Short message sending time optimization method

Abstract

The invention discloses a short message sending time optimization method, which comprises the following steps of: 1, defining influence factors of user behavior data, namely dividing the influence factors into user communication behavior data and user mobile phone interface operation behavior data; step 2, modeling method description and sampling time fragmentation are carried out, a time period in which short messages can be sent every day and the rest of the user is not influenced is segmented into N sampling time fragments, and in each time fragment, various behavior data of the user are counted; and step 3, sub-model definition: due to large difference between the behavior models of the user in workdays and holidays, defining two sub-models, namely a workday sub-model and a holiday sub-model, and the sampling time slices of the two sub-models are the same. The invention belongs to the technical field of optimization methods in a short message group sending scene, in particular to a short message sending time optimization method, and aims to perform behavior analysis modeling on each time slice of each user to obtain an idle estimated value of each time slice of each user.

Description

Technical Field

[0001] The present invention belongs to the technical field of optimization methods in SMS group sending scenarios, and specifically refers to a method for optimizing SMS sending time. Background Art

[0002] The existing group SMS technology cannot implement a mechanism to determine the best time to send SMS that is tailored to each individual, resulting in users not having enough time to open and read the SMS when they receive it, thus reducing the reading rate after the SMS is sent.

[0003] In order to increase the probability of users opening and reading text messages, we can analyze two aspects of user behavior data (communication behavior data and mobile phone interface operation behavior data), and define the influencing factors of these behavior data to determine whether the user is idle through big data modeling (divided into positive factors and negative factors, and define the strength of the influencing factors, such as: strong positive factors, weak negative factors, etc.). According to the superposition of these influencing factors, we can find the time periods when users are relatively idle, and arrange to send text messages during these idle time periods, thereby increasing the probability of users reading text messages. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a method for optimizing SMS sending time. This scheme aims to perform behavioral analysis modeling for each time slice of each user to obtain the idleness estimation of each time slice of each user.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides a method for optimizing the time of sending a short message, comprising the following steps:

[0006] Step 1: Define the influencing factors of user behavior data, which are divided into user communication behavior data and user mobile phone interface operation behavior data;

[0007] Step 2: Modeling method description: Sampling time slices: divide the time period during which SMS can be sent and will not affect the user's rest into N sampling time slices. In each time slice, various user behavior data are counted.

[0008] Step 3: Sub-model definition. Since the user behavior models on weekdays and holidays are quite different, two sub-models are defined, namely the weekday model and the holiday sub-model. The sampling time slices of the two sub-models are the same, but when the final output of each time slice value is calculated and output according to each sub-model;

[0009] Step 4: Define influence. Define four behavioral influence factors, namely strong positive factor, weak positive factor, strong negative factor, and weak negative factor. And assign influence to each factor.

[0010] Step 5: Influencing factor superposition algorithm: Taking the working day model as an example (the implementation method of the holiday sub-model is similar), take the time slice data of a specific user in the previous month to perform weight factor superposition calculation.

[0011] Furthermore, in the step 2, the user can be sent text messages during the custom time period without affecting the user's rest period. The sampling time is then defined as 15 minutes (the length of the time slice can be customized), so there are 48 time slices per day for counting various user behavior data.

[0012] Furthermore, in the working day model of step three, the final output result of the time slice is a negative value, indicating that the user is not idle; however, in the holiday sub-model, the final output result of the time slice is a positive value, indicating that the user is idle.

[0013] Furthermore, the step 4 may define the positive factor and the negative factor to have the same influence. In this case, the absolute values ​​of the strong positive factor and the strong negative factor are defined to be the same, but the positive and negative values ​​are opposite, such as: the values ​​are A and -A. The absolute values ​​of the weak positive factor and the weak negative factor are the same, but the positive and negative values ​​are opposite, such as: the values ​​are B and -B;

[0014] It is also possible to define the influence of positive factors and negative factors to be different, thereby implementing a model calculation method that emphasizes positive behavior or negative behavior.

