TWI798973B - Electronic device and method for estimating poor customer experience of wireless network - Google Patents

Abstract

An electronic device and a method for estimating a poor customer experience of a wireless network are provided. The method includes: receiving user feedback information, user signal quality information, user website browsing history corresponding to a user; counting a number of predefined strings in the user feedback information to generate first feature data; counting a number of predefined website in the user browsing history to generate second feature data; calculating an average of signal parameters according to the user signal quality information to generate third feature data; inputting the first feature data, the second feature data, and the third feature data to at least one machine learning model to generate an estimated result; and outputting the estimated result corresponding to the user.

Description

Electronic device and method for predicting user experience degradation of wireless network

The present invention relates to an electronic device and method for predicting user experience degradation of a wireless network.

Providers of wireless network services usually use the performance indicators of base stations to analyze and predict the failure probability and cause of failure of base stations. However, in some cases, the performance index of the base station cannot truly reflect the user experience of the customer. Even when the performance index of the base station is good, events of user experience degradation may still occur.

The present invention provides an electronic device and method for predicting user experience degradation of a wireless network, which can perform prediction of user experience degradation based on user feedback.

An electronic device for predicting user experience degradation in a wireless network of the present invention includes a processor and a transceiver. The processor is coupled to the transceiver, wherein the processor is configured to perform: receiving user feedback information corresponding to the user, user signal quality information, and user website browsing records through the transceiver; counting the number of predefined word strings in the user feedback information and Generate the first characteristic data; count the number of predefined websites in the user’s website browsing records to generate the second characteristic data; calculate the average value of the signal parameters according to the user’s signal quality information to generate the third characteristic data; combine the first characteristic data, the second The second characteristic data and the third characteristic data are input into at least one machine learning model to generate a prediction result; and the prediction result corresponding to the user is output through the transceiver.

In an embodiment of the present invention, the above-mentioned processor is further configured to perform: receiving historical user feedback information through a transceiver, wherein the historical user feedback information includes a plurality of feedback messages respectively corresponding to a plurality of users; The intersection of the feedback messages produces a predefined string.

In an embodiment of the present invention, the above-mentioned processor is further configured to perform: receiving historical user feedback information through a transceiver, wherein the historical user feedback information includes a plurality of website browsing records respectively corresponding to a plurality of users; The intersection of website browsing records generates a predefined website.

In an embodiment of the present invention, the above at least one machine learning model includes at least one of the following: random forest model, gradient boosting tree and neural network.

In an embodiment of the present invention, the above at least one machine learning model outputs at least one probability value according to the first characteristic data, the second characteristic data and the third characteristic data, wherein the processor generates a prediction result according to the at least one probability value.

In an embodiment of the present invention, the processor generates a prediction result in response to at least one probability value being greater than a threshold, wherein the prediction result indicates that user experience degradation is about to occur.

In an embodiment of the present invention, the above-mentioned processor is further configured to perform: receiving historical user feedback information, historical user signal quality information, and historical user website browsing records corresponding to historical user experience degradation through a transceiver; and according to the historical The user feedback information, historical user signal quality information and historical user website browsing records train at least one machine learning model.

In an embodiment of the present invention, the above-mentioned processor is further configured to perform: counting the number of predefined character strings in historical user feedback information to generate first historical characteristic data; counting the predefined word strings in historical user website browsing records The number of websites to generate the second historical characteristic data; calculate the historical average value of signal parameters according to the historical user signal quality information to generate the third historical characteristic data; and according to the first historical characteristic data, the second historical characteristic data and the third historical characteristic data The data trains at least one machine learning model.

In an embodiment of the present invention, the above signal parameters include a download rate and an upload rate.

A method for predicting user experience degradation of a wireless network of the present invention includes: receiving user feedback information corresponding to the user, user signal quality information, and user website browsing records; counting the number of predefined word strings in the user feedback information to generate The first characteristic data; count the number of predefined websites in the user's website browsing record to generate the second characteristic data; calculate the average value of the signal parameters according to the user signal quality information to generate the third characteristic data; combine the first characteristic data, the second The characteristic data and the third characteristic data are input into at least one machine learning model to generate a prediction result; and output the prediction result corresponding to the user.

Based on the above, the electronic device of the present invention can predict whether user experience degradation occurs in the wireless network according to feedback information such as user feedback information, user signal quality information, or user website browsing records.

