction model, reference may be made to related descriptions in step 2022 in the embodiment corresponding to FIG. 2, which will not be repeated here.
  • Step 304 Select candidate push information from the candidate push information set according to the corresponding first operation information and at least one second operation information.
  • the candidate push information in response to determining that the target probability corresponding to the candidate push information is greater than a preset probability threshold, the candidate push information may be selected.
  • the target probability may be used to represent the maximum value in the probability set corresponding to the candidate push information or the average value of each probability in the probability set.
  • the probability set may be composed of the probability represented by the first operation information corresponding to the candidate push information and the probability represented by the corresponding at least one second operation information respectively.
  • the probabilities respectively characterized by the first operation information and the at least one second operation information can be used as an aspect to consider whether to select candidate push information.
  • the probability represented by the first operation information and the at least one second operation information can be taken as one aspect, and other aspects are combined to comprehensively consider whether to select each candidate push information.
  • the format of the candidate push information needs to be considered, such as whether the candidate push information is video or text, or when the candidate push information is text, whether the candidate push information contains many images, etc. Therefore, it is possible to select appropriate candidate push information in combination with the network status of the target user, so as to avoid receiving video or text information containing more graphics and text when the user is using the data network.
  • Step 305 Push the selected candidate push information to the terminal device corresponding to the target user.
  • step 204 For the specific execution process of this step, reference may be made to the related description of step 204 in the embodiment corresponding to FIG. 2, which will not be repeated here.
  • FIG. 4 is a schematic diagram 400 of an application scenario of the method for pushing information according to this embodiment.
  • the above-mentioned execution subject may first obtain the location information 402 of the user's location based on GPS technology as "Wudaokou, Haidian District, Beijing".
  • the location information 402 can then be generalized to two location information 401 and 403 with different granularities. As shown in the figure, the location information 401 is "Beijing Haidian District", and the location information 403 is "Beijing”.
  • the above-mentioned execution subject may also obtain the information set 405 from the connected database 404. Then, for a piece of information in the information set 405, the click rate of the user based on the location information of three different granularities can be determined respectively.
  • the following takes a piece of information 4051 in the information set 405 as an example for specific description.
  • the user's click-through rate of historically pushed information related to "Beijing Haidian District” can be determined to represent the user's preference for information related to "Beijing Haidian District” .
  • the content of the information 4051 can be analyzed to determine the degree of relevance between the information 4051 and "Haidian District, Beijing".
  • the attention degree corresponding to the preset "Beijing Haidian District” can be obtained.
  • the matching information 406 can be composed of the click rate, relevance, and attention corresponding to the determined "Haidian District, Beijing”.
  • the matching information 406 can be input into the pre-trained click-through rate prediction model 407, and it is obtained that under the location information of "Beijing Haidian District", the predicted probability 408 that the user clicks on the information 4051 is 0.9.
  • the user's click-through rate of historically pushed information related to "Wudaokou, Haidian District, Beijing” can be determined to characterize the user's information related to "Wudaokou, Haidian District, Beijing” Degree of preference.
  • the content of the information 4051 can be analyzed to determine the degree of relevance between the information 4051 and "Wudaokou, Haidian District, Beijing”.
  • the attention degree corresponding to the preset "Wudaokou, Haidian District, Beijing” can be obtained.
  • the matching information 409 can be composed of the click rate, relevance, and attention corresponding to the determined "Wudaokou, Haidian District, Beijing".
  • the matching information 409 can be input to the pre-trained click-through rate prediction model 407, and it is obtained that the predicted probability 410 that the user clicks on the information 4051 is 0.8 under the location information "Wudaokou, Haidian District, Beijing".
  • the user's click-through rate of historically pushed information related to "Beijing” can be determined to characterize the user's preference for information related to "Beijing".
  • the content of the information 4051 can be analyzed to determine the degree of relevance between the information 4051 and "Beijing”.
  • the attention degree corresponding to the preset "Beijing” can be obtained.
  • the matching information 411 may be composed of the click rate, relevance, and attention corresponding to the determined "Beijing City”.
  • the matching information 411 can be input into the pre-trained click-through rate prediction model 407, and it is obtained that under the location information "Beijing", the predicted probability 412 of the user clicking the information 4051 is 0.2.
  • the maximum value of 0.9 can be selected from the respective upper prediction probabilities corresponding to the information 4051. Since 0.9 is greater than the preset threshold 0.6, the information 4051 can be pushed to the terminal device 413 corresponding to the user, and the information 4051 can be displayed on the terminal device 413.
  • the method provided by the above-mentioned embodiments of the present disclosure generalizes the user's position to multiple positions with different granularities after obtaining the user's position, thereby predicting the user's operation information on candidate push information for each granular position.
  • whether to push candidate push information can be comprehensively considered, which helps to further improve the accuracy of push information.
  • the present disclosure provides an embodiment of a device for pushing information.
  • the device embodiment corresponds to the method embodiment shown in FIG. It can be used in various electronic devices.
  • the apparatus 500 for pushing information includes an acquiring unit 501, an operation information determining unit 502, a selecting unit 503, and a pushing unit 504.
  • the obtaining unit 501 is configured to obtain the first location information used to indicate the location of the target user and obtain the candidate push information set
  • the operation information determining unit 502 is configured to perform the following determination for the candidate push information in the candidate push information set Step: Determine the first matching information of the candidate push information, where the first matching information is used to characterize the target user's preference for the candidate push information based on the first location information; determine the candidate push information according to the first matching information Corresponding first operation information, where the first operation information is used to characterize the probability that the target user performs the target operation on the candidate push information based on the first matching information;
  • the selection unit 503 is configured to, according to the corresponding first operation information, The candidate push information is selected collectively;
  • the pushing unit 504 is configured to push the selected candidate push information to the terminal device corresponding to the target user.
  • the specific processing of the acquiring unit 501, the operating information determining unit 502, the selecting unit 503, and the pushing unit 504 and the technical effects brought by them can be implemented with reference to FIG. 2 respectively.
  • the related descriptions of step 201, step 202, step 203, and step 204 in the example will not be repeated here.
  • the foregoing apparatus 500 for pushing information further includes: a location information determining unit (not shown in the figure) is configured to determine at least one second location based on the first location information Information; and the above-mentioned operation information determining unit 502, further configured to determine second matching information of the candidate push information for the second position information in the at least one second position information, wherein the second matching information is used to characterize the second position information based on the The second location information, the target user’s preference for the candidate push information; the second operation information corresponding to the candidate push information is determined according to the second matching information, where the second operation information is used to indicate that the target is based on the second matching information The probability that the user performs a target operation on the candidate push information; and the aforementioned selection unit 503 is further configured to select candidate push information from the candidate push information set according to the corresponding first operation information and at least one second operation information.
  • the above-mentioned location information determining unit is further configured to determine at least one piece of second location information according to the first location information, including: according to the first location information and the preset at least one distance The threshold is used to determine at least one piece of second location information, where the distance threshold is used to characterize the distance between the location indicated by the first location information and the location indicated by the second location information.
  • the above-mentioned operation information determination unit 502 is further configured to input the first matching information into a pre-trained information prediction model to obtain the first operation information corresponding to the candidate push information, where , The information prediction model is used to characterize the correspondence between the matching information and the operation information; and the above-mentioned operation information determination unit 502 is further configured to: input the second matching information into the information prediction model to obtain the second corresponding to the candidate push information Operational information.
  • the first matching information includes at least one of the following: information used to characterize the degree of preference of the target user for information related to the location indicated by the first position information, and information used to characterize the Information about the degree of relevance between the candidate push information and the location indicated by the first location information, and information used to characterize the degree of attention of the location indicated by the first location information.
  • the second matching information includes at least one of the following: information used to characterize the target user's preference for information related to the location indicated by the second location information, and information used to characterize the candidate push information and the second location information. 2. Information about the degree of relevance of the position indicated by the position information, and information used to characterize the degree of attention of the position indicated by the second position information.
  • the information prediction model is obtained by training in the following steps: obtaining a training sample set, where the training samples in the training sample set include sample matching information and sample indication information, where the sample matching information includes The matching information of the historical push information corresponding to the target user.
  • the sample indication information is used to indicate whether the target user has performed the target operation on the corresponding historical push information; obtain the pre-established initial model; use the machine learning method based on the training sample set and pre- Set the loss function, train the initial model, and determine the initial model after training as the information prediction model.
  • the selection unit 503 is further configured to select the candidate push information in the candidate push information set, in response to determining that the target probability corresponding to the candidate push information is greater than a preset probability threshold,
  • the candidate push information where the target probability is used to indicate the maximum value in the probability set corresponding to the candidate push information or the average value of each probability in the probability set, wherein the probability set is characterized by the first operation information corresponding to the candidate push information
  • the probability and the corresponding probability composition represented by at least one piece of second operation information respectively.
  • the above-mentioned obtaining unit 501 is further configured to obtain a candidate push information set according to the first position information, wherein the candidate push information in the candidate push information set and the information indicated by the first position information Location related.
  • the above-mentioned pushing unit 503 is further configured to push the selected candidate push information to the terminal device in response to detecting a preset operation of the target user, and to control the display on the terminal device.
  • the selected candidate push information, wherein the preset operation is used to request push of information related to the location indicated by the first location information.
  • the device provided by the above-mentioned embodiment of the present disclosure obtains the first location information used to indicate the location of the target user through the obtaining unit, and obtains the candidate push information set; the operation information determination unit performs the following for the candidate push information in the candidate push information set Determining step: determining the first matching information of the candidate push information, where the first matching information is used to characterize the target user’s preference for the candidate push information based on the first location information; determine the candidate push according to the first matching information The first operation information corresponding to the information, where the first operation information is used to represent the probability that the target user performs the target operation on the candidate push information based on the first matching information; the selection unit pushes the information from the candidate according to the corresponding first operation information Centralized selection of candidate push information; the push unit pushes the selected candidate push information to the terminal device corresponding to the target user, thereby helping to improve the accuracy of the push information and avoid pushing related information to users who are not interested in the information related to the corresponding location Information, reduce resource consumption in the process of
  • FIG. 6 shows a schematic structural diagram of an electronic device (for example, the server in FIG. 1) 600 suitable for implementing embodiments of the present disclosure.
  • the server shown in FIG. 6 is only an example, and should not bring any limitation to the function and scope of use of the embodiments of the present disclosure.
  • the electronic device 600 may include a processing device (such as a central processing unit, a graphics processor, etc.) 601, which can be loaded into a random access device according to a program stored in a read-only memory (ROM) 602 or from a storage device 608.
  • the program in the memory (RAM) 603 executes various appropriate actions and processing.
  • the RAM 603 also stores various programs and data required for the operation of the electronic device 600.
  • the processing device 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604.
  • An input/output (I/O) interface 605 is also connected to the bus 604.
  • the following devices can be connected to the I/O interface 605: including input devices 606 such as touch screen, touch pad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; including, for example, liquid crystal display (LCD), speakers, vibration Output device 607 such as a device; and a communication device 609.
  • the communication device 609 may allow the electronic device 600 to perform wireless or wired communication with other devices to exchange data.
  • FIG. 6 shows an electronic device 600 having various devices, it should be understood that it is not required to implement or have all the illustrated devices. It may alternatively be implemented or provided with more or fewer devices. Each block shown in FIG. 6 can represent one device, or can represent multiple devices as needed.
  • the process described above with reference to the flowchart can be implemented as a computer software program.
  • the embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart.
  • the computer program may be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602.
  • the processing device 601 the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.
  • the computer-readable medium described in the embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two.
  • the computer-readable storage medium may be, for example, but not limited to, an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device.
  • the computer-readable signal medium may include a data signal propagated in a baseband or as a part of a carrier wave, and a computer-readable program code is carried therein. This propagated data signal can take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.
  • the computer-readable signal medium may also be any computer-readable medium other than the computer-readable storage medium.
  • the computer-readable signal medium may send, propagate, or transmit the program for use by or in combination with the instruction execution system, apparatus, or device .
  • the program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to: wire, optical cable, RF (Radio Frequency), etc., or any suitable combination of the above.
  • the above-mentioned computer-readable medium may be included in the above-mentioned server; or it may exist separately without being installed in the server.
  • the above-mentioned computer-readable medium carries one or more programs.
  • the server is caused to: obtain the first location information used to indicate the location of the target user, and obtain the candidate push information set ;
  • For the candidate push information in the candidate push information set perform the following determining step: determine the first matching information of the candidate push information, where the first matching information is used to characterize the target user’s information on the candidate push information based on the first location information Degree of preference; according to the first matching information, determine the first operation information corresponding to the candidate push information, where the first operation information is used to characterize the probability of the target user performing the target operation on the candidate push information based on the first matching information;
  • the corresponding first operation information selects candidate push information from the candidate push information set; pushes the selected candidate push information to the terminal device corresponding to the target user.
  • the computer program code used to perform the operations of the embodiments of the present disclosure may be written in one or more programming languages or a combination thereof.
  • the programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and Conventional procedural programming language-such as "C" language or similar programming language.
  • the program code can be executed entirely on the user's computer, partly on the user's computer, executed as an independent software package, partly on the user's computer and partly executed on a remote computer, or entirely executed on the remote computer or server.
  • the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to Connect via the Internet).
  • LAN local area network
  • WAN wide area network
  • each block in the flowchart or block diagram can represent a module, program segment, or part of code, and the module, program segment, or part of code contains one or more for realizing the specified logical function Executable instructions.
  • the functions marked in the block may also occur in a different order from the order marked in the drawings. For example, two blocks shown in succession can actually be executed substantially in parallel, or they can sometimes be executed in the reverse order, depending on the functions involved.
  • each block in the block diagram and/or flowchart, and the combination of the blocks in the block diagram and/or flowchart can be implemented by a dedicated hardware-based system that performs the specified functions or operations Or it can be realized by a combination of dedicated hardware and computer instructions.
  • the units involved in the embodiments described in the present disclosure can be implemented in software or hardware.
  • the described unit may also be provided in the processor.
  • a processor includes an acquisition unit, an operation information determination unit, a selection unit, and a pushing unit.
  • the names of these units do not constitute a limitation on the unit itself under certain circumstances.
  • the obtaining unit can also be described as "obtaining the first location information used to indicate the location of the target user, and obtaining candidate push information Set unit".