[0015] The beneficial effects achieved by the present invention using the above structure are as follows:

[0016] 1. This solution uses big data modeling to analyze the behavior patterns and response time of each user in each time slice, and automatically determines the best SMS sending time for each user, thereby improving the user's reading and response rate DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0018] Embodiment 1: A method for optimizing the time of sending a short message, comprising the following steps:

[0019] Step 1: Definition of influencing factors of user behavior data;

[0020] Step 2: describe the modeling method, sample time slices, and count various user behavior data;

[0021] Step 3: Sub-model definition, calculation and output according to each sub-model;

[0022] Step 4: Define influence. Define four behavioral influence factors, namely strong positive factor, weak positive factor, strong negative factor, and weak negative factor. And assign influence to each factor.

[0023] Step 5: Influencing factor superposition algorithm: Taking the working day model as an example (the implementation method of the holiday sub-model is similar), take the time slice data of a specific user in the previous month to perform weight factor superposition calculation.

[0024] In the step 2, the user can be sent text messages during the custom time period without affecting the user's rest period. The sampling time is then segmented into 15 minutes (the segment length can be customized), so there are 48 time segments per day for counting various user behavior data.

[0025] In the working day model of step 3, the final output result of the time slice is a negative value, indicating that the user is not idle; however, in the holiday sub-model, the final output result of the time slice is a positive value, indicating that the user is idle.

[0026] In step 4, the positive factor and the negative factor can be defined to have the same influence. In this case, the absolute values ​​of the strong positive factor and the strong negative factor are defined to be the same, but the positive and negative values ​​are opposite, such as: the values ​​are A and -A. The absolute values ​​of the weak positive factor and the weak negative factor are the same, but the positive and negative values ​​are opposite, such as: the values ​​are B and -B;

[0027] It is also possible to define the influence of positive factors and negative factors to be different, thereby implementing a model calculation method that emphasizes positive behavior or negative behavior.

[0028] User communication behavior data, including the following four types:

[0029] 1. Not answering the phone / actively hanging up: A strong negative factor. The time when you cannot answer the phone is generally when the user is concentrating on handling important matters, and the possibility of being idle is extremely low.

[0030] 2. Answering the phone: a weak positive factor, because answering the phone is a passive behavior, it only indicates that the user is convenient to answer the phone during this period, and does not mean that the user is free.

[0031] 3. Making phone calls: a weak positive factor, which can only indicate that the user has time to arrange independently, but it is possible that the purpose of making a phone call is for work.

[0032] 4. Traffic generation: a weak positive factor, indicating that the user has free time to use traffic, but cannot distinguish whether the user uses traffic for entertainment or work

[0033] User mobile interface operation behavior data, including the following 5 types:

[0034] 1. Opening the SMS inbox: A strong positive factor, indicating that the user actively opens the SMS inbox and has the opportunity to see other SMS content.

[0035] 2. Opening an entertainment app: a strong positive factor, indicating that the user is idle and has entered the entertainment period.

[0036] 3. Unlock the screen: a weak positive factor. After unlocking the screen, there is a chance to see SMS notifications, but it is impossible to determine whether the user is free.

[0037] 4. Opening an office app: strongly negative, indicating that the user has started to handle office affairs and the possibility of being idle is extremely low.

[0038] 5. Opening the email app: Weakly negative, indicating that the user has started processing emails. It is impossible to determine whether the emails are related to office affairs, and the possibility of being idle is low.

[0039] Impact factor superposition algorithm: First, we need to define some symbols to describe this algorithm:

[0040] Bij represents the user's behavior data in the jth time slice on the i-th day, where i = 1, 2, ..., 30 (assuming there are 30 days in the last month), j = 1, 2, ..., N (N is the number of time slices).

[0041] Wij represents the weight of the user in the jth time slice of the i-th day. This weight can be set according to different time periods. For example, a smaller weight can be set during the user's lunch break.