FIG. 1 is a schematic diagram of an electronic device 100 for predicting user experience degradation of a wireless network according to an embodiment of the present invention.

The electronic device 100 is, for example, a central processing unit (central processing unit, CPU), or other programmable general purpose or special purpose micro control unit (micro control unit, MCU), microprocessor (microprocessor), digital signal processing Digital signal processor (DSP), programmable controller, application specific integrated circuit (ASIC), graphics processing unit (graphics processing unit, GPU), image signal processor (image signal processor, ISP) ), image processing unit (image processing unit, IPU), arithmetic logic unit (arithmetic logic unit, ALU), complex programmable logic device (complex programmable logic device, CPLD), field programmable logic gate array (field programmable gate array , FPGA) or other similar components or combinations of the above components. The processor 110 can be coupled to the storage medium 120 and the transceiver 120 , and access and execute multiple modules and various application programs stored in the storage medium 120 .

The storage medium 120 is, for example, any type of fixed or removable random access memory (random access memory, RAM), read-only memory (read-only memory, ROM), flash memory (flash memory) , hard disk drive (hard disk drive, HDD), solid state drive (solid state drive, SSD) or similar components or a combination of the above components, and are used to store multiple modules or various application programs executable by the processor 110 . In this embodiment, the storage medium 120 may store at least one machine learning model 121 including one or more machine learning models.

The transceiver 130 transmits and receives signals in a wireless or wired manner. The transceiver 130 may also perform operations such as low noise amplification, impedance matching, frequency mixing, up or down frequency conversion, filtering, amplification, and the like.

The processor 110 can receive feedback information corresponding to the user through the transceiver 130, such as user feedback information, user signal quality information, and user website browsing records, wherein the feedback information can come from the user's customer complaint record or from the customer service of the telecommunications company departmental information. In addition, the processor 110 can access the server of the telecom operator through the transceiver 130, so as to obtain the user signal quality information of the user during a specific time period from the server. The user signal quality information may include one or more signal parameters, where the signal parameters are, for example, download rate or upload rate.

Table 1 is an example of a customer complaint record. The processor 110 can obtain user feedback information from the feedback content in Table 1, and can obtain user website browsing records corresponding to a specific time period from the website browsing records in Table 1. Table 1 user time Feedback content Website Browsing History A 2021/08/09 01:03 Lasting for more than 30 days, indoor 0~2 grid, data-internet speed is slow, the customer said that this location is a company, and the local network reception is very poor for more than 1 month, the signal is weak and the Internet cannot be accessed, this location has multiple phone number requirements Improve the signal, please your organization to help deal with and reply, thank you. The user's website browsing records in the past 30 days

The storage medium 120 can store predefined character strings. The processor 110 can count the number of predefined word strings in the user feedback information to generate the first characteristic data. The predefined character strings in the storage medium 120 are expandable. Table 2 shows examples of predefined strings. Taking Table 1 and Table 2 as an example, in Table 1, the predefined string "over 30 days" appeared once, the predefined string "indoor 0~2 cells" appeared once, and the predefined string ""Indoor 3~5 cells" appeared 0 times, the predefined string "data-slow Internet access" appeared 1 time, and the predefined string "Internet reception is poor" appeared 1 time. Accordingly, the processor 110 can count the number of predefined word strings appearing in the feedback content in Table 1 to generate the first feature data corresponding to user A as shown in Table 3 . Table 2 last for more than 30 days Indoor 0~2 grid Indoor 3~5 grids Data - Slow Internet Internet reception is poor table 3 user last for more than 30 days Indoor 0~2 grid Indoor 3~5 grids Data - Slow Internet Internet reception is poor A 1 1 0 1 1

In one embodiment, the processor 110 may generate a predefined word string according to information fed back by multiple users. Specifically, the processor 110 may receive historical user feedback information through the transceiver 130, wherein the historical user feedback information may include multiple feedback messages respectively corresponding to multiple users. The processor 110 can generate a predefined word string according to the intersection of multiple feedback messages. For example, if there is an intersection of multiple feedback messages containing the string "continuing for more than 30 days". The processor 110 may set the string of "continuing for more than 30 days" as the predefined string.