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Telephone Function (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • User Interface Of Digital Computer (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本公开的实施例公开了用于推送信息的方法和装置。该方法的一具体实施方式包括：获取用于表示目标用户的位置的第一位置信息，获取候选推送信息集；对于候选推送信息集中的候选推送信息，确定该候选推送信息的第一匹配信息；根据第一匹配信息，确定该候选推送信息对应的第一操作信息，第一操作信息用于表征目标用户对该候选推送信息执行目标操作的概率；根据对应的第一操作信息，从候选推送信息集中选取候选推送信息；向目标用户对应的终端设备推送所选取的候选推送信息。该实施方式有助于提升推送信息的准确性。

Description

用于推送信息的方法和装置
相关申请的交叉引用
本申请基于申请号为201910261259.4、申请日为2019年04月02日的中国专利申请提出，并要求该中国专利申请的优先权，该中国专利申请的全部内容在此引入本申请作为参考。
技术领域
本公开的实施例涉及计算机技术领域，具体涉及用于推送信息的方法和装置。
背景技术
在一些信息推送类应用中，可以向用户推送的信息包括一些与地域相关的信息。例如，用于描述发生在某市的事件的信息、用于介绍某个街道的交通状况等等。
对于这类与地域相关的信息，通常只有对应地域的用户可能对这些信息比较感兴趣。基于此，一些信息推送类应用通常是基于用户所在的位置，向用户推送一些与用户所在的位置相关的信息。
发明内容
本公开的实施例提出了用于推送信息的方法和装置。
第一方面，本公开的实施例提供了一种用于推送信息的方法，该方法包括：获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；对于候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候选推送信息的第一匹配信息，其中，第一匹配信息用于表征基于第一位置信息，目标用户对该候选推送信息的偏好程度；根据第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，第一操作信息用于表征基于第一匹配信息，目标用户对该候选 推送信息执行目标操作的概率；根据对应的第一操作信息，从候选推送信息集中选取候选推送信息；向目标用户对应的终端设备推送所选取的候选推送信息。
在一些实施例中，上述方法还包括：根据第一位置信息，确定至少一个第二位置信息；以及确定步骤还包括：对于至少一个第二位置信息中的第二位置信息，确定该候选推送信息的第二匹配信息，其中，第二匹配信息用于表征基于该第二位置信息，目标用户对该候选推送信息的偏好程度；根据第二匹配信息，确定该候选推送信息对应的第二操作信息，其中，第二操作信息用于表征基于第二匹配信息，目标用户对该候选推送信息执行目标操作的概率；以及根据对应的第一操作信息，从候选推送信息集中选取候选推送信息，包括：根据对应的第一操作信息和至少一个第二操作信息，从候选推送信息集中选取候选推送信息。
在一些实施例中，根据第一位置信息，确定至少一个第二位置信息，包括：根据第一位置信息和预设的至少一个距离阈值，确定至少一个第二位置信息，其中，距离阈值用于表征第一位置信息指示的位置与第二位置信息指示的位置之间的距离。
在一些实施例中，根据第一匹配信息，确定该候选推送信息对应的第一操作信息，包括：将第一匹配信息输入至预先训练的信息预测模型，得到该候选推送信息对应的第一操作信息，其中，信息预测模型用于表征匹配信息与操作信息之间的对应关系；以及根据第二匹配信息，确定该候选推送信息对应的第二操作信息，包括：将第二匹配信息输入至信息预测模型，得到该候选推送信息对应的第二操作信息。
在一些实施例中，第一匹配信息包括以下至少一项：用于表征目标用户对与第一位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与第一位置信息指示的位置的相关度的信息、用于表征第一位置信息指示的位置的关注度的信息。
在一些实施例中，第二匹配信息包括以下至少一项：用于表征目标用户对与该第二位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与该第二位置信息指示的位置的相关度 的信息、用于表征该第二位置信息指示的位置的关注度的信息。
在一些实施例中，信息预测模型通过如下步骤训练得到：获取训练样本集，其中，训练样本集中的训练样本包括样本匹配信息和样本指示信息，其中，样本匹配信息包括目标用户对应的历史推送信息的匹配信息，样本指示信息用于指示目标用户是否对对应的历史推送信息执行过目标操作；获取预先建立的初始模型；利用机器学习的方法，基于训练样本集和预设的损失函数，对初始模型进行训练，以及将训练后的初始模型确定为信息预测模型。
在一些实施例中，根据对应的第一操作信息和至少一个第二操作信息，从候选推送信息集中选取候选推送信息，包括：对于候选推送信息集中的候选推送信息，响应于确定该候选推送信息对应的目标概率大于预设的概率阈值，选取该候选推送信息，其中，目标概率用于表示该候选推送信息对应的概率集中的最大值或概率集中的各个概率的平均值，其中，概率集由该候选推送信息对应的第一操作信息表征的概率和对应的至少一个第二操作信息分别表征的概率组成。
在一些实施例中，获取候选推送信息集，包括：根据第一位置信息，获取候选推送信息集，其中，候选推送信息集中的候选推送信息与第一位置信息指示的位置相关。
在一些实施例中，向目标用户对应的终端设备推送所选取的候选推送信息，包括：响应于检测到目标用户的预设操作，向终端设备推送所选取的候选推送信息，以及控制在终端设备上显示所选取的候选推送信息，其中，预设操作用于请求推送与第一位置信息指示的位置相关的信息。
第二方面，本公开的实施例提供了一种用于推送信息的装置，该装置包括：获取单元，被配置成获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；操作信息确定单元，被配置成对于候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候选推送信息的第一匹配信息，其中，第一匹配信息用于表征基于第一位置信息，目标用户对该候选推送信息的偏好程度；根据第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，第一操作信 息用于表征基于第一匹配信息，目标用户对该候选推送信息执行目标操作的概率；选取单元，被配置成根据对应的第一操作信息，从候选推送信息集中选取候选推送信息；推送单元，被配置成向目标用户对应的终端设备推送所选取的候选推送信息。
在一些实施例中，上述用于推送信息的装置还包括：位置信息确定单元，被配置成根据第一位置信息，确定至少一个第二位置信息；以及上述操作信息确定单元，进一步被配置成对于至少一个第二位置信息中的第二位置信息，确定该候选推送信息的第二匹配信息，其中，第二匹配信息用于表征基于该第二位置信息，目标用户对该候选推送信息的偏好程度；根据第二匹配信息，确定该候选推送信息对应的第二操作信息，其中，第二操作信息用于表征基于第二匹配信息，目标用户对该候选推送信息执行目标操作的概率；以及上述选取单元，进一步被配置成根据对应的第一操作信息和至少一个第二操作信息，从候选推送信息集中选取候选推送信息。
在一些实施例中，上述位置信息确定单元进一步被配置成根据第一位置信息，确定至少一个第二位置信息，包括：根据第一位置信息和预设的至少一个距离阈值，确定至少一个第二位置信息，其中，距离阈值用于表征第一位置信息指示的位置与第二位置信息指示的位置之间的距离。
在一些实施例中，上述操作信息确定单元进一步被配置成将第一匹配信息输入至预先训练的信息预测模型，得到该候选推送信息对应的第一操作信息，其中，信息预测模型用于表征匹配信息与操作信息之间的对应关系；以及上述操作信息确定单元进一步被配置成：将第二匹配信息输入至信息预测模型，得到该候选推送信息对应的第二操作信息。
在一些实施例中，第一匹配信息包括以下至少一项：用于表征目标用户对与第一位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与第一位置信息指示的位置的相关度的信息、用于表征第一位置信息指示的位置的关注度的信息。
在一些实施例中，第二匹配信息包括以下至少一项：用于表征目 标用户对与该第二位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与该第二位置信息指示的位置的相关度的信息、用于表征该第二位置信息指示的位置的关注度的信息。
在一些实施例中，信息预测模型通过如下步骤训练得到：获取训练样本集，其中，训练样本集中的训练样本包括样本匹配信息和样本指示信息，其中，样本匹配信息包括目标用户对应的历史推送信息的匹配信息，样本指示信息用于指示目标用户是否对对应的历史推送信息执行过目标操作；获取预先建立的初始模型；利用机器学习的方法，基于训练样本集和预设的损失函数，对初始模型进行训练，以及将训练后的初始模型确定为信息预测模型。
在一些实施例中，上述选取单元进一步被配置成对于候选推送信息集中的候选推送信息，响应于确定该候选推送信息对应的目标概率大于预设的概率阈值，选取该候选推送信息，其中，目标概率用于表示该候选推送信息对应的概率集中的最大值或概率集中的各个概率的平均值，其中，概率集由该候选推送信息对应的第一操作信息表征的概率和对应的至少一个第二操作信息分别表征的概率组成。
在一些实施例中，上述获取单元进一步被配置成根据第一位置信息，获取候选推送信息集，其中，候选推送信息集中的候选推送信息与第一位置信息指示的位置相关。
在一些实施例中，上述推送单元进一步被配置成响应于检测到目标用户的预设操作，向终端设备推送所选取的候选推送信息，以及控制在终端设备上显示所选取的候选推送信息，其中，预设操作用于请求推送与第一位置信息指示的位置相关的信息。
第三方面，本公开的实施例提供了一种服务器，该服务器包括：一个或多个处理器；存储装置，用于存储一个或多个程序；当一个或多个程序被一个或多个处理器执行，使得一个或多个处理器实现如第一方面中任一实现方式描述的方法。
第四方面，本公开的实施例提供了一种计算机可读介质，其上存储有计算机程序，该计算机程序被处理器执行时实现如第一方面中任一实现方式描述的方法。
对于一与位置相关的信息，通常只有长期生活在对应位置附近一定范围内的用户可能对该信息感兴趣。其中，可能会有部分生活在对应位置附近一定范围内、但对这类与位置相关的信息不感兴趣的用户。因此，对一与位置相关的信息感兴趣的用户的数目是相对较少的。在推送这类与位置相关的信息时，只需要向对这类信息感兴趣的用户推送，以避免不必要的信息推送带来额外的资源消耗。
