WO2019120037A1

WO2019120037A1 - Model construction method, network resource preloading method and apparatus, medium, and terminal

Info

Publication number: WO2019120037A1
Application number: PCT/CN2018/117157
Authority: WO
Inventors: 陈岩; 刘耀勇
Original assignee: Oppo广东移动通信有限公司
Priority date: 2017-12-18
Filing date: 2018-11-23
Publication date: 2019-06-27
Also published as: CN108134691B; CN108134691A

Abstract

Disclosed are a model construction method, a network resource preloading method and apparatus, a medium, and a terminal. The model construction method comprises: obtaining a historical network access record and the terminal state associated with same; determining a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state; and constructing an access prediction model on the basis of the state transition probability matrix and the observation probability matrix.

Description

Model construction method, network resource preloading method, device, medium and terminal

The present application claims the priority of the Chinese Patent Application No. PCT Application No.

Technical field

The embodiments of the present application relate to data processing technologies, for example, to a model construction method, a network resource preloading method, an apparatus, a medium, and a terminal.

Background technique

At present, mobile terminals provide communication services, living services, entertainment services, and the like for more and more users. For example, a user can install a news application on a mobile terminal to view news information through such an application.

An application installed on a mobile terminal provides users with the information they need by accessing network resources. However, because the network loading process takes a long time, the user needs to wait for the network data to load, and the application performance is not good, thereby affecting the user's viscosity of the application. Take the news application as an example. The news interface includes not only text content, but also pictures or short videos. When the mobile terminal detects that the user clicks on a news title to the final display of the news interface corresponding to the news title, the network data is downloaded by the set resource resource address, that is, the Uniform Resource Locator (URL). The time is as low as tens of milliseconds (in a network-free environment), hundreds of milliseconds or even seconds, and the user experience is not good.

Summary of the invention

The following is an overview of the topics detailed in this document. This Summary is not intended to limit the scope of the claims.

The embodiment of the present application provides a model construction method, a network resource preloading method, a device, a medium, and a terminal, which can shorten the loading time of network data.

In a first aspect, the embodiment of the present application provides a model construction method, including: acquiring a historical network access record and a terminal status associated with the historical network access record; determining a status according to the historical network access record and the terminal status The transition probability matrix and the observation probability matrix are constructed; the access prediction model is constructed based on the state transition probability matrix and the observation probability matrix.

In a second aspect, the embodiment of the present application further provides a network resource preloading method, including: acquiring a current network access record and a terminal status associated with a current network access record according to a set period; and the current network access record and the location Entering a pre-built access prediction model, wherein the access prediction model is a model of terminal state training corresponding to the historical network access record and the historical network access record; and the access prediction model output is to be accessed. The network resource identifier is used to download network data corresponding to the network resource identifier, and is stored in a local cache.

In a third aspect, the embodiment of the present application further provides a model construction apparatus, where the apparatus includes: a history acquisition module, configured to acquire a historical network access record and a terminal status associated with the historical network access record; and a matrix determination module, The method is configured to determine a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state; and the model building module is configured to construct an access prediction model based on the state transition probability matrix and the observation probability matrix.

In a fourth aspect, the embodiment of the present application further provides a network resource preloading device, where the device includes: a current record obtaining module, configured to acquire a current network access record according to a set period and associated with the current network access record. a terminal input module configured to input the current network access record and the terminal state into a pre-built access prediction model, wherein the access prediction model is based on a historical network access record and a terminal corresponding to the historical network access record a model of the state training; the preloading module is configured to acquire the network resource identifier to be accessed output by the access prediction model, and download the network data corresponding to the network resource identifier, and store the data in the local cache.

In a fifth aspect, the embodiment of the present application further provides a computer readable storage medium, where the computer program is stored, and when the program is executed by the processor, the model construction method according to the first aspect is implemented; or the program The network resource preloading method as described in the second aspect above is implemented when executed by the processor.

In a sixth aspect, the embodiment of the present application further provides a terminal, including a memory, a processor, and a computer program stored on the memory and operable on the processor, where the processor implements the computer program to implement the first aspect as described above The model construction method; or the network resource preloading method according to the second aspect described above is implemented when the processor executes the computer program.

Other aspects will be apparent upon reading and understanding the drawings and detailed description.

DRAWINGS

1 is a flowchart of a method for constructing a model provided by an embodiment of the present application;

2 is a flowchart of another method for constructing a model provided by an embodiment of the present application;

3 is a flowchart of a network resource preloading method provided by an embodiment of the present application;

4 is a flowchart of another network resource preloading method provided by an embodiment of the present application;

FIG. 5 is a schematic diagram of a network data loading process according to an embodiment of the present application; FIG.

6 is a structural block diagram of a model construction apparatus according to an embodiment of the present application;

FIG. 7 is a structural block diagram of a network resource preloading apparatus according to an embodiment of the present application;

FIG. 8 is a structural block diagram of a terminal according to an embodiment of the present disclosure;

FIG. 9 is a schematic structural diagram of a mobile terminal according to an embodiment of the present application.

Detailed ways

The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the application and are not intended to be limiting. In addition, it should be noted that, for the convenience of description, only some but not all of the structures related to the present application are shown in the drawings.

Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as a process or method depicted as a flowchart. Although the flowcharts describe the various steps as a sequential process, many of the steps can be implemented in parallel, concurrently, or concurrently. In addition, the order of the steps can be rearranged. The process may be terminated when its operation is completed, but may also have additional steps not included in the figures. The processing may correspond to methods, functions, procedures, subroutines, subroutines, and the like.

FIG. 1 is a flowchart of a method for constructing a model according to an embodiment of the present disclosure. The method may be implemented by a model building device, where the device may be implemented by at least one of software and hardware, and may be integrated into a terminal. For example in a mobile terminal. As shown in FIG. 1, the method includes steps 110 to 130.

In step 110, a historical network access record and a terminal status associated with the historical network access record are obtained.

The terminal may be a mobile terminal equipped with an operating system, including but not limited to a smart phone, a tablet computer, a handheld game console, a notebook computer, and a smart watch.

The historical network access record may be a network access record of an application accessing the network behavior in the local machine in the preset time interval, and the network access record includes but is not limited to a network resource address (ie, a URL address), and a timestamp for opening the URL address. (ie access time) and the application ID of the application accessing the URL address. The terminal status includes, but is not limited to, a network environment status, a charging status, remaining power, and location information. It should be noted that the terminal status in the embodiment of the present application is a network environment status, a charging status, and the like when the application accesses the URL address.

When an application accesses the network, the mobile terminal records the network access behavior as a historical network access record. At the same time, when the network access behavior is recorded, the state of the mobile terminal at the time when the corresponding application accesses the network can also be obtained. For example, when an application accesses the network, network access information such as a URL address, an application that issues a network access request, and a timestamp of opening a URL are recorded. It is also possible to collect terminal status information such as the network environment and the charging status of the mobile terminal at this time. The network environment may be a wireless network. For example, whether to connect WIreless-FIdelity (WIFI) or the like. The mobile terminal can store the historical network access record and the terminal status in the form of a data table, which is stored in the mobile terminal database.

Table 1. Network access record and data table of mobile terminal status associated with the historical network access record

It should be noted that one row in the above table may be recorded as a status. The features listed in the above table include WIFI, an application, a URL, a timestamp for opening the URL, and a charging status, but are not limited to the above features, and may also be based on The actual model construction needs to increase the remaining power, location information and other characteristics. The timestamp to open this URL can be converted to the year, month, day, hour, minute, and second format. For example, 1497590695469=2017/6/16 13:24:55.