On the other hand, the storage medium 120 can store predefined websites. The processor 110 can count the number of predefined websites in the user's website browsing record to generate the second characteristic data. The predefined websites in the storage medium 120 are expandable. Table 4 is an example of a predefined website. Assume that in Table 1, User A's "User Browsing Records of Websites in the Last 30 Days" has 1 predefined website "Website A", 1 predefined website "Website B", and 0 predefined websites Website "Site C". Accordingly, the processor 110 can count the number of predefined websites in the "website browsing records of the user in the last 30 days", so as to generate the second feature data corresponding to the user A as shown in Table 5. Table 4 Website A Website B Site C table 5 user Website A Website B Site C A 1 1 0

In one embodiment, the processor 110 may generate a predefined website according to information fed back by multiple users. Specifically, the processor 110 may receive historical user feedback information through the transceiver 130, wherein the historical user feedback information may include a plurality of website browsing records respectively corresponding to a plurality of users. The processor 110 can generate a predefined website according to the intersection of multiple website browsing records. For example, if the intersection of multiple website browsing records includes "Website A". The processor 110 can set "website A" as a predefined website. The processor 110 can calculate the average value of signal parameters according to the signal quality of the user to generate third feature data, wherein the third feature data can include an average download rate or an average upload rate.

After obtaining the first characteristic data, the second characteristic data and the third characteristic data, the processor 110 can input the first characteristic data, the second characteristic data and the third characteristic data into at least one machine learning model 121 to generate prediction results. The prediction result may indicate whether the wireless network user (for example: user A) experiences degradation in user experience. The processor 110 can output the prediction result through the transceiver 130 for the reference of the user of the electronic device 100 (eg, maintenance personnel).

In one embodiment, at least one machine learning model 121 may include a random forest model (random forest), a gradient boost tree (gradient boost tree) or a neural network (neural network). At least one machine learning model 121 can generate at least one probability value according to the first characteristic data, the second characteristic data and the third characteristic data. The processor 110 can generate a prediction result according to at least one probability value.

In one embodiment, the processor 110 may generate a prediction result in response to at least one probability value being greater than a threshold, wherein the prediction result indicates that user experience degradation is about to occur. For example, assume that at least one machine learning model 121 includes one machine learning model, and the threshold is 0.5. At least one machine learning model 121 can generate a probability value according to the first characteristic data, the second characteristic data and the third characteristic data. If the probability value is greater than 0.5, the prediction result generated by the processor 110 may indicate that user experience degradation is about to occur.

In one embodiment, the processor 110 may calculate an average value of at least one probability value, and may generate a prediction result in response to the average value being greater than a threshold, wherein the prediction result indicates that user experience degradation is about to occur. For example, assume that at least one machine learning model 121 includes two machine learning models, and the threshold is 0.5. The two machine learning models can respectively generate probability values according to the first characteristic data, the second characteristic data and the third characteristic data. Processor 110 may calculate an average of the two probability values. If the average value is greater than 0.5, the prediction result generated by the processor 110 may indicate that user experience degradation is about to occur.

In one embodiment, the processor 110 can train at least one machine learning model 121 according to the information previously fed back by the user. The processor 110 may receive historical user feedback information, historical user signal quality information, and historical user website browsing records corresponding to historical user experience degradation through the transceiver 130 . The processor 110 can train at least one machine learning model 121 according to historical user feedback information, historical user signal quality information, and historical user website browsing records.

Specifically, the processor 110 can count the number of predefined character strings in the historical user feedback information to generate the first historical feature data. The processor 110 can count the number of predefined websites in the historical user website browsing records to generate the second historical characteristic data. The processor 110 may calculate historical average values of signal parameters according to historical user signal quality information to generate third historical feature data. After obtaining the first historical characteristic data, the second historical characteristic data and the third historical characteristic data, the processor 110 can use the first historical characteristic data, the second historical characteristic data and the third historical characteristic data as training data to train at least one Machine Learning Models121.

FIG. 2 shows a flow chart of a method for predicting user experience degradation in a wireless network according to an embodiment of the present invention, wherein the method can be implemented by the electronic device 100 shown in FIG. 1 . In step S201, user feedback information corresponding to the user, user signal quality information, and user website browsing record corresponding to the user are received. In step S202, the number of predefined character strings in the user feedback information is counted to generate first characteristic data. In step S203, the number of predefined websites in the user's website browsing record is counted to generate second characteristic data. In step S204, an average value of signal parameters is calculated according to the user signal quality information to generate third feature data. In step S205, the first characteristic data, the second characteristic data and the third characteristic data are input into at least one machine learning model to generate a prediction result. In step S206, the prediction result corresponding to the user is output.