本公开的实施例提供的用于推送信息的方法和装置，基于用户在其所在的位置处，对候选推送信息的偏好程度来预测用户对候选推送信息执行目标操作的概率。预测的概率即可以反映用户对其所在位置相关的信息的感兴趣程度，进而，可以根据预测的概率有选择性的向用户推送信息，以尽可能地向对其所在位置相关的信息感兴趣的用户推送对应信息，也避免向对其所在位置相关的信息不感兴趣的用户推送对应信息，从而减少信息推送过程中由于向对其所在位置相关的信息不感兴趣的用户推送对应信息所造成的不必要的资源消耗。
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所作的详细描述，本公开的其它特征、目的和优点将会变得更明显：
图1是本公开的一个实施例可以应用于其中的示例性系统架构图；
图2是根据本公开的用于推送信息的方法的一个实施例的流程图；
图3是根据本公开的用于推送信息的方法的又一个实施例的流程图；
图4是根据本公开的实施例的用于推送信息的方法的一个应用场景的示意图；
图5是根据本公开的用于推送信息的装置的一个实施例的结构示意图；
图6是适于用来实现本公开的实施例的电子设备的结构示意图。
具体实施方式
下面结合附图和实施例对本公开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，本公开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公开。
图1示出了可以应用本公开的用于推送信息的方法或用于推送信息的装置的实施例的示例性架构100。
如图1所示，系统架构100可以包括终端设备101、102、103，网络104和服务器105。网络104用以在终端设备101、102、103和服务器105之间提供通信链路的介质。网络104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。
终端设备101、102、103通过网络104与服务器105交互，以接收或发送消息等。终端设备101、102、103上可以安装有各种客户端应用。例如，浏览器类应用、搜索类应用、即时通信工具、阅读类应用、信息分享类应用等。
终端设备101、102、103可以是硬件，也可以是软件。当终端设备101、102、103为硬件时，可以是具有显示屏的各种电子设备，包括但不限于智能手机、平板电脑、膝上型便携计算机和台式计算机等等。当终端设备101、102、103为软件时，可以安装在上述所列举的电子设备中。其可以实现成多个软件或软件模块(例如用来提供分布式服务的多个软件或软件模块)，也可以实现成单个软件或软件模块。在此不做具体限定。
服务器105可以是提供各种服务的服务器，例如为终端设备101、102、103上安装的一些应用(如信息分享类应用)提供支持的后端服务器。服务器105可以基于终端设备101、102、103对应的用户的位置和候选推送信息集，预测用户对各个候选推送信息的操作信息。进而可以根据对应的操作信息，从候选推送信息集中选取候选推送信息， 以及将选取的候选推送信息推送至终端设备101、102、103。
需要说明的是，本公开的实施例所提供的用于推送信息的方法一般由服务器105执行，相应地，用于推送信息的装置一般设置于服务器105中。
还需要指出的是，终端设备101、102、103中也可以安装有信息处理类应用，终端设备101、102、103也可以基于信息处理类应用对候选推送信息集中的候选推送信息进行处理。此时，用于推送信息的方法也可以由终端设备101、102、103执行，相应地，用于推送信息的装置也可以设置于终端设备101、102、103中。此时，示例性系统架构100可以不存在服务器105和网络104。
需要说明的是，服务器105可以是硬件，也可以是软件。当服务器105为硬件时，可以实现成多个服务器组成的分布式服务器集群，也可以实现成单个服务器。当服务器105为软件时，可以实现成多个软件或软件模块(例如用来提供分布式服务的多个软件或软件模块)，也可以实现成单个软件或软件模块。在此不做具体限定。
应该理解，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。
继续参考图2，其示出了根据本公开的用于推送信息的方法的一个实施例的流程200。该用于推送信息的方法包括以下步骤：
步骤201，获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集。
在本实施例中，目标用户可以指与用于推送信息的方法的执行主体(如图1所示的服务器105)通信连接的终端设备(如图1所示的终端设备101、102、103)对应的用户。第一位置信息可以是能够表示位置的各种信息。例如，第一位置信息可以是经纬度信息、通讯地址信息等等。
上述执行主体可以基于现有的各种定位技术确定目标用户的位置。例如，可以基于GPS(Global Positioning System，全球定位系统)技术确定目标用户的位置。可选地，可以直接将确定的目标用户的位置作为第一位置信息，也可以从本地或其它存储设备获取预设的用于 表示确定的目标用户的位置的信息作为第一位置信息。
在本实施例中，候选推送信息集可以是由技术人员预先指定的一些信息组成的集合，也可以是根据预设的筛选条件确定出的一些信息组成的集合。候选推送信息集中的候选推送信息可以是各种类型的信息。例如，候选推送信息可以是图像、文本、视频、音频等等。上述执行主体可以从本地或其它存储设备(如连接的数据库、第三方数据平台等)获取候选推送信息集。
在本实施例的一些可选的实现方式中，可以根据第一位置信息，获取候选推送信息集。其中，候选推送信息集中的候选推送信息与第一位置信息指示的位置相关。
由此，可以避免候选推送信息集中出现与第一位置信息指示的位置无关的信息，从而有助于减少后续需要处理的候选推送信息的数目，加快处理速度。
候选推送信息与第一位置信息指示的位置是否相关可以根据预先设置的相关条件来判断。例如，相关条件可以是候选推送信息描述的内容与第一位置信息指示的位置相关。作为示例，候选推送信息为一描述某种食物的文章，若所描述的食物为第一位置信息指示的位置所在的一定区域内特产的食物，那么可以认为候选推送信息与第一位置信息指示的位置相关。
又例如，相关条件可以是候选推送信息中出现有第一位置信息指示的位置所对应的预设关键词集中的预设关键词。其中，预设关键词集可以包括第一位置信息指示的位置的各种描述方式。预设关键词集也可以包括第一位置信息指示的位置所在的一定区域内所覆盖的所有位置分别对应的各种描述方式。
应当可以理解，一般地，目标用户的位置可以是非常具体的一个位置。例如，目标用户的位置可以精确到具体的经纬度。这种情况下，与目标用户的位置相关较高的信息可能较少。因此，根据实际的应用需求，可以将目标用户的位置泛化到目标用户的位置所在的一定区域内。此时，与目标用户的位置所在的一定区域内相关的信息都可以认为是与目标用户的位置相关的信息。
上述执行主体可以根据第一位置信息，从本地或其它存储设备(如连接的数据库、第三方数据平台等)获取与第一位置信息指示的位置相关的一些信息组成候选推送信息集。
步骤202，对于候选推送信息集中的候选推送信息，执行如下确定步骤：
步骤2021，确定该候选推送信息的第一匹配信息。
在本实施例中，第一匹配信息可以用于表征基于第一位置信息，目标用户对该候选推送信息的偏好程度。偏好程度可以根据用户和该候选推送信息的各种相关信息来衡量。
具体地，可以根据实际的应用场景和应用需求灵活选择不同的、一个或多个可以表征基于第一位置信息，目标用户对该候选推送信息的偏好程度的信息来作为第一匹配信息。
可选地，第一匹配信息可以包括用户在预设时间段内具有较高偏好的一些历史推送信息与该候选推送信息的相似度。其中，用户在预设时间段内具有较高偏好的一些历史推送信息可以根据在预设时间段内，用户在第一位置信息指示的位置处的历史行为数据来确定。
例如，可以从在预设时间段内，且用户在第一位置信息指示的位置处时，向用户推送的各个历史推送信息中选取用户点击或评论过的历史推送信息作为用户具有较高偏好的历史推送信息。
一般地，用户在预设时间段内具有较高偏好的一些历史推送信息与该候选推送信息的相似度越高，可以表示用户在第一位置信息指示的位置处，对该候选推送信息的偏好程度越高。
可选地，第一匹配信息可以包括以下至少一项：用于表征目标用户对与第一位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与第一位置信息指示的位置的相关度的信息、用于表征第一位置信息指示的位置的关注度的信息。
其中，目标用户对与第一位置信息指示的位置相关的信息的偏好程度可以根据实际的应用场景，采用各种方式来确定。
可选地，可以根据目标用户的、针对历史推送过的与第一位置信息指示的位置相关的信息的历史行为数据来分析目标用户对与第一位 置信息指示的位置相关的信息的偏好程度。
一般地，用户对偏好的推送信息会有一些交互操作。例如，点击操作、评论操作、收藏或分享推送信息的连接地址的操作等等。用户对于不同的推送信息所执行的交互操作的一些属性信息也会有所不同。例如点击操作的次数、浏览时长等等。因此，可以根据用户的历史行为数据来分析用户对推送的信息的偏好程度。通常情况下，用户执行交互操作越多或越频繁，可以表示用户对对应的推送信息的偏好程度越高。
作为示例，可以统计目标历史时间段内，向目标用户推送过的、与第一位置信息指示的位置相关的信息的总数目作为第一数目，以及在推送的这些信息中，被目标用户执行过预设操作的信息的数目作为第二数目。之后，可以利用第二数目与第一数目的商值来表征目标用户对与第一位置信息指示的位置相关的信息的偏好程度。
其中，目标历史时间段可以是由技术人员预先指定的任意时间段。例如，目标时间段可以是以当前时间为起点，在当前时间之前的三个月所对应的时间段。预设操作可以是由技术人员预先指定的各种用户操作。
可选地，可以利用现有的一些开源的用于预测的模型(如点击率预测模型等)来预测目标用户对与第一位置信息指示的位置相关的信息的操作信息(如点击率)。基于此，可以利用预测的操作信息来表征目标用户对与第一位置信息指示的位置相关的信息的偏好程度。
其中，用于预测的模型也可以由技术人员预先利用大量的用户历史行为数据训练得到。举例来说，信息预测模型可以根据用于表征用户的特征向量和用于表征候选推送信息的特征向量来预测用户对候选推送信息的操作信息(如点击率等)。
在本实施例中，候选推送信息与第一位置信息指示的位置的相关度可以指候选推送信息与第一位置信息指示的位置之间的关联程度。举例来说，候选推送信息″A″是一篇描述第一位置信息指示的位置所在的区域中的景点的文章。候选推送信息″B″是一篇出现有表示第一位置信息指示的位置的名称、但是内容与第一位置信息指示的位 置无关的文章。那么，一般可以认为候选推送信息″A″与第一位置信息指示的位置的相关度高于候选推送信息″B″与第一位置信息指示的位置的相关度。
