CN112188004B - Obstacle call detection system based on machine learning and control method thereof - Google Patents
Obstacle call detection system based on machine learning and control method thereof Download PDFInfo
- Publication number
- CN112188004B CN112188004B CN202011039833.0A CN202011039833A CN112188004B CN 112188004 B CN112188004 B CN 112188004B CN 202011039833 A CN202011039833 A CN 202011039833A CN 112188004 B CN112188004 B CN 112188004B
- Authority
- CN
- China
- Prior art keywords
- call
- data
- call connection
- quality
- machine learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000010801 machine learning Methods 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000001514 detection method Methods 0.000 title claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 89
- 238000013473 artificial intelligence Methods 0.000 claims abstract description 36
- 238000010606 normalization Methods 0.000 claims description 17
- 230000004888 barrier function Effects 0.000 claims description 14
- 238000013527 convolutional neural network Methods 0.000 claims description 13
- 238000013480 data collection Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 description 15
- 238000004891 communication Methods 0.000 description 13
- 230000008569 process Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000000903 blocking effect Effects 0.000 description 5
- 239000000284 extract Substances 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/22—Arrangements for supervision, monitoring or testing
- H04M3/2218—Call detail recording
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/22—Arrangements for supervision, monitoring or testing
- H04M3/2236—Quality of speech transmission monitoring
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/22—Arrangements for supervision, monitoring or testing
- H04M3/2272—Subscriber line supervision circuits, e.g. call detection circuits
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Medical Informatics (AREA)
- Telephonic Communication Services (AREA)
Abstract
The invention discloses a machine learning-based obstacle call detection system and a control method thereof. The control method of the obstacle call detection system according to the present invention includes: a step of collecting basic call connection processing data including call quality per unit time for each call connection; generating a distribution map of call quality based on the call quality per unit time included in the basic call connection processing data; determining each parameter of an artificial intelligence system by performing machine learning using at least one piece of information included in the basic call connection processing data and the generated call quality profile as input values; and a step of judging an obstacle call by applying at least one of real-time call connection data and CDR data to the artificial intelligence system.
Description
Technical Field
The present invention relates to a system and a method for detecting a fault call, and more particularly, to a system and a method for detecting a fault call based on machine learning.
Background
With the recent development of communication technology, wired, internet phone, wireless phone, etc. are widely used, and the communication environment has changed rapidly, but it is impossible to add appropriate equipment or for various other reasons, and thus, communication service users (clients) have been dissatisfied with the quality of communication.
However, conventionally, when detecting an obstacle call that is a state in which the communication quality is degraded, manual work by the operator of the communication service company is actually relied upon.
If the barrier call is found, it needs to be quickly investigated and solved to bring more customers and increase customer loyalty, but the manual work performed by the operator is not quick and correct.
That is, in the case of the above-described cases, the determination of whether or not a call is a fault call is not accurate simply by manually analyzing a conventional Call Detail Record (CDR) generated for each call connection, because countless communication environment changes and the call quality changes every moment during the period when a call is started after the call connection.
(prior art document) Korean laid-open patent No. 10-2009-0008948
Disclosure of Invention
Technical problem to be solved by the invention
The present invention has been made to solve the above-described conventional problems, and an object of the present invention is to provide a system and a control method thereof for calculating appropriate statistical data based on collected call connection processing-related data and detecting a faulty call by machine learning.
Technical scheme for solving problems
In order to achieve the above object, an obstructed-call detection system according to the present invention includes: a data collection unit that collects basic call connection processing data including call quality per unit time for each call connection; a statistical data generating unit that generates a distribution map of call quality based on the call quality per unit time included in the basic call connection processing data; a machine learning processing unit that determines each parameter of the artificial intelligence system by performing machine learning using, as input values, at least one piece of information included in basic call connection processing data collected from the basic call connection processing data and the distribution map of call quality generated by the statistical data generation unit; and a determination unit which determines an obstacle call by applying at least one of real-time call connection data and CDR (CallDetailRecord) data to the artificial intelligence system.
In order to achieve the above object, a method for controlling an obstacle call detection system according to the present invention includes: a step of collecting basic call connection processing data including call quality per unit time for each call connection; generating a distribution map of call quality based on the call quality per unit time included in the basic call connection processing data; determining each parameter of an artificial intelligence system by performing machine learning using at least one piece of information included in the basic call connection processing data and the generated call quality profile as input values; and a step of judging an obstacle call by applying at least one of real-time call connection data and CDR (Call Detail record) data to the artificial intelligence system.