To sum up, the electronic device of the present invention can predict whether user experience degradation occurs in a wireless network according to feedback information such as user feedback information, user signal quality information, or user website browsing records. The electronic device can perform prediction of user experience degradation through machine learning models such as random forest models, gradient boosting trees, or neural networks. Compared with the traditional use of performance indicators of base stations to analyze the faults of base stations, the electronic device of the present invention can generate prediction results of user experience degradation directly related to user feedback, so as to assist wireless network maintenance personnel to eliminate as soon as possible Possible root causes of user experience degradation, thereby improving the customer's user experience.

100: Electronic device 110: Processor 120: storage media 121:At least one machine learning model 130: Transceiver S201, S202, S203, S204, S205, S206: steps

FIG. 1 is a schematic diagram of an electronic device for predicting user experience degradation of a wireless network according to an embodiment of the present invention. FIG. 2 shows a flow chart of a method for predicting user experience degradation in a wireless network according to an embodiment of the present invention.

S201, S202, S203, S204, S205, S206: steps

Claims (10)

An electronic device for predicting user experience degradation of a wireless network, comprising: transceivers; and a processor coupled to the transceiver, wherein the processor is configured to perform: receiving user feedback information corresponding to the user, user signal quality information, and user website browsing records corresponding to the user through the transceiver; counting the number of predefined character strings in the user feedback information to generate first characteristic data; Counting the number of predefined websites in the user's website browsing records to generate second characteristic data; calculating an average value of signal parameters according to the user signal quality information to generate third characteristic data; inputting the first characteristic data, the second characteristic data and the third characteristic data into at least one machine learning model to generate a prediction result; and The prediction result corresponding to the user is output through the transceiver. The electronic device as claimed in claim 1, wherein the processor is further configured to perform: receiving historical user feedback information through the transceiver, wherein the historical user feedback information includes a plurality of feedback messages respectively corresponding to a plurality of users; and The predefined character string is generated according to the intersection of the plurality of feedback messages. The electronic device as claimed in claim 1, wherein the processor is further configured to perform: receiving historical user feedback information through the transceiver, wherein the historical user feedback information includes a plurality of website browsing records respectively corresponding to a plurality of users; and The predefined website is generated according to the intersection of the plurality of website browsing records. The electronic device according to claim 1, wherein the at least one machine learning model includes at least one of the following: random forest model, gradient boosting tree, and neural network. The electronic device according to claim 1, wherein said at least one machine learning model outputs at least one probability value according to said first characteristic data, said second characteristic data and said third characteristic data, wherein said processor The prediction result is generated according to the at least one probability value. The electronic device according to claim 5, wherein the processor generates the prediction result in response to the at least one probability value being greater than a threshold, wherein the prediction result indicates that the user experience degradation is about to occur. The electronic device as claimed in claim 1, wherein the processor is further configured to perform: receiving historical user feedback information corresponding to historical user experience degradation, historical user signal quality information, and historical user website browsing records through the transceiver; and The at least one machine learning model is trained according to the historical user feedback information, the historical user signal quality information, and the historical user website browsing records. The electronic device as claimed in claim 7, wherein the processor is further configured to perform: counting the number of the predefined character strings in the historical user feedback information to generate first historical characteristic data; counting the number of the predefined websites in the historical user website browsing records to generate second historical feature data; calculating historical average values of the signal parameters according to the historical user signal quality information to generate third historical characteristic data; and The at least one machine learning model is trained according to the first historical characteristic data, the second historical characteristic data and the third historical characteristic data. The electronic device as claimed in claim 1, wherein the signal parameters include a download rate and an upload rate. A method for predicting user experience degradation in a wireless network, comprising: Receive user feedback information, user signal quality information and user website browsing records corresponding to the user; counting the number of predefined character strings in the user feedback information to generate first characteristic data; Counting the number of predefined websites in the user's website browsing records to generate second characteristic data; calculating an average value of signal parameters according to the user signal quality information to generate third characteristic data; inputting the first characteristic data, the second characteristic data and the third characteristic data into at least one machine learning model to generate a prediction result; and outputting the prediction result corresponding to the user.

Images