其中，相关度的确定方法可以根据不同的应用场景选取不同的方法。例如，可以先获取候选推送信息对应的评论信息集。其中，评论信息集可以包括候选推送信息的一些评论信息。之后，可以获取评论信息集中的各条评论信息分别对应的用户所在的位置信息，得到位置信息集。其中，评论信息对应的用户可以指发布评论信息的终端设备对应的用户。之后，可以将第一位置信息指示的位置所在的一定区域确定为目标区域。然后，可以统计位置信息集中的、指示的位置属于目标区域的位置信息的数目作为第一数目，以及确定位置信息集包括的位置信息的总数目作为第二数目。进而，可以利用第一数目与第二数目的商值来表征候选推送信息与第一位置信息指示的位置的相关度。一般地，第一数目与第二数目的商值越大，可以表示候选推送信息与第一位置信息指示的位置的相关度越高。
由于对与第一位置信息指示的位置相关的信息感兴趣的用户通常是长期在第一位置信息指示的位置所在的一定区域范围内的用户。那么，对于与第一位置信息指示的位置相关的信息来说，会对这些信息进行评论的用户也是长期在第一位置信息指示的位置所在的一定区域范围内的用户。因此，可以根据候选推送信息的各条评论信息对应的用户的位置信息来分析候选推送信息与第一位置信息指示的位置的相关度。
可选地，可以利用现有的各种关键词提取方法提取候选推送信息的关键词，得到候选推送信息对应的关键词集。然后，可以获取预设的与第一位置信息指示的位置对应的关键词集。之后，可以利用候选推送信息对应的关键词集和第一位置信息指示的位置对应的关键词集的相似度来表征候选推送信息与第一位置信息指示的位置的相关度。一般地，候选推送信息对应的关键词集和第一位置信息指示的位置对应的关键词集的相似度越高，可以表征候选推送信息与第一位置信息指示的位置的相关度越高。
其中，确定两个关键词集的相似度的算法可以采用现有的各种相关算法。例如，基于TF-IDF(Term Frequency-Inverse Document Frequency，词频-逆文本频率指数)的算法、基于SimRank(一种衡量任意两个对象相似程度的模型)的算法、基于WMD(Word Mover’s Distance，一种衡量两个对象之间的距离的方法)的算法等。
可选地，可以利用预先训练的相关度确定模型来分析候选推送信息与第一位置信息指示的位置的相关度。其中，相关度确定模型可以用于根据候选推送信息和第一位置信息确定候选推送信息和第一位置信息之间的相关度。具体地，相关度确定模型可以基于大量的训练样本训练得到。每个训练样本可以包括历史推送信息、位置信息，以及历史推送信息和位置信息之间的相关度。其中，训练样本中的历史推送信息和位置信息之间的相关度可以利用前述一些相关度确定方法确定，也可以由技术人员预先进行标注。
在本实施例中，第一位置信息指示的位置的关注度可以用于表征第一位置信息指示的位置的受关注程度。例如，一般地，景点所在位置的关注度通常要高于非景点所在位置的关注度。
由于一般情况下，各个位置的关注度都比较稳定。因此，可以预先设置各个位置对应的关注度。当然，在一些情况下，某些位置的关注度需要动态调整。例如，某个位置所在区域近期开展了比较知名的活动，这种情况下，就可以适当调高这个位置的关注度。
应当可以理解，用于表征目标用户对与一位置信息指示的位置相关的信息的偏好程度的信息可以是各种形式的信息。例如，在使用点击率表征目标用户对与一位置信息指示的位置相关的信息的偏好程度时，点击率就可以是表征目标用户对与一位置信息指示的位置相关的信息的偏好程度的信息。此时，还可以预先设置不同点击率对应的表示信息，进而获取点击率对应的表示信息来作为表征目标用户对与一位置信息指示的位置相关的信息的偏好程度的信息。
类似地，用于表征该候选推送信息与第一位置信息指示的位置的相关度的信息也可以是各种形式的信息。用于表征第一位置信息指示的位置的关注度的信息也可以是各种形式的信息。具体的表示方法可 以由技术人员根据实际的应用需求灵活设置。
步骤2022，根据第一匹配信息，确定该候选推送信息对应的第一操作信息。
在本实施例中，第一操作信息可以用于表征基于第一匹配信息，目标用户对该候选推送信息执行目标操作的概率。其中，目标操作可以是预先指定的用户操作(如用于表示点击操作的用户操作，用于表示评论操作的用户操作等)，也可以是根据候选推送信息确定的用户操作(如目标操作可以是候选推送信息支持的任意用户操作等)。
在不同的应用场景，根据第一匹配信息，可以选取不同的确定候选推送信息对应的第一操作信息的方法。
可选地，可以选取第一匹配信息包括的各个信息指示的最大值或各个信息指示的数值的平均值作定为第一操作信息。
在本实施例的一些可选的实现方式中，可以将第一匹配信息输入至预先训练的信息预测模型，得到候选推送信息对应的第一操作信息。其中，信息预测模型可以用于表征匹配信息与操作信息之间的对应关系。
在本实施例的一些可选的实现方式中，信息预测模型可以通过如下步骤训练得到：
步骤一，获取训练样本集。其中，训练样本集中的训练样本可以包括样本匹配信息和样本指示信息。其中，样本匹配信息可以包括目标用户对应的历史推送信息的匹配信息。样本指示信息可以用于指示目标用户是否对对应的历史推送信息执行过目标操作。
在本步骤中，样本匹配信息根据不同的应用需求，可以包括以下至少一项：用于表征目标用户对目标用户对应的位置相关的信息的偏好程度的信息、用于表征历史推送信息与目标用户对应的位置的相关度的信息、用于表征目标用户对应的位置的关注度的信息。其中，目标用户对应的位置可以指向用户推送历史推送信息时所确定的目标用户的位置。
可选地，历史推送信息可以指与与目标用户对应的位置相关的信息。
步骤二，获取预先建立的初始模型。
在本步骤中，初始模型可以是未经训练或训练完成的各种人工神经网络，也可以是各种人工神经网络进行组合所得到的模型。初始模型也可以由技术人员根据实际的应用需求(如需要哪些网络层、每个网络层的参数、卷积核的大小等)，利用一些深度学习框架(如TensorFlow、Caffe等)进行搭建。
可选地，可以基于现有的一些回归模型(如基于逻辑回归的模型、基于逐步回归的模型等)来搭建初始模型。
步骤三，利用机器学习的方法，基于训练样本集和预设的损失函数，对初始模型进行训练，以及将训练后的初始模型确定为信息预测模型。
在本步骤中，损失函数可以由技术人员预先设计。损失函数通常用于根据初始模型的输出与对应期望输出之间的差异。一般地，希望损失函数的值尽可能的小。
损失函数的选择也可以根据初始模型进行确定。不同的初始模型可以对应选择不同的损失函数。例如，对于基于逻辑回归的初始模型，可以选择现有的通过极大似然估计推导得到的损失函数。
具体地，可以每次从训练样本集中选取至少一个训练样本。然后将选取的训练样本中的样本匹配信息分别输入至初始模型，得到选取的训练样本分别对应的输出操作信息。之后，可以根据预设的概率阈值确定输出操作信息表征的是目标用户会对输入的样本匹配信息对应的历史推送信息执行目标操作或目标用户不会对输入的样本匹配信息对应的历史推送信息执行目标操作。
举例来说，若概率阈值为0.5。那么，若输出操作信息表征的概率大于0.5，则可以认为目标用户会对输入的样本匹配信息对应的历史推送信息执行目标操作。对应地，若输出操作信息表征的概率小于0.5，则可以认为目标用户不会对输入的样本匹配信息对应的历史推送信息执行目标操作。
然后，可以根据输出操作信息和样本指示信息，确定预设的损失函数的值。之后，根据损失函数的值确定初始模型是否训练完成。若 根据损失函数的值确定初始模型训练完成，可以将训练完成的初始模型确定为信息预测模型。其中，可以预先设置在损失函数的值符合预设条件时，确定初始模型训练完成。
若根据损失函数的值确定初始模型未训练完成，可以基于损失函数的值，利用反向传播和梯度下降算法调整初始模型的参数。同时，可以从训练样本集中重新选取至少一个训练样本，并将调整后的初始模型确定为初始模型，继续进行训练。
一般地，对初始模型的训练都需要反复地多次迭代才能训练完成。其中，可以基于每次选取的各个训练样本对应的损失函数的值的平均值，调整初始模型的参数。
可选地，信息预测模型可以通过如下步骤训练得到：获取训练样本集。其中，训练样本集中的训练样本可以包括样本匹配信息和样本操作信息。其中，样本匹配信息可以包括目标用户对应的历史推送信息的匹配信息。样本操作信息可以用于指示目标用户对对应的历史推送信息执行目标操作的概率。样本操作信息可以由技术人员预先进行标注。之后，可以采用曲线拟合的方法，基于训练样本集进行曲线拟合，以及将得到的拟合曲线作为信息预测模型。
其中，可以利用现有的各种能够用于曲线拟合的方法(如基于最小二乘法的拟合算法等)或应用，基于训练样本集进行曲线拟合。
步骤203，根据对应的第一操作信息，从候选推送信息集中选取候选推送信息。
在本实施例中，由于第一操作信息可以用于表征基于第一匹配信息，目标用户对候选推送信息执行目标操作的概率。因此，可以按照对应的第一操作信息从大到小的顺序，从候选推送信息集中选取候选推送信息目标数目个候选推送信息。
其中，目标数目可以由技术人员预先设置，也可以根据设置的确定条件确定(如将候选推送信息集包括的候选推送信息的总数目的百分之十确定为目标数目等)。
可选地，还可以先根据预设的概率阈值，从候选推送信息集中选取对应的第一操作信息表征的概率大于概率阈值的候选推送信息，得 到候选推送子集。之后，可以从候选推送子集中随机选取预设数目个候选推送信息。
步骤204，向目标用户对应的终端设备推送所选取的候选推送信息。
在本实施例的一些可选的实现方式中，响应于检测到目标用户的预设操作，向终端设备推送所选取的候选推送信息，以及控制在终端设备上显示所选取的候选推送信息。其中，预设操作可以用于请求推送与第一位置信息指示的位置相关的信息。
其中，预设操作可以包括由技术人员预先设置的任意用户操作。例如，预设操作可以包括刷新操作、点击终端设备显示的页面上的预设控件的操作等等。
如前述分析，由于对与第一位置信息指示的位置相关的信息感兴趣的用户的数目相对较少，因此可以在接收到用户的推送请求时，再向用户推送与用户对应的位置相关的信息。由此可以尽量的避免不必要的信息推送，减少信息推送过程中的资源消耗。
可选地，通过这种方式还可以在接收到用户的推送请求时，可以统一集中在终端设备上显示候选推送信息，以便于用户快速的浏览到相关信息。举例来说，可以在用户浏览信息的页面上设置两个按钮，用于切换不同的显示页面。一个按钮可以控制用于展示与用户对应的位置相关的信息。另一个按钮可以控制用于展示其它的推送信息。
基于此，使得在用户期望浏览与用户对应的位置相关的信息时，可以快速的在用户对应的终端设备上进行显示，以便于用户查看。
本公开的上述实施例提供的方法基于用户在其所在的位置处，对候选推送信息的偏好程度来预测用户对候选推送信息执行目标操作的概率。进而，可以根据预测的概率，选取推送信息向用户进行推送，从而有助于提升推送信息的准确性，避免向对与对应的位置相关的信息不感兴趣的用户推送相关信息，减少信息推送过程中的资源消耗。在此基础上，还可以根据用户的推送请求，向用户推送与用户对应的位置相关的信息并控制在用户对应的终端设备上进行显示，从而实现在用户想要浏览与用户对应的位置相关的信息时，可以及时的查看到 相关信息。