Drawings
Fig. 1 is a schematic configuration diagram of an overall system including an obstruction call detection system of an embodiment of the present invention;
FIG. 2 is a functional block diagram of the obstruction call detection system of FIG. 1;
FIG. 3 is a diagram showing an example of data that is available in real time for a network call;
fig. 4 is a diagram showing an example of data in which statistical data generated by the obstacle call detection system according to the embodiment of the present invention is added to the data in fig. 3;
fig. 5 is a diagram showing an example of a call quality distribution map;
FIG. 6 is a diagram showing a CNN processing structure;
FIG. 7 is a diagram illustrating a process of cumulatively storing data available in real time for a network call and performing machine learning based on the cumulatively stored data;
fig. 8 is a diagram showing a procedure of accumulating necessary data in CDR data stored after a PSTN call is made, and performing machine learning based on the accumulated data.
Best mode for carrying out the invention
The present invention is described in detail below with reference to the accompanying drawings.
The alphabet of '(a)' and the like is included in a portion of the claims for convenience, but does not specify the order of the steps.
A schematic configuration of an overall system including the obstruction call detection system 100 of an embodiment of the present invention is shown in fig. 1.
In this drawing, the calling terminal 300 is a terminal for making a call to a counterpart, and the called terminal 400 is a terminal for receiving a call from a counterpart.
As described above, the call connection processing system 200 is present in the middle of the communication path between the calling terminal 300 and the called terminal 400, and the call connection processing system 200 functions to receive the call connection invitation from the calling terminal 300, perform the authentication processing and the call connection processing with the called terminal 400, and further perform the function of managing the call connection between the calling terminal 300 and the called terminal 400 and the information related to the call, that is, the call connection processing data.
As described above, the procedure of making an outgoing call, making an incoming call, making a call connection, and storing call connection processing data among the calling terminal 300, the called terminal 400, and the call connection processing system 200 corresponds to a known technique, and thus, detailed description thereof is omitted.
The barrier call detection system 100 is configured to determine whether or not each call connection is a barrier call after a predetermined time has elapsed from the start of call connection or after the call connection is established after communication with the call connection processing system 200.
In particular, in order to detect a blocking call in real time, a device for packet mirroring or port mirroring between the call connection processing system 200 and the blocking call detection system may be formed, which is also a well-known technology, and thus, a detailed description thereof will be omitted.
In particular, the barrier call detection system 100 functions to detect whether a barrier call is performed after machine learning is performed.
Fig. 2 shows an example of detailed functional blocks of the above-described obstacle call detection system 100.
As shown in the drawing, the obstruction call detection system 100 includes the following structural formation: a data collection unit 110, a data normalization processing unit 130, a statistical data generation unit 120, a machine learning processing unit 140, and a determination unit 150.
First, the data collection unit 110 functions to collect basic call connection processing data on each call connection.
Here, the Call connection processing data may be real-time data of a starting point at which a Call connection occurs, or Call Detail Record (CDR) data stored after the Call connection is terminated.
In particular, the basic call connection processing data collected by the data collection unit 110 may include call quality per unit time, and may further include a caller identification number, a call start time, a call end time, a call time, and the like.
The basic call connection processing data collected by the data collection unit 110 may include customer request information corresponding to the corresponding call connection, which is information indicating whether or not there is a customer's dissatisfaction or dissatisfaction with respect to the call quality of the corresponding call connection.
Here, the call quality per unit time refers to the call quality measured at a predetermined time interval, and for example, if the call time is 100 seconds, the predetermined time interval is 20 seconds, and a total of 5 call qualities are measured.
Here, the call quality may be at least one of lq (listing quality) which is a value not considering delay occurring during a call and cq (conversion quality) which is a value considering delay occurring during a call, and the measurement value of the call quality and the measurement method thereof are in accordance with a known technique, and thus, a detailed description thereof will be omitted.
However, in this embodiment, the speech quality can distinguish between the uplink and the downlink.
Of course, the call quality may be generated separately from the other components of the present embodiment, but the present embodiment is included in the basic call connection processing data as an example.