[0042] Pij represents the influence value superposition of the user's positive behavior data in the j-th time slice on the i-th day.

[0043] Nij represents the influence value superposition of the user's negative behavior data in the j-th time slice on the i-th day.

[0044] Based on the above definition, we can write the following formula to calculate the monthly behavior evaluation value Aij of each time slice, which represents the monthly behavior evaluation value of the jth time slice on the i-th day:

[0045] Aij=(Pij+Nij)×Wij

[0046] Note: Since the influence of negative behavior is defined as a negative value, (Pij+Nij) can be used to directly represent the superposition of influences;

[0047] Then, we can define a positive value Y. When Aij>Y, it is considered that the user is idle at this moment, and it is suitable to send text messages to the user in this time slice. It is expected that the user is more likely to open and read the text messages in this time slice.

[0048] Embodiment 2: A method for optimizing the time of sending short messages. In the specific implementation process:

[0049] Taking the working day model as an example, a user behavior is assigned the following influence:

[0050] Strong positive: 2

[0051] Weak positive: 1

[0052] Strong negative: -2

[0053] Weak negative: -1

[0054] Define the time slice as 15 minutes. After the impact factors of a 15-minute slice are added up, the 15-minute period of each day of the month is added up and then divided by the number of working days in the month. Finally, the user idleness valuation of this 15-minute period is 2. Therefore, if we define the idleness valuation greater than 1 as an idle period, this 15-minute period is marked as an idle period. It is recommended to send SMS to this user in this time slice.

[0055] This is an implementation example of a time slice of a user. When we calculate the idle estimates of all time slices of all users, we can obtain a differentiated SMS sending strategy for each user.

[0056] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0057] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0058] The present invention and its implementation methods are described above, and such description is not restrictive. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design structures and embodiments similar to the technical solution without creativity, which should all fall within the protection scope of the present invention.

Claims

1. A method for optimizing SMS sending time, characterized in that: The following steps are involved: Step 1: Define the influencing factors of user behavior data, which are divided into user communication behavior data and user mobile phone interface operation behavior data; Step 2: Modeling method description: Sampling time slices: divide the time period during which SMS can be sent and will not affect the user's rest into N sampling time slices. In each time slice, various user behavior data are counted. Step 3: Sub-model definition. Since the user behavior models on weekdays and holidays are quite different, two sub-models are defined, namely the weekday model and the holiday sub-model. The sampling time slices of the two sub-models are the same, but when the final output of each time slice value is calculated and output according to each sub-model; Step 4: Define influence. Define four behavioral influence factors, namely strong positive factor, weak positive factor, strong negative factor, and weak negative factor. And assign influence to each factor. Step 5: Influencing factor superposition algorithm: Taking the working day model as an example (the implementation method of the holiday sub-model is similar), take the time slice data of a specific user in the previous month to perform weight factor superposition calculation.

2. A method for optimizing the time of sending short messages according to claim 1, characterized in that: In the step 2, the user can be sent text messages during the custom time period without affecting the user's rest period. The sampling time is then segmented into 15 minutes (the segment length can be customized), so there are 48 time segments per day for counting various user behavior data.

3. A method for optimizing the time of sending short messages according to claim 2, characterized in that: In the working day model of step 3, the final output result of the time slice is a negative value, indicating that the user is not idle; however, in the holiday sub-model, the final output result of the time slice is a positive value, indicating that the user is idle.

4. The method for optimizing the time of sending short messages according to claim 3, characterized in that: In step 4, the positive factor and the negative factor can be defined to have the same influence. In this case, the absolute values ​​of the strong positive factor and the strong negative factor are defined to be the same, but the positive and negative values ​​are opposite, such as: the values ​​are A and -A. The absolute values ​​of the weak positive factor and the weak negative factor are the same, but the positive and negative values ​​are opposite, such as: the values ​​are B and -B; It is also possible to define the influence of positive factors and negative factors to be different, thereby implementing a model calculation method that emphasizes positive behavior or negative behavior.