In step 120, a state transition probability matrix and an observation probability matrix are determined according to the historical network access record and the terminal state.

It should be noted that the mobile terminal needs to preprocess the historical network access record and the terminal state in the above table to obtain a state set and an observation set. The status set may be determined according to the WIFI, the application, the timestamp of the open URL, and the charging status in Table 1. The observation set may be determined according to the URL.

It should be noted that the state transition probability can be regarded as the probability that the state q _{i is} transferred to q _j in the state set, that is, the state transition probability can be understood as the state corresponding to the previous moment in the two adjacent moments is q _i and The state corresponding to the next moment is the probability of q _j .

In the embodiment of the present application, the manner of calculating the state transition probability of each state in the state set is a numerator as the number of times the state q _{i is} transferred to q _j in the state set (it is required that q _i may be the same as q _j The state may be a different state. The division is performed by using the number of occurrences of the state q _i in the state set as a denominator. It should be noted that the state set can be represented in the form of a state matrix. Illustratively, the state transition probability for each state in the set of states can be calculated using the following formula:

Where q _i represents the current state of the current time, q _j represents the adjacent state at the next moment, and P(i _t+1 =q _j |i _t =q _i ) indicates that the state in the state set is transferred from the current state to The probability of the adjacent state, N(i _t =q _i , i _i+1 =q _j ), represents the number of occurrences of the state in the state set being q _i and the neighboring state being q _j , N(i _t =q _i ) represents the number of times the current state in the set of states is q _i .

It should be noted that the state transition probability between two adjacent moments in the state set may be respectively calculated by using the above formula, and the state transition probability matrix is formed according to the state transition probability. For example, the state transition probability matrix may include a state transition state occurring at a next moment or a state transition probability of a current state, and a state transition probability at which a state occurring at a next moment is a remaining state in the state matrix. That is, if all possible sets of states are represented by Q, ie Q = {q ₁ , q ₂ , q ₃ , ..., q _n }, where n represents the number of states, then the state transition probability matrix A can represent A = [a _ij ] _{n × n} can be used to record the probability of a jump between all states, including a state transition to itself or a transition to a state in the state set.

It can be understood that the state transition probability may exist in the form of a transition transition probability matrix as shown in the above example, and may also exist and construct an access prediction model in other ways. For example, the state transition probability table may be stored in association with the state transition probability; or the state transition probability may be stored in the first table, the state may be stored in the second table, and the mapping between the first table and the second table may be established. Wait.

It should be noted that the observation probability can be regarded as the probability of generating a preset observation value in a certain state, that is, if a certain state is represented by q _x , the manner of determining the observation value corresponding to the state may be, times the number of occurrences of the observation state q _x k values divided by the occurrence of a state is obtained in a state q _x q _x k generated observation probability of the observed values. Exemplarily, one state in the state set may be sequentially acquired as the current state, and the number of occurrences of the current state may be counted. According to the observation value corresponding to the current state and the number of occurrences of the current state, the following formula is used to calculate the observation probability of the current state:

Wherein, the observed values k ∈ {v ₁ , v ₂ , . . . , v _m }, {v ₁ , v ₂ , . . . , v _m } are observation sets,

Indicates the probability that the observation value k is generated in the current state q _x , and N(i _t =q _x , v=k) represents the number of times the observation value k is generated in the current state q _x .

In step 130, an access prediction model is constructed based on the state transition probability matrix and the observation probability matrix.

It should be noted that the access prediction model in the embodiment of the present application may be a probability model. The access prediction model includes, but is not limited to, a hidden Markov model. The hidden Markov model describes a hidden sequence and a corresponding observable sequence, called a sequence of states, which can be represented by a set of states. This observable sequence is called an observation sequence and can be represented by an observation set.

Exemplarily, if Q represents a state set and V represents an observation set, then Q={q ₁ , q ₂ , q ₃ , . . . , q _n }, V={v ₁ , v ₂ , v ₃ , ..., v _m }, where n is the range of values for the number of states and m is the range of values for the observed values. Since A represents the state transition probability matrix, there is A=[a _ij ] _n×n , where 1≤i≤n, 1≤j≤n, a _ij =P(i _t+1 =q _j |i _t = q _i ), a _ij represents the probability of transitioning from state q _i to q _j . The observation probability matrix is a set of observation probabilities, that is, if the observation probability matrix is represented by B, the observation probability matrix can be expressed as

Thus, the access prediction model can be constructed by the state set Q, the observation set V, the state transition probability matrix A, and the observation probability matrix B, that is, the access prediction model can be expressed as [Q, V, A, B].

The technical solution of the embodiment obtains a historical network access record and a terminal state associated with the historical network access record; determines a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state; and based on the state transition probability matrix And the observation probability matrix constructs an access prediction model; thus, the observation probability of the observation value corresponding to the next state can be determined by accessing the prediction model based on the current state, since the observation value represents the network resource address, and thus, the user has predicted before accessing the network resource The target network resources are improved, and the intelligence of the mobile terminal is improved.

FIG. 2 is a flowchart of another method for constructing a model according to an embodiment of the present application. As shown in FIG. 2, the method includes steps 201 to 210.

In step 201, a historical network access record and a terminal status associated with the historical network access record are obtained.

The historical network access record includes an application for accessing the network, a network resource address, and an access time; and the terminal status includes network environment information and a charging status.

It should be noted that the collection of the original data (ie, the historical network access record) is completed on the mobile terminal. For example, real-time access to the network access record of the mobile terminal and the status of the terminal when the network access behavior occurs, including an application accessing the network, a network resource address (ie, a URL address), a URL access time, and a WIFI status when the application accesses the network, In the charging state, etc., the network access record and the terminal state are stored in a preset database of the mobile terminal, and the storage form can be as shown in Table 1.

In step 202, the program number of the application that accesses the network and the network number corresponding to the network resource address are matched according to a preset rule.

The preset rule may be that the application number is in a preset order, and the network resource address number is in a preset order. The preset order can be chronological, or other user-specified order.

For example, take the data in Table 1 as an example: applications that access the network include applications such as QQ, WeChat, and Today's headlines, assigning a number to each application and replacing the application with a number. It should be noted that different applications are numbered, and the same applications have the same number. For example, the program number assigned to QQ is 0, the program number assigned to WeChat is 1, and the program number assigned to today's headline is 2, ....., and the maximum number of program numbers depends on the number of applications in Table 1. The application may correspond to multiple network resource addresses and assign different network numbers to different network resource addresses. Different applications may access the same network resource address, and the same network resource address has the same network number. Each network accessed by the application is assigned a network number, ie the network number k∈[0,1,2...], and the maximum network number depends on how many different URLs have appeared in Table 1.

In step 203, the working days and the rest days are divided according to the access time, and different date numbers are assigned to the working days and the rest days, respectively.

It should be noted that the access time is the timestamp of opening the URL in Table 1, and the timestamp can be converted into a year, month, day, hour, minute, and second format, so that the time for opening the URL address is determined to be a working day or a rest according to the access time. day. You can set the timestamp of opening the URL to 1 in the case of opening the URL. If the time to open the URL is 1 day, the timestamp of opening the URL in Table 1 is the value of the timestamp. Is 0.

In step 204, the time number is determined according to the time period to which the access time belongs.

It should be noted that it is necessary to divide the preset time interval in the natural day in advance to obtain the time period, and the time period is stored in association with the time number. For example, 24 hours in the natural day are divided into several time periods in advance. For example, if the time interval is 10 minutes, then a natural day has 24*60/10=144 time segments for 24 hours, and the time segments are sequentially numbered, and the time number corresponding to the access time is t∈[0,1,2 , 3...143]. Since the time period has been numbered, the access time of the terminal to access the URL may be determined according to the timestamp of opening the URL in any one of the records in Table 1, thereby, according to the time period to which the access time belongs and the time period and time stored in advance association. The number can be used to determine the time number corresponding to the record.