Also, the caller identification number and the callee identification number may include identification numbers about devices located on the communication path in addition to identification numbers about respective call partner terminals.
That is, the caller identification number may include not only the identification number of the calling terminal 300 (for example, the telephone number of the calling terminal 300) but also the identification number of the device corresponding to the caller, and the caller identification number may include not only the identification number of the called terminal 400 (for example, the telephone number of the called terminal 400) but also the identification number of the device corresponding to the caller.
Therefore, even if the call quality is the same, if the respective communication paths are different, the machine learning is executed including the case where the degree of the client's perception is different.
The statistical data generator 120 functions to generate a distribution map of call quality based on the call quality per unit time included in the basic call connection processing data.
More specifically, the statistical data generating unit 120 selects a preset call quality section for each call quality per unit time included in the basic call connection processing data, and generates a distribution map of the call quality using a ratio of each call quality section for the entire corresponding call connection.
For example, when the call quality 100 is taken as a reference, the call quality interval is divided into 5 stages: that is, when the number is less than 30, more than 30 to less than 50, more than 50 to less than 70, more than 70 to less than 90, or more than 90, the statistical data generating unit 120 determines to which of the 5 stages the call quality measured per unit time, for example, at a preset time such as 100ms, 1s, or 10 seconds, belongs, and determines the ratio of the call quality measured for the entire call to each of the sections, thereby generating a distribution map of the call quality.
Fig. 3 shows an example of basic call connection processing data, and fig. 4 shows a state in which data relating to a profile of call quality is added by the above-described basic call connection processing data. That is, the data added to fig. 3 is shaded in fig. 4.
Fig. 3 and 4 show the data ethical structure, and do not necessarily correspond to the database tables included in a specific database, and at least a part of the data is stored or generated in another form in a memory or the like.
Fig. 5 is a graph showing a call quality distribution diagram for each of the uplink and the downlink.
Referring to fig. 5, the overall call quality per unit time collected for the uplink and the downlink is 11.11% for less than 30, 11.11% for more than 30 and less than 50, 0% for more than 50 and less than 70, 44.44% for more than 70 and less than 90, and 33.33% for more than 90.
Fig. 5 shows an example in which the uplink and the downlink are both the same profile, but the call quality profiles of the uplink and the downlink may be different depending on the communication environment.
The data normalization processing unit 130 functions to normalize at least one piece of information included in the basic call connection processing data collected by the data collection unit 110 and the distribution map of the call quality generated by the statistical data generation unit 120 so as to have the same length, according to a preset algorithm.
For example, the data normalization processing section 130 functions to extract or receive pre-stored call connection processing data and the generated call quality distribution map, and to perform normalization processing according to a preset algorithm for each field so as to have the same length.
Here, the data normalization processing unit 130 may extract the call connection processing data when the data is stored by itself, or may receive the data from the external server 200.
In particular, the data normalization processing unit 130 normalizes the call connection processing data and the call quality distribution map by field, for example, when the normalized size of the caller id field is 15 bits and the caller id included in the call connection processing data is' 010-.
In particular, as described above, the data normalization processing unit 130 normalizes not only the extracted and received basic call connection processing data but also the distribution map of the call quality generated by the statistical data generation unit 120.
The machine learning processing unit 140 has a function of determining each parameter of the artificial intelligence system by performing machine learning, as input values, of at least one piece of information included in the basic call connection processing data collected from the basic call connection processing data and the distribution map of the call quality generated from the statistical data generating unit 120.
In particular, as described above, after the normalization processing section performs normalization processing for each field, the machine learning processing section 140 performs machine learning using the normalized basic call connection processing data and the normalized profile of call quality as input values, and determines each parameter of the artificial intelligence system.
That is, the artificial intelligence system, particularly, deep learning, which is a kind of machine learning, may differ in the result according to the parameter values of each layer forming the neural network, and the function of the machine learning processing unit 140 is to determine the parameter values of each layer through machine learning and reflect them to the corresponding artificial intelligence system.
The process of machine learning corresponds to the process of calculating the parameter values (for example, row and column values) of each layer of the artificial intelligence system as described above, and is a known technique, and therefore, a detailed description thereof will be omitted.