进一步参考图3，其示出了用于推送信息的方法的又一个实施例的流程300。该用于推送信息的方法的流程300，包括以下步骤：
步骤301，获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集。
本步骤的具体的执行过程可参考图2对应实施例中的步骤201的相关说明，在此不再赘述。
步骤302，根据第一位置信息，确定至少一个第二位置信息。
在本实施例中，可以预先设置存储有第一位置信息和至少一个第二位置信息之间的对应关系的对应关系表。此时，在确定第一位置信息后，可以在对应关系表查找与第一位置信息对应的至少一个第二位置信息。
在本实施例的一些可选的实现方式中，可以根据第一位置信息和预设的至少一个距离阈值，确定至少一个第二位置信息。其中，距离阈值可以用于表征第一位置信息指示的位置与第二位置信息指示的位置之间的距离。
其中，距离阈值可以由技术人员进行预先设置。例如，距离阈值可以是十公里，那么可以将距离第一位置信息指示的位置十公里以内且包含第一位置信息指示的位置的位置确定为一个第二位置，从而可以得到第二位置信息。
一般地，由于位置信息是具有一定的层级关系的。例如，XX省XX市XX区XX街道。其中，XX省可以是一个位置信息，XX市也可以是一个位置信息，同样地，XX区和XX街道同样都是位置信息。即位置信息可以具有不同粒度的表示。因此，在实际应用中，如何选择位置信息的表示粒度也是需要考量的一个方面。
举例来说，获取用户的位置在XX区，但是用户可能对XX区相关的信息并不感兴趣，而对XX市相关的信息比较感兴趣，或者也可能对XX街道相关的信息比较感兴趣。类似地，一个推送信息可以是与XX区相关的信息，也同时可以是XX街道相关的信息。
因此，在确定目标用户的位置之后，可以考虑将确定的用户的位 置泛化到不同粒度表示的多个位置，进而可以从不用粒度表示的多个位置综合衡量目标用户对候选推送信息的操作信息，进一步提升预测的操作信息的准确性。
需要说明的是，在本公开中，为了便于描述不同的位置，分贝命名为第一位置、第二位置等，本领域技术人员应该理解，其中的第一或第二并不构成对位置的特殊限定。
步骤303，对于候选推送信息集中的候选推送信息，执行如下确定步骤：
步骤3031，确定该候选推送信息的第一匹配信息。
步骤3032，根据第一匹配信息，确定该候选推送信息对应的第一操作信息。
上述步骤3031和3032的具体的执行过程可参考图2对应实施例中的步骤2021和2022的相关说明，在此不再赘述。
步骤3033，对于至少一个第二位置信息中的第二位置信息，执行如下步骤：
步骤30331，确定该候选推送信息的第二匹配信息。
在本步骤中，第二匹配信息可以用于表征基于该第二位置信息，目标用户对该候选推送信息的偏好程度。偏好程度可以根据用户和该候选推送信息的各种相关信息来衡量。
与第一匹配信息类似地，可以根据实际的应用场景和应用需求灵活选择不同的、一个或多个可以表征基于第二位置信息，目标用户对该候选推送信息的偏好程度的信息来作为第二匹配信息。
可选地，第二匹配信息可以包括用户在预设时间段内具有较高偏好的一些历史推送信息与该候选推送信息的相似度。其中，用户在预设时间段内具有较高偏好的一些历史推送信息可以根据在预设时间段内，用户在第二位置信息指示的位置处的历史行为数据来确定。
例如，可以从在预设时间段内，且用户在第二位置信息指示的位置处时，向用户推送的各个历史推送信息中选取用户点击或评论过的历史推送信息作为用户具有较高偏好的历史推送信息。
一般地，用户在预设时间段内具有较高偏好的一些历史推送信息 与该候选推送信息的相似度越高，可以表示用户在第二位置信息指示的位置处，对该候选推送信息的偏好程度越高。
可选地，第二匹配信息包括以下至少一项：用于表征目标用户对与该第二位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与该第二位置信息指示的位置的相关度的信息、用于表征该第二位置信息指示的位置的关注度的信息。
其中，第二匹配信息的确定方法可以参考图2对应实施例中的步骤2021中的关于确定第一匹配信息的相关说明，在此不再赘述。
步骤30332，根据第二匹配信息，确定该候选推送信息对应的第二操作信息。
在本步骤中，第二操作信息可以用于表征基于第二匹配信息，目标用户对该候选推送信息执行目标操作的概率。
在本实施例的一些可选的实现方式中，可以将第二匹配信息输入至信息预测模型，得到该候选推送信息对应的第二操作信息。
其中，确定第二操作信息的具体过程，以及信息预测模型可以参考图2对应实施例中的步骤2022中的相关说明，在此不再赘述。
步骤304，根据对应的第一操作信息和至少一个第二操作信息，从候选推送信息集中选取候选推送信息。
在本实施例的一些可选的实现方式中，对于候选推送信息集中的候选推送信息，响应于确定该候选推送信息对应的目标概率大于预设的概率阈值，可以选取该候选推送信息。其中，目标概率可以用于表示该候选推送信息对应的概率集中的最大值或概率集中的各个概率的平均值。其中，概率集可以由该候选推送信息对应的第一操作信息表征的概率和对应的至少一个第二操作信息分别表征的概率组成。
在本实施例中，在得到第一操作信息和至少一个第二操作信息时，可以根据第一操作信息和至少一个第二操作信息分别表征的概率作为考量是否选取候选推送信息的一个方面。在一些应用场景中，需要是否选取一个候选推送信息需要考量的因素可以由很多方面。此时，可以将第一操作信息和至少一个第二操作信息分别表征的概率作为其中一个方面，结合其它方面，综合考量是否选取每个候选推送信息。
例如，在一些情况下，需要考虑候选推送信息的格式，例如候选推送信息是视频还是文本，或候选推送信息是文本时，候选推送信息中包含的图像是否较多等等。从而可以结合目标用户的网络状态，选取合适的候选推送信息，以避免在用户在使用数据网络时，接收视频或包含图文较多的文本信息等。
步骤305，向目标用户对应的终端设备推送所选取的候选推送信息。
本步骤的具体的执行过程可参考图2对应实施例中的步骤204的相关说明，在此不再赘述。
继续参见图4，图4是根据本实施例的用于推送信息的方法的应用场景的一个示意图400。在图4的应用场景中，上述执行主体可以先基于GPS技术获取用户的位置的位置信息402为″北京市海淀区五道口″。然后可以将位置信息402泛化到不同粒度的两个位置信息401和403。如图中所示，位置信息401为″北京市海淀区″，位置信息403为″北京市″。
上述执行主体还可以从连接的数据库中404获取信息集405。然后对于信息集405中的一信息，可以分别确定基于三个不同粒度的位置信息，用户对该信息的点击率。下面以信息集405中的一个信息4051作为示例具体说明。
对于位置信息401，即″北京市海淀区″，可以确定用户对历史推送的与″北京市海淀区″相关的信息的点击率来表征用户对与″北京市海淀区″相关的信息的偏好程度。之后，可以分析信息4051的内容以确定信息4051与″北京市海淀区″的相关度。之后，可以获取预设的″北京市海淀区″对应的关注度。然后，可以由确定的″北京市海淀区″对应的点击率、相关度、关注度组成匹配信息406。接着，可以将匹配信息406输入至预先训练的点击率预测模型407，得到在″北京市海淀区″这一个位置信息下，预测的用户点击信息4051的概率408为0.9。
对于位置信息402，即″北京市海淀区五道口″，可以确定用户对历史推送的与″北京市海淀区五道口″相关的信息的点击率来表征用 户对与″北京市海淀区五道口″相关的信息的偏好程度。之后，可以分析信息4051的内容以确定信息4051与″北京市海淀区五道口″的相关度。之后，可以获取预设的″北京市海淀区五道口″对应的关注度。然后，可以由确定的″北京市海淀区五道口″对应的点击率、相关度、关注度组成匹配信息409。接着，可以将匹配信息409输入至预先训练的点击率预测模型407，得到在″北京市海淀区五道口″这一个位置信息下，预测的用户点击信息4051的概率410为0.8。
对于位置信息403，即″北京市″，可以确定用户对历史推送的与″北京市″相关的信息的点击率来表征用户对与″北京市″相关的信息的偏好程度。之后，可以分析信息4051的内容以确定信息4051与″北京市″的相关度。之后，可以获取预设的″北京市″对应的关注度。然后，可以由确定的″北京市″对应的点击率、相关度、关注度组成匹配信息411。接着，可以将匹配信息411输入至预先训练的点击率预测模型407，得到在″北京市″这一个位置信息下，预测的用户点击信息4051的概率412为0.2。
之后，可以从信息4051分别对应的上预测概率中选取最大值为0.9。由于0.9大于预设的阈值0.6，所以可以向用户对应的终端设备413推送信息4051，并在终端设备413上显示信息4051。
本公开的上述实施例提供的方法通过在得到用户的位置之后，将用户的位置泛化到不同的粒度的多个位置，从而对于各个粒度的位置，分别预测用户对候选推送信息的操作信息，从而可综合考虑是否推送候选推送信息，有助于进一步地提升推送信息的准确性。
进一步参考图5，作为对上述各图所示方法的实现，本公开提供了用于推送信息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各种电子设备中。
如图5所示，本实施例提供的用于推送信息的装置500包括获取单元501、操作信息确定单元502、选取单元503和推送单元504。其中，获取单元501被配置成获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；操作信息确定单元502被配置成对于候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候 选推送信息的第一匹配信息，其中，第一匹配信息用于表征基于第一位置信息，目标用户对该候选推送信息的偏好程度；根据第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，第一操作信息用于表征基于第一匹配信息，目标用户对该候选推送信息执行目标操作的概率；选取单元503被配置成根据对应的第一操作信息，从候选推送信息集中选取候选推送信息；推送单元504被配置成向目标用户对应的终端设备推送所选取的候选推送信息。
在本实施例中，用于推送信息的装置500中：获取单元501、操作信息确定单元502、选取单元503和推送单元504的具体处理及其所带来的技术效果可分别参考图2对应实施例中的步骤201、步骤202、步骤203和步骤204的相关说明，在此不再赘述。
在本实施例的一些可选的实现方式中，上述用于推送信息的装置500还包括：位置信息确定单元(图中未示出)被配置成根据第一位置信息，确定至少一个第二位置信息；以及上述操作信息确定单元502，进一步被配置成对于至少一个第二位置信息中的第二位置信息，确定该候选推送信息的第二匹配信息，其中，第二匹配信息用于表征基于该第二位置信息，目标用户对该候选推送信息的偏好程度；根据第二匹配信息，确定该候选推送信息对应的第二操作信息，其中，第二操作信息用于表征基于第二匹配信息，目标用户对该候选推送信息执行目标操作的概率；以及上述选取单元503进一步被配置成根据对应的第一操作信息和至少一个第二操作信息，从候选推送信息集中选取候选推送信息。