It can be understood that, since the user cannot use the mobile terminal 24 hours a day, the time interval in which the user uses the mobile terminal can be divided according to the usage habit of the user. For example, if the user is in a sleep state between 12 am and 6 am, and the mobile terminal is not used, the time interval outside the rest time interval can be divided to obtain a time period.

In step 205, it is determined whether the terminal accesses the wireless network according to the network environment information, and determines a network state value according to the determination result.

The network environment information includes information such as whether the terminal accesses the network or not, and how to access the network, including but not limited to whether to access the WIFI.

It may be pre-defined that if the mobile terminal accesses the wireless network, the network status value is set to 1, and if the mobile terminal does not access the wireless network, the network status value is set to zero. For example, if the WIFI of the mobile terminal is turned on and the wireless network is connected, the network status value is set to 1, and if the WIFI of the mobile terminal is turned off, the network status value is set to 0.

In step 206, it is judged whether the terminal is charging according to the state of charge, and the state of charge value is determined according to the determination result.

It can be pre-specified that if the mobile terminal is in the charging state, the charging state value is set to 1, and if the mobile terminal is not in the charging state, the charging state value is 0. The charging state value when the application accesses the URL is determined according to whether the mobile terminal is charging.

The following table can be obtained based on the above-described preprocessing of historical network access records and terminal status.

Table 2, sample collection table

WIFIWIFI	应用程序application	是否工作日Whether working day	时间段period	充电状态charging	URL(网址)URL (URL)
11	00	11	8181	11	00
00	11	11	8282	00	11
…...	…...	…...	…...	…...	…...
…...	…...	…...	…...	…...	…...

It should be noted that each row in Table 2 represents a state.

In step 207, a state set is generated based on the program number, date number, time number, network state value, and state of charge value, and an observation set is constructed based on the network number.

It should be noted that the program number, the date number, the time number, the network status value, and the charging status value constitute a state set, and the network number constitutes an observation set.

Exemplarily, taking the sample set in Table 2 as an example, the state set Q is composed of the first five columns of corresponding feature values, and the observation set V is composed of the last column of URLs. Since at least two weeks of historical network access records and corresponding terminal states are acquired during model construction, the records in the state matrix corresponding to the state set are highly random, not necessarily chronologically arranged. In an embodiment, the state set Q = {q ₁ = [0, 0, _0, 0, 0], q ₂ = [0, 0, 0, 0, 1], ..., q _i = [1 ,0,1,81,1],q _j =[0,1,1,82,0],...,q _n }, observation set V={0,1,2,...,v _m }. If the state set and the observation set are represented in a matrix form, they can be as follows:

as well as,

The number of records of Q can be expressed as 2*2*144*2*, the number of applications accessing the network on the terminal, and the size of V is the number of URLs visited.

In step 208, a state transition probability for each state in the set of states is calculated, and a state transition probability matrix is generated based on the state transition probability.

It should be noted that the state transition probability in the embodiment of the present application calculates the current state and the transition probability of the state in which the next row may occur, that is, the current state and the next state appear in pairs in the state set, and the next state The next line in the row where the current state is located. For example, in some states, the current state is q _i =[1,0,1,81,1], and the next state is q _j =[0,1,1,82,0], while in others In the state, the current state is q _i =[1,0,1,81,1], but the next state q _j =[0,0,0,0,1], etc., the two q listed above can be _{j is} considered to be the next next state of the current state. Since each user has their own network access habits, if a user visits some URLs at a certain time of the day, the probability of the user accessing the same URL in the same time period is higher in the sampling period of at least two weeks. Therefore, there is a recurring state in the state matrix. Exemplarily, if the state transition probability of the current state q _i to the next state q _j is to be calculated, the state of the state matrix Q in which the previous row is q _i and the next row is q _j may be counted, and The number of occurrences of the q _i state in the above state matrix Q is also counted. Thus, the result of the state in which the upper row of the state matrix Q is q _i and the next row is q _j is divided by the number of q _i states is taken as the state transition probability of jumping from the current state q _i to the next state q _j . It should be noted that q _j includes the remaining state of removing q _i from the state matrix, and also includes the state q _i itself. The state transition probability is calculated for each state in the state matrix, and a state transition probability matrix of n*n is obtained, which may be [a _ij ] _n×n described above.

In step 209, an observation probability of an observation value corresponding to each state is calculated according to the state set and the observation set, and an observation probability matrix is generated according to the observation probability.

In the embodiment of the present application, a row state in the state matrix is sequentially acquired as a current state, an observation value corresponding to the current state is acquired, and the number of occurrences of the same observation value N (i _t =q _x , v=k) is counted, wherein , x ∈ [1, 2, 3, ..., n], k ∈ [1, 2, 3, ..., v _m ]. Then, N (i _t =q _x , v=k) is divided by the operation result of the current state of the state matrix in the state matrix as the observation probability of the observation value k generated by the current state.

According to the above manner, each state in the state matrix is taken as the current state, and the probability of generating the observation value in the observation set in the current state is respectively determined to constitute the observation probability matrix B of n*m.

In step 210, an access prediction model is constructed.

In the embodiment of the present application, the access prediction model is a hidden Markov model. According to the state set, the observation set, the state transition probability matrix A and the observation probability matrix B, the hidden Markov model λ can be constructed, and the access prediction model can be expressed as λ=[Q, V, A, B]. The access prediction model is built in the terminal. Within the setting, it is set to predict in advance which resources corresponding to the URL need to be downloaded in advance.

It should be noted that, since the randomness of the user accessing the network resource by the application on the mobile terminal is very large, the sample set for predicting the network resource address that the mobile terminal may access in the future is not static, and the model may be set according to actual needs. The condition is updated so that the access prediction model is updated when the update condition is met. Exemplarily, the update time is set. When the system time is detected to satisfy the update time, the historical network access behavior and the mobile terminal state information in the preset time interval are acquired, and the access prediction model is updated. Specifically, the update period may be set, and when the update period is reached, it is determined that the model update condition is satisfied. For example, you can set a week or day as the update cycle. If the model is updated once a week, when it is detected that the update interval is satisfied, the network behavior record and the terminal state of the mobile terminal in the latest week are obtained by means of incremental update. It can be understood that the update date can also be set, for example, Monday or every other week of Monday. The sample update operation may also be triggered by a user, for example, the user inputs a model update indication, performs a model update operation based on the update indication, and the like.

The technical solution of the embodiment, by learning the historical network access behavior of the terminal and the terminal state associated with the historical network access behavior, constructs an access prediction model, improves the adaptive capability of the model, and has good applicability to different users.

3 is a flowchart of a network resource preloading method provided by an embodiment of the present application, where the method may be implemented by a network resource preloading device, where the device may be implemented by at least one of software and hardware, and generally integrated. In the terminal, for example in a mobile terminal. As shown in FIG. 3, the method includes steps 310 to 330.

In step 310, the current network access record and the terminal status associated with the current network access record are obtained according to a set period.

It should be noted that the current network access record includes the application that accesses the network at the current time, the network resource address (ie, the URL address), and the timestamp at which the URL is opened. The terminal status associated with the current network access record includes the current network environment information and the current charging status, including but not limited to whether the monitoring terminal is currently connected to the WIFI, and the monitoring terminal recognizes that the charging is performed.