The machine learning processing unit 140 performs a function peculiar to the machine learning processing, that is, forms data normalized by the data normalization processing unit into one-dimensional image data, performs machine learning on the corresponding one-dimensional image data by the convolutional Neural network CNN (volumetric Neural network), determines each parameter of the CNN, and reflects the parameter.
As an example of a process of forming a one-dimensional image, the data normalization processing unit 130 will be described using a part of the data in fig. 4, in a case where the total call quality per unit time collected in correspondence with the CALLER IP (CALLER _ IP), the called IP (CALLER _ IP), the uplink is less than 30, more than 30 to less than 50, more than 50 to less than 70, more than 70 to less than 90, or more than 90, and in a case where the total call quality per unit time collected in correspondence with the downlink is less than 30, more than 30 to less than 50, more than 50 to less than 70, more than 70 to less than 90, or more than 90, the data normalization processing unit generates normalization data, i.e., 121111000001, 3944, 3933, 121111000001, 3933, for each of the data in this order, and generates normalization data for each of the data, i.e., for each of the cases where the data is greater than 30, greater than 50, less than 70, less than 90, greater than 90, and is equal to the case where the data is generated in this order of '121.111.0.1', '212.0.0.112', '11.e.,' 11., '212000000112', '1111', '1111', '1111', '0000', '4444', '3333', '1111', '1111', '0000', '4444', '3333', and generates data '1211110000012120000001121111111100004444333311111111000044443333' in which the above normalized data are connected side by side in a row, and one-dimensional image thereof is digitized.
Here, the one-dimensional image is an image in which pixels are connected to each other only in one direction (for example, in the horizontal direction) and are not connected to each other in the other direction (for example, in the vertical direction).
As described above, the machine learning processing unit 140 that performs the one-dimensional imaging process performs machine learning on the corresponding one-dimensional image by the CNN, determines and reflects the parameters of the CNN.
Fig. 6 shows a process of processing a one-dimensional image having n pixels by applying the one-dimensional image to the CNN algorithm.
Referring to fig. 6, as an AI (artificial intelligence) model, a neural network is configured by an Input (Input) which inputs data encoded in a normalized manner with respect to the above-described real-time data or CDR data and a call quality profile, a Layer (Layer), a prediction (Predict Result), an actual value (Target), a Loss Function (Loss Function), and an Optimizer (Optimizer).
The Layer (Layer) is a Layer constituting a neural network, and is modeled in a manner most suitable for a one-dimensional convolutional Layer (Layer1 Dimension CNN) algorithm in order to be suitable for real-time data processing, CDR data processing, and call quality profile processing.
A Loss Function (Loss Function) is an important component defining a feedback signal used in learning, and according to a deep learning guideline, Binary Cross entropy (Binary Cross entropy) is applied if classified into two categories, category Cross entropy (category Cross entropy) is applied if classified into various categories, mean square error is applied if regression is applied, Connection Temporal Classification (CTC) is applied if sequence is applied, and there are many categories, and thus category Cross entropy (category Cross entropy) is applied.
An Optimizer (Optimizer) is a component for determining a learning progression method, determines a weight update (update) of a neural network based on a loss function, and applies a Stochastic Gradient Descent (SGD).
In order to make the CNN model, data subjected to the normalization operation as described above is arranged in a one-dimensional image, and an optimum value for the layer parameter is derived by machine learning of repeatedly performing convolution operation on the image.
In addition, when performing the machine learning, the machine learning processing unit 140 determines whether or not the obstacle call is determined based on the client request information included in the basic call connection processing data, in order to indicate that information (AI _ LABEL in fig. 4) about whether or not the obstacle call is determined is necessary.
That is, in the case where the machine learning of the present embodiment is performed to determine various layer parameters for the obstacle call determination, and the estimated result (ResultY' in fig. 6) of the data input at the time of the machine learning is compared with the obstacle call determination value (TargetY in fig. 6) to gradually predict the direction in which the difference decreases, the obstacle call determination value can be determined based on the client request information. That is, if there is a customer request, it is determined as a barrier call.
As another example, the device learning processing unit 140 bases the rate of the barrier call determination value required for performing the device learning on the basis of the standard quality or less (hereinafter referred to as "low quality rate") preset in the call quality per unit time.
For example, when a call quality value is stored at 20-second intervals for a call connection of 10 minutes, if the measurement rate of the call quality of a preset value (for example, 50 minutes) or less is 15% or more based on 100 minutes, it can be determined as a barrier call for the corresponding call connection.