在本实施例的一些可选的实现方式中，上述位置信息确定单元进一步被配置成根据第一位置信息，确定至少一个第二位置信息，包括：根据第一位置信息和预设的至少一个距离阈值，确定至少一个第二位置信息，其中，距离阈值用于表征第一位置信息指示的位置与第二位置信息指示的位置之间的距离。
在本实施例的一些可选的实现方式中，上述操作信息确定单元502进一步被配置成将第一匹配信息输入至预先训练的信息预测模型，得到该候选推送信息对应的第一操作信息，其中，信息预测模型 用于表征匹配信息与操作信息之间的对应关系；以及上述操作信息确定单502进一步被配置成：将第二匹配信息输入至信息预测模型，得到该候选推送信息对应的第二操作信息。
在本实施例的一些可选的实现方式中，第一匹配信息包括以下至少一项：用于表征目标用户对与第一位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与第一位置信息指示的位置的相关度的信息、用于表征第一位置信息指示的位置的关注度的信息。
在一些实施例中，第二匹配信息包括以下至少一项：用于表征目标用户对与该第二位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与该第二位置信息指示的位置的相关度的信息、用于表征该第二位置信息指示的位置的关注度的信息。
在本实施例的一些可选的实现方式中，信息预测模型通过如下步骤训练得到：获取训练样本集，其中，训练样本集中的训练样本包括样本匹配信息和样本指示信息，其中，样本匹配信息包括目标用户对应的历史推送信息的匹配信息，样本指示信息用于指示目标用户是否对对应的历史推送信息执行过目标操作；获取预先建立的初始模型；利用机器学习的方法，基于训练样本集和预设的损失函数，对初始模型进行训练，以及将训练后的初始模型确定为信息预测模型。
在本实施例的一些可选的实现方式中，上述选取单元503进一步被配置成对于候选推送信息集中的候选推送信息，响应于确定该候选推送信息对应的目标概率大于预设的概率阈值，选取该候选推送信息，其中，目标概率用于表示该候选推送信息对应的概率集中的最大值或概率集中的各个概率的平均值，其中，概率集由该候选推送信息对应的第一操作信息表征的概率和对应的至少一个第二操作信息分别表征的概率组成。
在本实施例的一些可选的实现方式中，上述获取单元501进一步被配置成根据第一位置信息，获取候选推送信息集，其中，候选推送信息集中的候选推送信息与第一位置信息指示的位置相关。
在本实施例的一些可选的实现方式中，上述推送单元503进一步 被配置成响应于检测到目标用户的预设操作，向终端设备推送所选取的候选推送信息，以及控制在终端设备上显示所选取的候选推送信息，其中，预设操作用于请求推送与第一位置信息指示的位置相关的信息。
本公开的上述实施例提供的装置，通过获取单元获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；操作信息确定单元对于候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候选推送信息的第一匹配信息，其中，第一匹配信息用于表征基于第一位置信息，目标用户对该候选推送信息的偏好程度；根据第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，第一操作信息用于表征基于第一匹配信息，目标用户对该候选推送信息执行目标操作的概率；选取单元根据对应的第一操作信息，从候选推送信息集中选取候选推送信息；推送单元向目标用户对应的终端设备推送所选取的候选推送信息，从而有助于提升推送信息的准确性，避免向对与对应的位置相关的信息不感兴趣的用户推送相关信息，减少信息推送过程中的资源消耗。
下面参考图6，其示出了适于用来实现本公开的实施例的电子设备(例如图1中的服务器)600的结构示意图。图6示出的服务器仅仅是一个示例，不应对本公开的实施例的功能和使用范围带来任何限制。
如图6所示，电子设备600可以包括处理装置(例如中央处理器、图形处理器等)601，其可以根据存储在只读存储器(ROM)602中的程序或者从存储装置608加载到随机访问存储器(RAM)603中的程序而执行各种适当的动作和处理。在RAM 603中，还存储有电子设备600操作所需的各种程序和数据。处理装置601、ROM 602以及RAM 603通过总线604彼此相连。输入/输出(I/O)接口605也连接至总线604。
通常，以下装置可以连接至I/O接口605：包括例如触摸屏、触摸板、键盘、鼠标、摄像头、麦克风、加速度计、陀螺仪等的输入装置606；包括例如液晶显示器(LCD)、扬声器、振动器等的输出装置607；以及通信装置609。通信装置609可以允许电子设备600与其他设备 进行无线或有线通信以交换数据。虽然图6示出了具有各种装置的电子设备600，但是应理解的是，并不要求实施或具备所有示出的装置。可以替代地实施或具备更多或更少的装置。图6中示出的每个方框可以代表一个装置，也可以根据需要代表多个装置。
特别地，根据本公开的实施例，上文参考流程图描述的过程可以被实现为计算机软件程序。例如，本公开的实施例包括一种计算机程序产品，其包括承载在计算机可读介质上的计算机程序，该计算机程序包含用于执行流程图所示的方法的程序代码。在这样的实施例中，该计算机程序可以通过通信装置609从网络上被下载和安装，或者从存储装置608被安装，或者从ROM 602被安装。在该计算机程序被处理装置601执行时，执行本公开的实施例的方法中限定的上述功能。
需要说明的是，本公开的实施例所描述的计算机可读介质可以是计算机可读信号介质或者计算机可读存储介质或者是上述两者的任意组合。计算机可读存储介质例如可以是——但不限于——电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。计算机可读存储介质的更具体的例子可以包括但不限于：具有一个或多个导线的电连接、便携式计算机磁盘、硬盘、随机访问存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑磁盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。在本公开的实施例中，计算机可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。而在本公开的实施例中，计算机可读信号介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了计算机可读的程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。计算机可读信号介质还可以是计算机可读存储介质以外的任何计算机可读介质，该计算机可读信号介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。计算机可读介质上包含的程序代码可以用任何适当的介质传输，包括但不限于：电线、光缆、RF(射频)等等，或者上 述的任意合适的组合。
上述计算机可读介质可以是上述服务器中所包含的；也可以是单独存在，而未装配入该服务器中。上述计算机可读介质承载有一个或者多个程序，当上述一个或者多个程序被该服务器执行时，使得该服务器：获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；对于候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候选推送信息的第一匹配信息，其中，第一匹配信息用于表征基于第一位置信息，目标用户对该候选推送信息的偏好程度；根据第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，第一操作信息用于表征基于第一匹配信息，目标用户对该候选推送信息执行目标操作的概率；根据对应的第一操作信息，从候选推送信息集中选取候选推送信息；向目标用户对应的终端设备推送所选取的候选推送信息。
可以以一种或多种程序设计语言或其组合来编写用于执行本公开的实施例的操作的计算机程序代码，程序设计语言包括面向对象的程序设计语言一诸如Java、Smalltalk、C++，还包括常规的过程式程序设计语言一诸如″C″语言或类似的程序设计语言。程序代码可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或服务器上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络——包括局域网(LAN)或广域网(WAN)——连接到用户计算机，或者，可以连接到外部计算机(例如利用因特网服务提供商来通过因特网连接)。
附图中的流程图和框图，图示了按照本公开各种实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段、或代码的一部分，该模块、程序段、或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注意，在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个接连地表示的方框实际上可以基本并行地执行，它们有时 也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或操作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
描述于本公开的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。所描述的单元也可以设置在处理器中，例如，可以描述为：一种处理器包括获取单元、操作信息确定单元、选取单元和推送单元。其中，这些单元的名称在某种情况下并不构成对该单元本身的限定，例如，获取单元还可以被描述为″获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集的单元″。
以上描述仅为本公开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，本公开的实施例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上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。例如上述特征与本公开的实施例中公开的(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 (22)