It should be noted that the setting period may be the default of the system, or may be a time interval set by the user for invoking the access prediction model to predict the next state according to the current state. For example, the access prediction model may be set to be called every 10 minutes to predict the network resource address that the next state may access according to the current state, and pre-download the network data associated with the predicted network resource address.

It can be understood that after the current network access record and its associated terminal status are obtained, the current network access record and the terminal status need to be pre-processed. The pre-processing manner is the same as that in the foregoing embodiment, and is no longer As a result, get the current state that matches the input data format of the access prediction model.

In step 320, the current network access record and terminal status are entered into a pre-built access prediction model.

The access prediction model is a model of terminal state training corresponding to the historical network access record and the historical network access record. It can be understood that the access prediction model herein may be a model constructed by the scheme described in the above embodiment.

In this embodiment, since the current network access record and the terminal state are preprocessed to obtain the current state, the current state is directly outputted to the access prediction model, so that the access prediction model predicts the next state that may occur at the next moment. And, respectively, an observation probability of each observation value (ie, a network resource identifier, and the network resource identifier may include a network resource address) corresponding to the next state is calculated, and the observation probability may be considered as an output of the access prediction model. Exemplarily, the current state q _{i is} input into the access prediction model, and the state transition probability of all the next states corresponding to the current state q _{i is} determined according to the state transition probability matrix, so that the state with the highest state transition probability is the most The next state q _j that may appear. Then, the reference observation probability corresponding to q _{j is} determined according to q _j and the observation probability matrix B. The reference observation probability is arranged in descending order, and the target observation probability of the set number prior to the ranking is selected by the reference observation probability, and the network resource identifier corresponding to the target observation probability is used as the network resource identifier most likely to be accessed in the next state, And output.

In step 330, the network resource identifier to be accessed that is output by the access prediction model is obtained, and the network data corresponding to the network resource identifier is downloaded and stored in the local cache.

It should be noted that the network resource identifier may be a network resource address. For example, if the set number is 100, the 100 network resource addresses corresponding to the next state output by the access prediction model are obtained, and the 100 network resource addresses are considered to be preloaded.

It can be understood that after the network resource address is predicted, if the network data corresponding to the network resource address has been loaded, the network data does not need to be repeatedly loaded to avoid excessive occupation of the cache. Exemplarily, after predicting the foregoing 100 network resource addresses, it is determined whether the network data corresponding to the network resource address is predicted to be preloaded; if it is predicted that the network data corresponding to the network resource address has been preloaded, then discarding Performing a download operation for the network data, if it is predicted that the network data corresponding to the network resource address is not preloaded, download the network data corresponding to the network resource identifier, and store the data in the local cache.

It should be noted that the method for downloading the network data corresponding to the network resource identifier may be that a communication connection is established between the server corresponding to the network resource address through the Internet, and the network data corresponding to the network resource address is downloaded by the server. It can be understood that the manner of downloading the network data corresponding to the network resource identifier is not limited to the manner enumerated in the above examples.

The technical solution of the embodiment obtains the current network access record and the terminal status associated with the current network access record according to the set period, and inputs the current network access record and the terminal status into the access prediction model to obtain a to-be-accessed The network resource identifier is configured to download the network data corresponding to the network resource identifier, and store the network data in the local cache, and preload the network data corresponding to the network resource identifier that may be accessed before acquiring the network data according to the identifier of the network resource, The network data is read in the local cache to shorten the loading time of the network data, thereby avoiding the situation that the loading time is long caused by loading data from the external network.

It should be noted that the pre-loaded network data is cached in the local cache, and the network data corresponding to the network access request may be directly downloaded from the local cache when the application obtains the network access request. Among them, the application includes an application with network access capabilities. For example, a news application, an instant messaging application, or a multimedia application. Get a network access request for the application. The network access request may be understood as a message that the application requests the network corresponding to the specified network resource address (ie, the URL) to obtain the network data. The network access request can be generated by the application and reported to the system network function module. When the system network function module obtains the message requesting to obtain the network data of the specified URL, the system knows that the application needs to acquire the network resource. It should be noted that the system network function module can be understood as the code and protocol stack of the system to implement the network access function.

Then, the local cache is queried according to the network access request, and a preload state of the network data corresponding to the network access request is determined. Exemplarily, the network access request includes a network resource address (URL address), and when the network access request of the application is obtained, the network resource address is extracted. The local cache is queried according to the network resource address, and it is determined whether the network data corresponding to the network resource address exists. If the network data corresponding to the network resource address does not exist, it is determined that the network data corresponding to the network access request is not preloaded, the network access process is executed, and the network data corresponding to the network access request is downloaded by the external network. If the network data corresponding to the network resource address exists, it is determined that the network data corresponding to the network access request is preloaded, and the network data is read by the local cache, and is directly returned to the application that sends the network access request to prevent the mobile terminal from passing. The Internet downloads network data from the server, which shortens the time for loading network data.

FIG. 4 is a flowchart of another network resource preloading method provided by an embodiment of the present application. As shown in FIG. 4, the method includes steps 401 to 416.

In step 401, the current network access record and the terminal status associated with the current network access record are obtained according to a set period.

The setting period may be the default of the system, or may be set by the user according to the actual needs, and is used to trigger the operation of predicting the network resource identifier that may be accessed at the next moment according to the current time network access record and the terminal state.

In step 402, the current network access record and the terminal status are preprocessed to obtain a current status.

In step 403, the current state is input into a pre-built access prediction model to predict a next state corresponding to the next moment by using the access prediction model, and respectively determine an observation of a network resource address corresponding to the next state. Probability.

In step 404, the network resource address to be accessed corresponding to the next state output by the access prediction model is obtained.

In step 405, it is determined whether the terminal accesses the wireless network. If the terminal accesses the wireless network, step 406 is performed. If the terminal does not access the wireless network, step 407 is performed.

The wireless network in the embodiment of the present application includes but is not limited to WIFI. That is to say, after obtaining the network resource address outputted by the access prediction model, the network state of the terminal can be obtained, and whether the terminal is connected to the WIFI is determined.

In step 406, the network data corresponding to the network resource identifier is downloaded and stored in a local cache.

It can be understood that, in some embodiments, after the execution of this step is completed, step 413 may be skipped. In some embodiments, after the step is performed, the process may go to step 410, where the step 410 is performed to obtain the first network included in the network resource identifier through the network crawler after downloading the network data corresponding to the network resource identifier. The second network resource address of the resource address loads the network data corresponding to the second network resource address, and is stored in the local cache.

In step 407, status information of the cellular mobile network data of the terminal is obtained.

The status information in this embodiment includes, but is not limited to, a switch state of a cellular mobile network data switch, a daily traffic upper limit, a current used traffic, and a remaining traffic state.

In step 408, it is determined whether the state information meets the preset condition. If the state information meets the preset condition, step 406 is performed. If the state information does not meet the preset condition, step 409 is performed.

It should be noted that the preset condition may be that the cellular mobile network data switch is in an open state and the remaining traffic exceeds a preset traffic threshold. In an embodiment, the preset condition may also be that the cellular mobile network data switch is in an open state and the currently used traffic does not reach a preset daily traffic upper limit or the like.

In step 409, the target network data is determined according to at least one of a data type and a data amount of the network data corresponding to the network resource identifier, and the target network data is loaded.