  1. 一种用于推送信息的方法，包括：
    获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；
    对于所述候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候选推送信息的第一匹配信息，其中，所述第一匹配信息用于表征基于所述第一位置信息，所述目标用户对该候选推送信息的偏好程度；根据所述第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，所述第一操作信息用于表征基于所述第一匹配信息，所述目标用户对该候选推送信息执行目标操作的概率；
    根据对应的第一操作信息，从所述候选推送信息集中选取候选推送信息；
    向所述目标用户对应的终端设备推送所选取的候选推送信息。
  2. 根据权利要求1所述的方法，其中，所述方法还包括：
    根据所述第一位置信息，确定至少一个第二位置信息；以及
    所述确定步骤还包括：
    对于所述至少一个第二位置信息中的第二位置信息，确定该候选推送信息的第二匹配信息，其中，所述第二匹配信息用于表征基于该第二位置信息，所述目标用户对该候选推送信息的偏好程度；根据所述第二匹配信息，确定该候选推送信息对应的第二操作信息，其中，所述第二操作信息用于表征基于所述第二匹配信息，所述目标用户对该候选推送信息执行目标操作的概率；以及
    所述根据对应的第一操作信息，从所述候选推送信息集中选取候选推送信息，包括：
    根据对应的第一操作信息和至少一个第二操作信息，从所述候选推送信息集中选取候选推送信息。
  3. 根据权利要求2所述的方法，其中，所述根据所述第一位置信 息，确定至少一个第二位置信息，包括：
    根据所述第一位置信息和预设的至少一个距离阈值，确定至少一个第二位置信息，其中，距离阈值用于表征所述第一位置信息指示的位置与第二位置信息指示的位置之间的距离。
  4. 根据权利要求2所述的方法，其中，所述根据所述第一匹配信息，确定该候选推送信息对应的第一操作信息，包括：
    将所述第一匹配信息输入至预先训练的信息预测模型，得到该候选推送信息对应的第一操作信息，其中，所述信息预测模型用于表征匹配信息与操作信息之间的对应关系；以及
    所述根据所述第二匹配信息，确定该候选推送信息对应的第二操作信息，包括：
    将所述第二匹配信息输入至所述信息预测模型，得到该候选推送信息对应的第二操作信息。
  5. 根据权利要求1所述的方法，其中，所述第一匹配信息包括以下至少一项：用于表征所述目标用户对与所述第一位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与所述第一位置信息指示的位置的相关度的信息、用于表征所述第一位置信息指示的位置的关注度的信息。
  6. 根据权利要求2所述的方法，其中，所述第二匹配信息包括以下至少一项：用于表征所述目标用户对与该第二位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与该第二位置信息指示的位置的相关度的信息、用于表征该第二位置信息指示的位置的关注度的信息。
  7. 根据权利要求4所述的方法，其中，所述信息预测模型通过如下步骤训练得到：
    获取训练样本集，其中，所述训练样本集中的训练样本包括样本 匹配信息和样本指示信息，其中，样本匹配信息包括所述目标用户对应的历史推送信息的匹配信息，样本指示信息用于指示所述目标用户是否对对应的历史推送信息执行过所述目标操作；
    获取预先建立的初始模型；
    利用机器学习的方法，基于所述训练样本集和预设的损失函数，对所述初始模型进行训练，以及将训练后的初始模型确定为信息预测模型。
  8. 根据权利要求2所述的方法，其中，所述根据对应的第一操作信息和至少一个第二操作信息，从所述候选推送信息集中选取候选推送信息，包括：
    对于所述候选推送信息集中的候选推送信息，响应于确定该候选推送信息对应的目标概率大于预设的概率阈值，选取该候选推送信息，其中，所述目标概率用于表示该候选推送信息对应的概率集中的最大值或所述概率集中的各个概率的平均值，其中，所述概率集由该候选推送信息对应的第一操作信息表征的概率和对应的至少一个第二操作信息分别表征的概率组成。
  9. 根据权利要求1所述的方法，其中，所述获取候选推送信息集，包括：
    根据所述第一位置信息，获取所述候选推送信息集，其中，所述候选推送信息集中的候选推送信息与所述第一位置信息指示的位置相关。
  10. 根据权利要求1-9之一所述的方法，其中，所述向所述目标用户对应的终端设备推送所选取的候选推送信息，包括：
    响应于检测到所述目标用户的预设操作，向所述终端设备推送所选取的候选推送信息，以及控制在所述终端设备上显示所选取的候选推送信息，其中，所述预设操作用于请求推送与所述第一位置信息指示的位置相关的信息。
  11. 一种用于推送信息的装置，包括：
    获取单元，被配置成获取用于表示目标用户的位置的第一位置信息，以及获取候选推送信息集；
    操作信息确定单元，被配置成对于所述候选推送信息集中的候选推送信息，执行如下确定步骤：确定该候选推送信息的第一匹配信息，其中，所述第一匹配信息用于表征基于所述第一位置信息，所述目标用户对该候选推送信息的偏好程度；根据所述第一匹配信息，确定该候选推送信息对应的第一操作信息，其中，所述第一操作信息用于表征基于所述第一匹配信息，所述目标用户对该候选推送信息执行目标操作的概率；
    选取单元，被配置成根据对应的第一操作信息，从所述候选推送信息集中选取候选推送信息；
    推送单元，被配置成向所述目标用户对应的终端设备推送所选取的候选推送信息。
  12. 根据权利要求11所述的装置，其中，所述装置还包括：
    位置信息确定单元，被配置成根据所述第一位置信息，确定至少一个第二位置信息；以及
    所述操作信息确定单元进一步被配置成：
    对于所述至少一个第二位置信息中的第二位置信息，确定该候选推送信息的第二匹配信息，其中，所述第二匹配信息用于表征基于该第二位置信息，所述目标用户对该候选推送信息的偏好程度；根据所述第二匹配信息，确定该候选推送信息对应的第二操作信息，其中，所述第二操作信息用于表征基于所述第二匹配信息，所述目标用户对该候选推送信息执行目标操作的概率；以及
    所述选取单元进一步被配置成：
    根据对应的第一操作信息和至少一个第二操作信息，从所述候选推送信息集中选取候选推送信息。
  13. 根据权利要求12所述的装置，其中，所述位置信息确定单元进一步被配置成：
    根据所述第一位置信息和预设的至少一个距离阈值，确定至少一个第二位置信息，其中，距离阈值用于表征所述第一位置信息指示的位置与第二位置信息指示的位置之间的距离。
  14. 根据权利要求12所述的装置，其中，所述操作信息确定单元进一步被配置成：
    将所述第一匹配信息输入至预先训练的信息预测模型，得到该候选推送信息对应的第一操作信息，其中，所述信息预测模型用于表征匹配信息与操作信息之间的对应关系；以及
    所述操作信息确定单元进一步被配置成：
    将所述第二匹配信息输入至所述信息预测模型，得到该候选推送信息对应的第二操作信息。
  15. 根据权利要求11所述的装置，其中，所述第一匹配信息包括以下至少一项：用于表征所述目标用户对与所述第一位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与所述第一位置信息指示的位置的相关度的信息、用于表征所述第一位置信息指示的位置的关注度的信息。
  16. 根据权利要求12所述的装置，其中，所述第二匹配信息包括以下至少一项：用于表征所述目标用户对与该第二位置信息指示的位置相关的信息的偏好程度的信息、用于表征该候选推送信息与该第二位置信息指示的位置的相关度的信息、用于表征该第二位置信息指示的位置的关注度的信息。
  17. 根据权利要求14所述的装置，其中，所述信息预测模型通过如下步骤训练得到：
    获取训练样本集，其中，所述训练样本集中的训练样本包括样本 匹配信息和样本指示信息，其中，样本匹配信息包括所述目标用户对应的历史推送信息的匹配信息，样本指示信息用于指示所述目标用户是否对对应的历史推送信息执行过所述目标操作；
    获取预先建立的初始模型；
    利用机器学习的方法，基于所述训练样本集和预设的损失函数，对所述初始模型进行训练，以及将训练后的初始模型确定为信息预测模型。
  18. 根据权利要求12所述的装置，其中，所述选取单元进一步被配置成：
    对于所述候选推送信息集中的候选推送信息，响应于确定该候选推送信息对应的目标概率大于预设的概率阈值，选取该候选推送信息，其中，所述目标概率用于表示该候选推送信息对应的概率集中的最大值或所述概率集中的各个概率的平均值，其中，所述概率集由该候选推送信息对应的第一操作信息表征的概率和对应的至少一个第二操作信息分别表征的概率组成。
  19. 根据权利要求11所述的装置，其中，所述获取单元进一步被配置成：
    根据所述第一位置信息，获取所述候选推送信息集，其中，所述候选推送信息集中的候选推送信息与所述第一位置信息指示的位置相关。
  20. 根据权利要求11-19之一所述的装置，其中，所述推送单元进一步被配置成：
    响应于检测到所述目标用户的预设操作，向所述终端设备推送所选取的候选推送信息，以及控制在所述终端设备上显示所选取的候选推送信息，其中，所述预设操作用于请求推送与所述第一位置信息指示的位置相关的信息。
  21. 一种服务器，包括：
    一个或多个处理器；
    存储装置，其上存储有一个或多个程序；
    当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如权利要求1-10中任一所述的方法。
  22. 一种计算机可读介质，其上存储有计算机程序，其中，该程序被处理器执行时实现如权利要求1-10中任一所述的方法。
PCT/CN2019/126769 2019-04-02 2019-12-19 用于推送信息的方法和装置 WO2020199662A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910261259.4A CN109981785B (zh) 2019-04-02 2019-04-02 用于推送信息的方法和装置
CN201910261259.4 2019-04-02