It should be noted that, when the state information of the cellular mobile network data does not meet the preset condition, the data type of the network data corresponding to the network resource identifier output by the access prediction model is acquired, where the data type includes but is not limited to text, Pictures, animations, videos and audio. In an embodiment, the data volume of the network data corresponding to the network resource identifier (that is, the size of the network data) may also be acquired. For example, the data amount of the network data D1 is 5 MB, the data amount of the network data D2 is 50 MB, and the like. The target network data may refer to network data that meets preset filtering conditions, wherein the filtering condition may be that network data whose data type is text and picture can be used as target network data. In an embodiment, the screening condition may also be that the network data whose data amount is less than the preset data amount threshold may be used as the target network data.

The data type of the network data corresponding to the obtained network resource identifier is matched with the preset screening condition. If the matching is successful, it is determined that the successfully matched network data is the target network data, and the target network data is loaded. In an embodiment, the data quantity of the network data corresponding to the acquired network resource identifier is matched with the preset screening condition. If the matching is successful, determining that the successfully matched network data is the target network data, and performing the target network data load. It can be understood that the above data types and data amounts can be combined for use in determining target network data, or can be used alone.

In step 410, the second network resource address linked to the first network resource address included in the network resource identifier is obtained by the network crawler, and the network data corresponding to the second network resource address is loaded and stored in the local cache.

It should be noted that the network data corresponding to some network resource addresses may include links of other network resource addresses, and these additional network resource addresses may be obtained through the network crawler, and data preloading may be performed. For example, when downloading the target network data, it is found that there is a second network resource address linked to the first network resource address (the second network resource address may not belong to the set of network resource identifiers output by the access prediction model), and may be downloaded together. The network data corresponding to the second network resource address. It should be noted that, when downloading the second network resource address, the network data actually needed to be downloaded may also be determined according to the preset screening conditions described in step 409.

It can be understood that this step is not a step that must be performed. In some embodiments, this step may not be performed after step 409 is performed.

In step 411, the storage space occupied by the network data in the local cache is obtained.

It can be understood that, because the network data is preloaded and stored in the local cache, as the amount of preloaded network data increases, the more local cache is occupied, which may affect the operating efficiency of the terminal. In order to avoid caching too much network data locally, a set policy can be used to clear part of the data in the local cache to reduce the amount of network data stored in the local cache. It should be noted that there are many ways to trigger the execution of the data clearing event, which is not specifically limited in the embodiment of the present application. For example, the size of the data amount of the network data in the local cache (that is, the size of the storage space) can be obtained according to a set period. For another example, the time for clearing the network data cache may be preset, such as clearing the locally cached network data from 0:00 to 12:00.

In step 412, when the storage space exceeds a set threshold, data to be cleared is determined according to at least one of a cache time and a use frequency of the network data, and the data to be cleared is cleared.

The setting threshold may be the system default, or may be set by the user according to actual needs. Exemplarily, if the total amount of cached data exceeds 100 MB, the set cache file is cleared. The setting cache file may be determined according to the cache time of the cache file. For example, the cache data whose cache time exceeds 1 hour may be considered as setting a cache file. In addition, setting the cache file can also be determined according to the frequency of use. For example, cached data whose frequency is less than 5 times/hour can be regarded as a set cache file. It should be noted that the above cache time and frequency of use can be combined to determine the set cache file, or can be used alone.

It should be noted that the execution order of step 411 and step 412 is not limited to the order described in the above example (ie, after step 410), and steps 411 and 412 are performed as long as it is detected that the data clear event is triggered.

In step 413, a network access request of the application is obtained.

It should be noted that the execution order of this step is not limited, and is triggered by the operation of the network access request input by the user in an application. That is, the user operation is monitored, and when the user inputs a network access request operation, the terminal acquires a network access request of the application.

In step 414, the local cache is queried according to the network access request, and a preload state of the network data corresponding to the network access request is determined.

The preload status includes loading to the local cache and not loading. Querying the local cache according to the network resource address included in the network access request, and determining whether the network data corresponding to the network resource address is loaded. If the network data corresponding to the network resource address is loaded, step 415 is performed, if the network resource address corresponds to The network data is not loaded, and step 416 is performed.

In step 415, the network data is read and returned to the application.

In step 416, the network data is downloaded by an external network.

The technical solution provided in this embodiment pre-judges the network environment of the terminal before downloading the network data corresponding to the network resource identifier. If the network environment is in a wireless network environment (for example, in a WIFI environment), the network data may be directly preloaded. When the network access request of the application input by the user is detected, the network data is directly read by the local cache to improve the data loading speed. In the cellular mobile network data environment, all or part of the network data is downloaded according to the setting policy, so as to simultaneously increase the data loading speed without causing excessive traffic consumption. In addition, in addition to the network data corresponding to the network resource identifier outputted by the access prediction model, the network data corresponding to other links included in the network data is also loaded, and the preloaded data is enriched to avoid data omission. At the same time, by formulating an appropriate cache data clearing mechanism, it is possible to avoid too much network data being cached locally and affect the execution efficiency of the terminal.

FIG. 5 is a schematic diagram of a network data loading process provided by an embodiment of the present application. As shown in FIG. 5, the network subsystem of the terminal (which may also be referred to as a system network function module) uses the current network access record and the terminal status associated with the current network access record according to the set period, and utilizes the access prediction model (which may be hidden). The Markov model) predicts the most likely network resource address to be accessed based on the current network access record and the terminal state, and outputs the network resource address as a decision. That is to say, the output of the access prediction model is the set number of network resource addresses that the terminal is most likely to access next. The network data corresponding to the network resource address is downloaded by the server in the external network according to the network resource address, and the network data is stored in the local cache, and the content corresponding to the pre-loaded URL is stored in the local cache. If the network access request sent by the current application is detected, the network resource address included in the network access request is extracted, and it is determined whether the content corresponding to the network resource address is preloaded locally. If it is preloaded, the cache file corresponding to the network resource address is read by the local cache and directly returned to the application, and the network data is not required to be downloaded by the server in the external network. However, if the network data corresponding to the network resource address is not preloaded, the network access request continues, and the network data corresponding to the network resource address is downloaded by the external network server.

The technical solution of the embodiment, by calling the access prediction model according to the set period, predicting the network resource address that is most likely to be accessed next based on the current network access record and the terminal status associated with the current network access record, and downloading the network resource in advance The network data corresponding to the network resource address may be used to determine whether the network resource address of the network access request has corresponding network data in the local cache when the network access request of the application is detected, and if so, the network is directly read by the local cache. Accessing the network data corresponding to the request, shortening the loading time of the application for the network resource, improving the performance of the application, and avoiding the situation that the loading time is long caused by loading the data by the external network.

FIG. 6 is a structural block diagram of a model construction apparatus according to an embodiment of the present application. The device may be implemented by at least one of software and hardware, and may be integrated into the terminal and configured to perform the model construction method provided by the embodiment of the present application. As shown in FIG. 6, the apparatus includes a history acquisition module 610, a matrix determination module 620, and a model construction module 630.

The history acquisition module 610 is configured to obtain a historical network access record and a terminal status associated with the historical network access record.

The matrix determination module 620 is configured to determine a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state.

The model construction module 630 is configured to construct an access prediction model based on the state transition probability matrix and the observation probability matrix.

The technical solution of the embodiment provides a model construction device, which acquires a historical network access record and a terminal state associated with the historical network access record; and determines a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state; The access prediction model is constructed based on the state transition probability matrix and the observation probability matrix; thus, the observation probability of the observation value corresponding to the next state can be determined by accessing the prediction model based on the current state, since the observation value represents the network resource address, and further, the user The target network resources have been predicted before accessing the network resources, which improves the intelligence of the mobile terminal.

In an embodiment, the matrix determination module 620 includes a pre-processing sub-module, a first computing sub-module, and a second computing sub-module.

The pre-processing sub-module is configured to pre-process the historical network access record and the terminal state to obtain a state set and an observation set.

The first calculation submodule is configured to calculate a state transition probability of each state in the state set, and generate a state transition probability matrix according to the state transition probability.

The second calculation submodule is configured to calculate an observation probability of the observation value corresponding to each state according to the state set and the observation set, and generate an observation probability matrix according to the observation probability.

In an embodiment, the state transition probability of each state in the set of states may be calculated using the following formula:

In an embodiment, the second calculation sub-module is configured to: sequentially acquire one state in the state set as a current state, and count the number of occurrences of the current state; and obtain an observation value corresponding to the current state and the current state. The number of occurrences, using the following formula to calculate the observation probability of the current state:

Indicates the probability that the observation value k is generated in the current state q _x , and N(i _t =q _x , v=k) represents the number of times the observation value k is generated in the current state q _x . In an embodiment, the historical network access record includes an application for accessing the network, a network resource address, and an access time; the terminal status includes network environment information and a charging status.

And the pre-processing sub-module is configured to: match the program number of the application that accesses the network according to the preset rule, and the network number corresponding to the network resource address; and distinguish the working day and the rest day according to the access time, respectively The work date and the rest day are assigned different date numbers; the time number is determined according to the time period to which the access time belongs, wherein the time period is obtained by equally dividing the preset time interval in the natural day, and the time period is stored in association with the time number. Determining whether the terminal accesses the wireless network according to the network environment information, determining a network status value according to the determination result; determining, according to the charging status, whether the terminal is charging, determining a charging status value according to the determination result; according to the program number, date number, time The number, the network status value, and the state of charge value generate a set of states, and form an observation set based on the network number.

FIG. 7 is a structural block diagram of a network resource preloading apparatus according to an embodiment of the present application. The device may be implemented by using at least one of software and hardware, and may be integrated into the terminal and configured to perform the network resource preloading method provided by the embodiment of the present application. As shown in FIG. 7, the apparatus includes a current record acquisition module 710, a model input module 720, and a preload module 730.

The current record obtaining module 710 is configured to acquire a current network access record and a terminal status associated with the current network access record according to a set period.

The model input module 720 is configured to input the current network access record and the terminal state into a pre-built access prediction model, wherein the access prediction model is based on a historical network access record and a terminal state training corresponding to the historical network access record. Model.

The preloading module 730 is configured to acquire the network resource identifier to be accessed that is output by the access prediction model, and download the network data corresponding to the network resource identifier, and store the data in the local cache.

The technical solution of the embodiment provides a network resource preloading device, which can preload the network data corresponding to the network resource identifier that may be accessed before acquiring the network data according to a certain network resource identifier, so as to directly read the network from the local cache. Data, shortening the loading time of network data, thus avoiding the situation that the loading time is longer due to loading data from the external network.

In an embodiment, the model input module 720 is configured to: preprocess the current network access record and the terminal state to obtain a current state; and input the current state into a pre-built access prediction model to pass the access prediction The model predicts a next state corresponding to the next moment, and respectively determines an observation probability of the network resource address corresponding to the next state.

In an embodiment, the method further includes: downloading a policy selection module, configured to determine whether the terminal accesses the wireless network before the preloading module downloads the network data corresponding to the network resource identifier; and if the terminal accesses the wireless network, And causing the preloading module to perform an operation of downloading network data corresponding to the network resource identifier; if the terminal does not access the wireless network, acquiring state information of the cellular mobile network data of the terminal.

When the status information meets the preset condition, causing the preloading module to perform an operation of downloading network data corresponding to the network resource identifier;

When the status information does not meet the preset condition, the preloading module 730 is configured to: determine the target network data according to at least one of a data type and a data amount of the network data corresponding to the network resource identifier, The target network data is loaded.

In an embodiment, the method further includes: an address obtaining module, configured to: after the preloading module downloads the network data corresponding to the network resource identifier, acquire, by using a network crawler, a first network resource included in the network resource identifier The second network resource address of the address loads the network data corresponding to the second network resource address, and is stored in the local cache.

In an embodiment, the method further includes: a cache clearing module, configured to acquire a storage space occupied by network data in the local cache; and when the storage space exceeds a set threshold, according to a cache time and a frequency of use of the network data At least one of determining data to be cleared and clearing the data to be cleared.

In an embodiment, the method further includes: requesting an acquisition module, configured to acquire a network access request of the application; and a cache query module, configured to query the local cache according to the network access request, and determine a network corresponding to the network access request a data pre-loading state, configured to: if the network data corresponding to the network access request is preloaded, read the network data, and return to the application; the data downloading module is set to If the network data corresponding to the network access request is not preloaded, the network data is downloaded by the external network.

Embodiments of the present application also provide a storage medium including computer executable instructions that, when executed by a computer processor, are configured to perform a model construction method, the method comprising: obtaining a historical network access record and The historical network access records the associated terminal state; determining the state transition probability matrix and the observation probability matrix according to the historical network access record and the terminal state; and constructing the access prediction model based on the state transition probability matrix and the observation probability matrix.

In addition, the embodiment of the present application further provides another storage medium including computer executable instructions, which are set to execute a network resource preloading method when executed by a computer processor, the method comprising: Obtaining a current network access record and a terminal status associated with the current network access record; inputting the current network access record and terminal status into a pre-built access prediction model, wherein the access prediction model is accessed according to a historical network Recording a model of the terminal state training corresponding to the historical network access record; acquiring the network resource identifier to be accessed output by the access prediction model, downloading the network data corresponding to the network resource identifier, and storing the network data in the local cache.

Storage media - any of a variety of types of memory devices or storage devices. The term "storage medium" is intended to include: a mounting medium such as a Compact Disc Read-Only Memory (CD-ROM), a floppy disk or a tape device; a computer system memory or a random access memory such as a dynamic random access memory; (Dynamic Random Access Memory, DRAM), Double Data Rate Random Access Memory (DDR RAM), Static Random Access Memory (SRAM), Extended Data Output Random Access Memory ( Extended Data Output Random Access Memory (EDO RAM), Rambus RAM, etc.; non-volatile memory such as flash memory, magnetic media (eg, hard disk or optical storage); registers or other similar types of memory elements, and the like. The storage medium may also include other types of memory or a combination thereof. Additionally, the storage medium may be located in a first computer system in which the program is executed, or may be located in a different second computer system, the second computer system being coupled to the first computer system via a network, such as the Internet. The second computer system can provide program instructions to the first computer for execution. The term "storage medium" can include two or more storage media that can reside in different locations (eg, in different computer systems connected through a network). The storage medium may store program instructions (eg, embodied as a computer program) executable by the at least one processor.

Certainly, a storage medium containing computer executable instructions provided by the embodiments of the present application, the computer executable instructions are not limited to the operation of model construction as described above, and may also execute the model construction method provided by any embodiment of the present application. Related operations in .

It should be noted that the storage medium including the computer executable instructions provided by the embodiment of the present application is not limited to the operation of the network resource preloading as described above, and may also perform any of the embodiments of the present application. Related operations in the provided network resource preloading method.

The embodiment of the present application provides a terminal in which the model construction apparatus provided in the embodiment of the present application can be integrated. The embodiment of the present application provides another terminal, where the network resource preloading device provided by the embodiment of the present application can be integrated. It should be noted that the model can be constructed by the first terminal, and the model is transplanted to the second terminal that performs the network resource preloading operation, and the model construction and the network resource preloading operation can also be performed in the same terminal. Among them, the terminal includes a smart phone, a tablet computer, a handheld game machine, a notebook computer, and a smart watch. FIG. 8 is a structural block diagram of a terminal according to an embodiment of the present application. As shown in FIG. 8, the terminal may include a memory 810 and a processor 820. The memory 810 is configured to store a computer program, a historical network access record, a terminal status, and an access prediction model; the processor 820 reads and executes the computer program stored in the memory 810. The processor 820, when executing the computer program, implements the steps of: acquiring a historical network access record and a terminal status associated with the historical network access record; determining a state transition probability matrix according to the historical network access record and the terminal status and Observing probability matrix; constructing an access prediction model based on the state transition probability matrix and the observation probability matrix.