Publications (1)

Publication Number Publication Date
WO2020199662A1 true WO2020199662A1 (zh) 2020-10-08

Family

ID=67082477

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/126769 WO2020199662A1 (zh) 2019-04-02 2019-12-19 用于推送信息的方法和装置

Country Status (2)

Country Link
CN (1) CN109981785B (zh)
WO (1) WO2020199662A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112307966A (zh) * 2020-10-30 2021-02-02 京东方科技集团股份有限公司 事件展示方法及装置、存储介质及电子设备
CN112819526A (zh) * 2021-01-29 2021-05-18 北京骑胜科技有限公司 待推送资源确定方法、装置、电子设备、介质及程序产品
CN112991008A (zh) * 2021-03-04 2021-06-18 北京嘀嘀无限科技发展有限公司 一种位置推荐方法、装置及电子设备
CN113192619A (zh) * 2021-05-26 2021-07-30 北京京东拓先科技有限公司 一种对象匹配方法、装置、设备及存储介质
CN113239175A (zh) * 2021-06-10 2021-08-10 中国平安人寿保险股份有限公司 一种显示候选语句列表的方法及终端设备
CN116474234A (zh) * 2023-03-14 2023-07-25 北京阿叟阿巴科技有限公司 一种基于视觉辅助的孤独症干预系统及方法

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109981785B (zh) * 2019-04-02 2020-08-14 北京字节跳动网络技术有限公司 用于推送信息的方法和装置
CN113793163B (zh) * 2020-11-06 2024-05-24 京东安联财产保险有限公司 信息推送方法、装置与电子设备
CN112839107B (zh) * 2021-02-25 2023-04-18 北京梧桐车联科技有限责任公司 推送内容的确定方法、装置、设备及计算机可读存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102215452A (zh) * 2011-05-30 2011-10-12 宇龙计算机通信科技(深圳)有限公司 基于地理位置的信息推送方法、服务器和终端
CN108229994A (zh) * 2016-12-21 2018-06-29 北京金山云网络技术有限公司 一种信息推送方法及装置
US20190014185A1 (en) * 2017-07-04 2019-01-10 Baidu Online Network Technology (Beijing) Co., Ltd Method and apparatus for pushing information
CN109460513A (zh) * 2018-10-31 2019-03-12 北京字节跳动网络技术有限公司 用于生成点击率预测模型的方法和装置
CN109981785A (zh) * 2019-04-02 2019-07-05 北京字节跳动网络技术有限公司 用于推送信息的方法和装置

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104703125B (zh) * 2013-12-05 2020-11-13 腾讯科技（深圳）有限公司 基于即时通信的信息推荐方法、装置和终端
EP3295411A1 (en) * 2015-09-09 2018-03-21 Google LLC Systems and methods for providing content
CN107203894B (zh) * 2016-03-18 2021-01-01 百度在线网络技术（北京）有限公司 信息推送方法和装置
CN107305551A (zh) * 2016-04-18 2017-10-31 百度在线网络技术（北京）有限公司 推送信息的方法和装置
CN108629608B (zh) * 2017-03-22 2023-02-24 腾讯科技（深圳）有限公司 用户数据处理方法及装置
CN109257398A (zh) * 2017-07-12 2019-01-22 阿里巴巴集团控股有限公司 一种业务对象的推送方法及设备
CN109167816B (zh) * 2018-08-03 2021-11-16 广州虎牙信息科技有限公司 信息推送方法、装置、设备和存储介质
CN109086439B (zh) * 2018-08-15 2022-02-25 腾讯科技（深圳）有限公司 信息推荐方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102215452A (zh) * 2011-05-30 2011-10-12 宇龙计算机通信科技(深圳)有限公司 基于地理位置的信息推送方法、服务器和终端
CN108229994A (zh) * 2016-12-21 2018-06-29 北京金山云网络技术有限公司 一种信息推送方法及装置
US20190014185A1 (en) * 2017-07-04 2019-01-10 Baidu Online Network Technology (Beijing) Co., Ltd Method and apparatus for pushing information
CN109460513A (zh) * 2018-10-31 2019-03-12 北京字节跳动网络技术有限公司 用于生成点击率预测模型的方法和装置
CN109981785A (zh) * 2019-04-02 2019-07-05 北京字节跳动网络技术有限公司 用于推送信息的方法和装置

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112307966A (zh) * 2020-10-30 2021-02-02 京东方科技集团股份有限公司 事件展示方法及装置、存储介质及电子设备
CN112819526A (zh) * 2021-01-29 2021-05-18 北京骑胜科技有限公司 待推送资源确定方法、装置、电子设备、介质及程序产品
CN112991008A (zh) * 2021-03-04 2021-06-18 北京嘀嘀无限科技发展有限公司 一种位置推荐方法、装置及电子设备
CN113192619A (zh) * 2021-05-26 2021-07-30 北京京东拓先科技有限公司 一种对象匹配方法、装置、设备及存储介质
CN113239175A (zh) * 2021-06-10 2021-08-10 中国平安人寿保险股份有限公司 一种显示候选语句列表的方法及终端设备
CN116474234A (zh) * 2023-03-14 2023-07-25 北京阿叟阿巴科技有限公司 一种基于视觉辅助的孤独症干预系统及方法
CN116474234B (zh) * 2023-03-14 2024-02-02 北京阿叟阿巴科技有限公司 一种基于视觉辅助的孤独症干预系统及方法

Also Published As

Publication number Publication date
CN109981785B (zh) 2020-08-14
CN109981785A (zh) 2019-07-05

Similar Documents

Publication Publication Date Title
WO2020199662A1 (zh) 用于推送信息的方法和装置
CN109460513B (zh) 用于生成点击率预测模型的方法和装置
TWI790443B (zh) 用於機器學習模型建立之技術
US20150032739A1 (en) System and method for predicting a geographic origin of content and accuracy of geotags related to content obtained from social media and other content providers
WO2020156236A1 (zh) 关联信息的推送方法、装置，电子设备及存储介质
US10976904B2 (en) Graphical chatbot interface facilitating user-chatbot interaction
JP6911603B2 (ja) ユーザによって訪問される施設のカテゴリの予測モデルを生成する方法、プログラム、サーバ装置、及び処理装置
TW201214173A (en) Methods and apparatus for displaying content
CN110704751A (zh) 数据处理方法、装置、电子设备及存储介质
US11281992B2 (en) Predicting geofence performance for optimized location based services
CN114217948A (zh) 分布式存储系统中的性能监测
JP2016045504A (ja) 曖昧性を解消する教師データの生成方法、生成システム、及び生成プログラム
WO2017016122A1 (zh) 一种信息推送方法及装置
US10929491B2 (en) Social sharing path user interface insights
CN111104599B (zh) 用于输出信息的方法和装置
WO2020199659A1 (zh) 用于确定推送优先级信息的方法和装置
CN112182281B (zh) 一种音频推荐方法、装置及存储介质
CN112328889A (zh) 推荐搜索词确定方法、装置、可读介质及电子设备
US20230315990A1 (en) Text detection method and apparatus, electronic device, and storage medium
US20180032945A1 (en) Effective utilization of idle cycles of users
WO2017024684A1 (zh) 用户行为意图的获取方法、装置、设备及非易失性计算机存储介质
US9760933B1 (en) Interactive shopping advisor for refinancing product queries
US10169474B2 (en) Mobile application discovery using an electronic map
CN111026849A (zh) 数据处理方法和装置
US11734371B2 (en) Multi-sensory content authorship aid

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: 19922723

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 04.02.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19922723

Country of ref document: EP

Kind code of ref document: A1