It should be noted that the memory 810 of the terminal may also be configured to store another computer program and pre-loaded network data. Processor 820 can also be configured to read and execute computer programs stored in said memory 810. The processor 820, when executing the computer program, implements the steps of: acquiring a current network access record and a terminal status associated with the current network access record according to a set period; and inputting the current network access record and terminal status in advance The access prediction model is configured, wherein the access prediction model is a model of terminal state training corresponding to the historical network access record and the historical network access record; acquiring the network resource identifier to be accessed output by the access prediction model, and downloading The network resource identifies the corresponding network data and stores the data in the local cache.

The memory and processor listed in the above examples are all components of the terminal, and the terminal may further include other components. Taking the mobile terminal as an example, the possible structure of the above terminal is explained. FIG. 9 is a schematic structural diagram of a mobile terminal according to an embodiment of the present application. As shown in FIG. 9, the mobile terminal may include: a memory 901, a central processing unit (CPU) 902 (also referred to as a processor), a peripheral interface 903, a radio frequency (RF) circuit 905, and audio. Circuitry 906, speaker 911, power management chip 908, input/output (I/O) subsystem 909, other input/control devices 910, and external port 904 are communicated via at least one communication bus or signal line 907.

It should be understood that the illustrated mobile terminal 900 is merely one example of a mobile terminal, and that the mobile terminal 900 may have more or fewer components than those shown in the figures, may combine at least two components, or may Has a different component configuration. The various components shown in the figures can be implemented in hardware, software, or a combination of hardware and software, including at least one of signal processing and application specific integrated circuits, at least in combination with hardware, software, or a combination of hardware and software. One.

Taking the mobile terminal as a mobile phone as an example, the mobile terminal is described in detail.

The memory 901 can be accessed by the CPU 902, the peripheral interface 903, etc., and the memory 901 can include a high speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or Other volatile solid-state storage devices. The computer program is stored in the memory 901, and the historical network access record, the terminal status, the access prediction model, the preloaded network data, and the like can also be stored.

Peripheral interface 903, which can connect the input and output peripherals of the device to CPU 902 and memory 901.

I/O subsystem 909, which can connect input and output peripherals on the device, such as touch screen 912 and other input/control devices 910, to peripheral interface 903. The I/O subsystem 909 can include a display controller 9091 and at least one input controller 9092 configured to control other input/control devices 910. At least one of the input controllers 9092 receives electrical signals from other input/control devices 910 or transmits electrical signals to other input/control devices 910. Other input/control devices 910 may include physical buttons (press buttons, rocker buttons, etc.) , dial, slide switch, joystick, click on the wheel. It is worth noting that the input controller 9092 can be connected to any of the following: a keyboard, an infrared port, a USB interface, and a pointing device such as a mouse.

The touch screen 912 is an input interface and an output interface between the user terminal and the user, and displays the visual output to the user. The visual output may include graphics, text, icons, videos, and the like.

Display controller 9091 in I/O subsystem 909 receives an electrical signal from touch screen 912 or an electrical signal to touch screen 912. The touch screen 912 detects the contact on the touch screen, and the display controller 9091 converts the detected contact into an interaction with the user interface object displayed on the touch screen 912, that is, realizes human-computer interaction, and the user interface object displayed on the touch screen 912 may be running. The icon of the game, the icon of the network to the corresponding network, and the like. It is worth noting that the device may also include a light mouse, which is a touch sensitive surface that does not display a visual output, or an extension of a touch sensitive surface formed by the touch screen.

The RF circuit 905 is mainly configured to establish communication between the mobile phone and the wireless network (ie, the network side), and implement data reception and transmission between the mobile phone and the wireless network. For example, sending and receiving short messages, emails, and the like. In one embodiment, RF circuit 905 receives and transmits an RF signal, also referred to as an electromagnetic signal, and RF circuit 905 converts the electrical signal into an electromagnetic signal or converts the electromagnetic signal into an electrical signal, and through the electromagnetic signal and communication network And other devices to communicate. RF circuitry 905 may include known circuitry configured to perform these functions including, but not limited to, an antenna system, an RF transceiver, at least one amplifier, a tuner, at least one oscillator, a digital signal processor, a codec (COder- DECoder, CODEC) Chipset, Subscriber Identity Module (SIM), etc.

The audio circuit 906 is primarily configured to receive audio data from the peripheral interface 903, convert the audio data into an electrical signal, and transmit the electrical signal to the speaker 911.

The speaker 911 is arranged to restore the voice signal received by the mobile phone from the wireless network through the RF circuit 905 to sound and play the sound to the user.

The power management chip 908 is configured to provide power and power management for the hardware connected to the CPU 902, the I/O subsystem, and the peripheral interface.

The terminal provided by the embodiment of the present application can determine the observation probability of the observation value corresponding to the next state by using the access prediction model based on the current state, and predict the target network resource before the user accesses the network resource, thereby improving the intelligence of the mobile terminal.

The terminal provided by the embodiment of the present application may preload the network data corresponding to the network resource identifier that may be accessed before acquiring the network data according to the identifier of the network resource, so as to directly read the network data from the local cache, and shorten the loading of the network data. Time, and thus, avoiding the fact that loading time is caused by loading data from the external network.

The model building device, the storage medium and the terminal provided in the above embodiments can execute the model building method provided by the embodiment of the present application, and have the corresponding functional modules and beneficial effects of executing the method. For technical details that are not described in detail in the above embodiments, reference may be made to the model construction method provided by the embodiments of the present application.

The network resource preloading device, the storage medium, and the terminal provided in the foregoing embodiments may perform the network resource preloading method provided by the embodiment of the present application, and have the corresponding functional modules and beneficial effects of executing the method. For the technical details that are not described in detail in the foregoing embodiments, refer to the network resource preloading method provided by the embodiment of the present application.

Claims

A model construction method, including:

Obtaining a historical network access record and a terminal status associated with the historical network access record;

Determining a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state;

An access prediction model is constructed based on the state transition probability matrix and the observation probability matrix.
The method of claim 1, wherein determining the state transition probability matrix and the observation probability matrix according to the historical network access record and the terminal state comprises:

Performing pre-processing on the historical network access record and the state of the terminal to obtain a state set and an observation set;

Calculating a state transition probability of each state in the set of states, and generating a state transition probability matrix according to the state transition probability;

Observing the observation probability of the observation value corresponding to each state according to the state set and the observation set, and generating an observation probability matrix according to the observation probability.
The method of claim 2, wherein calculating a state transition probability for each state in the set of states comprises:

The state transition probability for each state in the set of states is calculated using the following formula:

Where q i represents the current state of the current time, q j represents the adjacent state at the next moment, and P(i t+1 =q j |i t =q i ) indicates that the state in the state set is transferred from the current state to The probability of the adjacent state, N(i t =q i , i i+1 =q j ), represents the number of occurrences of the state in the state set being q i and the neighboring state being q j , N(i t =q i ) represents the number of times the current state in the set of states is q i .
The method according to claim 3, wherein the observation probability of the observation value corresponding to each state is calculated according to the state set and the observation set, including:

Obtaining one state in the state set as a current state, and counting the number of occurrences of the current state;

According to the observation value corresponding to the current state and the number of occurrences of the current state, the following formula is used to calculate the observation probability of the current state:

Wherein, the observed values k ∈ {v 1 , v 2 , . . . , v m }, {v 1 , v 2 , . . . , v m } are observation sets,
Indicates the probability that the observation value k is generated in the current state q x , and N(i t =q x , v=k) represents the number of times the observation value k is generated in the current state q x .
The method according to any one of claims 2 to 4, wherein the historical network access record comprises an application accessing a network, a network resource address and an access time; the terminal status comprises network environment information and a charging status;

Performing pre-processing on the historical network access record and the terminal state to obtain a state set and an observation set, including:

Matching, by a preset rule, a program number of the application that accesses the network, and a network number corresponding to the network resource address;

Differentiating the working day and the rest day according to the access time, respectively assigning different date numbers to the working day and the rest day;

Determining a time number according to the time period to which the access time belongs, wherein the time period is obtained by equally dividing the preset time interval in the natural day, and the time period is stored in association with the time number;

Determining, according to the network environment information, whether the terminal accesses the wireless network, and determining a network state value according to the determination result;

Determining whether the terminal is charging according to the state of charge, and determining a state of charge value according to the determination result;

A state set is generated based on the program number, date number, time number, network state value, and charge state value, and an observation set is constructed based on the network number.
A network resource preloading method includes:

Obtaining a current network access record and a terminal status associated with the current network access record according to a set period;

Inputting the current network access record and the terminal status into a pre-built access prediction model, where the access prediction model is a model of terminal state training corresponding to the historical network access record and the historical network access record;

Obtaining a network resource identifier to be accessed output by the access prediction model, downloading network data corresponding to the network resource identifier, and storing the network data in a local cache.
The method of claim 6, wherein the inputting the current network access record and the terminal status into a pre-built access prediction model comprises:

Performing pre-processing on the current network access record and the status of the terminal to obtain a current state;

And inputting the current state into the pre-built access prediction model, so as to predict a next state corresponding to the next moment by using the access prediction model, and determine an observation probability of the network resource address corresponding to the next state.
The method of claim 6, before downloading the network data corresponding to the network resource identifier, the method further includes:

Determine whether the terminal accesses the wireless network;

And in response to determining that the terminal accesses the wireless network, performing an operation of downloading network data corresponding to the network resource identifier;

Obtaining state information of the cellular mobile network data of the terminal in response to determining that the terminal does not access the wireless network;

When the status information meets the preset condition, performing an operation of downloading network data corresponding to the network resource identifier;

When the status information does not meet the preset condition, the network data corresponding to the network resource identifier is downloaded, including:

Determining the target network data according to at least one of a data type and a data amount of the network data corresponding to the network resource identifier, and loading the target network data.
The method of claim 6, after downloading the network data corresponding to the network resource identifier, further comprising:

And obtaining, by the network crawler, the second network resource address that is linked to the first network resource address included in the network resource identifier, and loading the network data corresponding to the second network resource address, and storing the network data in the local cache.
The method of claim 6 further comprising:

Obtaining a storage space occupied by network data in the local cache;

When the storage space exceeds a set threshold, determining data to be cleared according to at least one of a cache time and a use frequency of the network data, and clearing the data to be cleared.
The method of any of claims 6-10, further comprising:

Obtain a network access request for the application;

Querying the local cache according to the network access request, and determining a preload state of the network data corresponding to the network access request;

Responding to the network data corresponding to the network access request being preloaded, reading the network data, and returning to the application;

The network data is downloaded by the external network in response to the network data corresponding to the network access request not being preloaded.
A model building device comprising:

a history acquisition module configured to acquire a historical network access record and a terminal status associated with the historical network access record;

a matrix determining module, configured to determine a state transition probability matrix and an observation probability matrix according to the historical network access record and the terminal state;

The model building module is configured to construct an access prediction model based on the state transition probability matrix and the observation probability matrix.
The apparatus according to claim 12, wherein the matrix determining module comprises a preprocessing submodule, a first computing submodule and a second computing submodule;

The pre-processing sub-module is configured to pre-process the historical network access record and the terminal state to obtain a state set and an observation set.

The first calculation submodule is configured to calculate a state transition probability of each state in the state set, and generate a state transition probability matrix according to the state transition probability.

The second calculation submodule is configured to calculate an observation probability of an observation value corresponding to each state according to the state set and the observation set, and generate an observation probability matrix according to the observation probability.
The apparatus according to claim 13, wherein the historical network access record comprises an application for accessing a network, a network resource address, and an access time; the terminal status includes network environment information and a charging status;

The preprocessing submodule is configured to:

Matching, by a preset rule, a program number of the application that accesses the network, and a network number corresponding to the network resource address;

Differentiating the working day and the rest day according to the access time, respectively assigning different date numbers to the working day and the rest day;

Determining a time number according to the time period to which the access time belongs, wherein the time period is obtained by equally dividing the preset time interval in the natural day, and the time period is stored in association with the time number;

Determining, according to the network environment information, whether the terminal accesses the wireless network, and determining a network state value according to the determination result;

Determining whether the terminal is charging according to the state of charge, and determining a state of charge value according to the determination result;

A state set is generated based on the program number, date number, time number, network state value, and charge state value, and an observation set is constructed based on the network number.
A network resource preloading device includes:

The current record obtaining module is configured to acquire a current network access record and a terminal state associated with the current network access record according to a set period;

a model input module, configured to input the current network access record and the terminal status into a pre-built access prediction model, wherein the access prediction model is trained according to a historical network access record and a terminal state corresponding to the historical network access record model;

The preloading module is configured to obtain the network resource identifier to be accessed that is output by the access prediction model, and download the network data corresponding to the network resource identifier, and store the data in the local cache.
The device of claim 15 further comprising:

The downloading the policy selection module is configured to determine whether the terminal accesses the wireless network before the preloading module downloads the network data corresponding to the network resource identifier;

If the terminal accesses the wireless network, the preloading module is configured to perform an operation of downloading network data corresponding to the network resource identifier;

Obtaining state information of the cellular mobile network data of the terminal if the terminal does not access the wireless network;

When the status information meets the preset condition, causing the preloading module to perform an operation of downloading network data corresponding to the network resource identifier;

When the status information does not meet the preset condition, the preloading module is configured to: determine the target network data according to at least one of a data type and a data amount of the network data corresponding to the network resource identifier, The target network data is loaded.
The device of claim 15 further comprising:

An address obtaining module, configured to: after downloading the network data corresponding to the network resource identifier, acquire, by using a network crawler, a second network resource address that is linked to the first network resource address included in the network resource identifier, and load the second network The network data corresponding to the resource address is stored in the local cache.
A device according to any one of claims 15-17, the device further comprising:

The requesting module is configured to obtain a network access request of the application; the cache query module is configured to query the local cache according to the network access request, and determine a preload state of the network data corresponding to the network access request; a module, configured to: if the network data corresponding to the network access request is preloaded, read the network data, and return to the application; and the data downloading module is configured to: if the network access request corresponds to the network data If not preloaded, the network data is downloaded by the external network.
A computer readable storage medium having stored thereon a computer program, the program being executed by a processor to implement the model construction method according to any one of claims 1 to 5; or the program being implemented by the processor The network resource preloading method of any one of 6 to 11.
A terminal comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, the processor executing the computer program to implement the model construction according to any one of claims 1 to 5 The method or the network resource preloading method according to any one of claims 6 to 11 is implemented when the processor executes the computer program.