Further, the machine learning processing unit 140 may determine the obstacle call determination by combining the above-described customer request information and the low quality ratio.
For example, when the administrator inputs the customer request information at the upper, middle, and lower levels, the administrator combines the low quality ratio and the scores assigned to the upper, middle, and lower levels at a predetermined ratio to determine the obstacle call determination.
In the case of the configuration shown in fig. 6, the machine learning process by the CNN algorithm corresponds to a known technique, and therefore, a more detailed description thereof will be omitted.
The determination unit 150 is configured to apply at least one of real-time call connection data and cdr (call Detail record) data to an artificial intelligence system to determine a blocking call.
That is, as described above, after determining each parameter of the artificial intelligence system through machine learning, the determination unit 150 can determine whether or not the call is a barrier call by transmitting real-time call connection data or CDR data as an input value to the corresponding artificial intelligence system.
In particular, the determination part 150 may perform a differentiated process according to a process of a call connection manner, extract call connection data in real time if the call connection is a connection through a Network, apply the corresponding call connection data to the artificial intelligence system to determine a blocking call, and apply the stored CDR information to the artificial intelligence system to determine a blocking call if the call connection is a connection through a Public Switched Telephone Network (PSTN).
Fig. 7 and 8 show the overall processing method according to the above-described call connection method.
Fig. 7 shows a process when a SIP call connection occurs through the network.
Referring to the drawing, if a packet transmitted to and received from the call connection processing system 200 is received through a packet mirror, the obstacle call detection system 100 extracts real-time call connection related information from a real-time extraction module of a real-time data extraction block and transmits the same to an AI detection block, and the AI detection block applies the same to a constructed AI model (i.e., an artificial intelligence system that determines and reflects the above parameters) to detect an obstacle call, and additionally performs machine learning processing for corresponding real-time data.
The additional execution of the machine learning process described above means an upgrade of parameters for the artificial intelligence system, thereby enabling continuous tracking management even if the barrier call pattern changes.
Fig. 8 shows a process when a call connection occurs through the PSTN.
Referring to the drawing, the CDR collecting block of the obstruction call detection system 100 periodically collects the cumulatively stored CDR data after the occurrence of the PSTN call connection and transfers it to the AI detection block, and the AI detection block applies the above CDR data to the constructed AI model to detect an obstruction call, and additionally performs machine learning processing for the corresponding CDR data.
It is to be understood that the processes of the above-described embodiments may be executed by a program or an application program stored in a predetermined recording medium (for example, a computer-readable medium). The recording medium includes all of electronic recording media such as ram (random Access memory), magnetic recording media such as hard disks, and optical recording media such as cd (compact disk).
In this case, the program stored in the recording medium may be executed by hardware such as a computer or a smart phone to execute the above-described embodiments. In particular, at least one of the functional blocks of the obstacle call detection system (100) of the present invention described above may be implemented by a program or an application program as described above.
The present invention is not limited to the specific embodiments described above, and various modifications and alterations may be made without departing from the scope of the present invention.
In particular, although the above-described embodiment has been described mainly in the case where a call is connected, it goes without saying that the case where a call fails without being normally connected may be included. At this time, the communication quality information per unit time can fill the blank.
Industrial applicability of the invention
As described above, according to the present invention, it is possible to improve the accuracy of the obstacle call detection and automate the machine learning of the obstacle call pattern, thereby enabling the obstacle call detection even if the form of the obstacle call changes.
In particular, when performing machine learning, the speech time of each speech connection has high variability, so if the speech quality per unit time of the original each speech connection is utilized, the accuracy of the artificial intelligence module generated as a result of machine learning is greatly reduced, therefore, when the profile of speech quality is first generated based on the speech quality per unit time as in the present invention, normalization processing can be consistently performed, and the accuracy of the artificial intelligence module generated as a result of machine learning is improved.
Further, the accuracy of the artificial intelligence module can be improved by performing machine learning based on the obstacle call determination while considering the customer request information and the low quality ratio.
Claims (4)
1. A method for controlling a barrier call detection system, comprising:
(a) a step of collecting basic call connection processing data including call quality per unit time for each call connection;
(b) generating a distribution map of call quality based on the call quality per unit time included in the basic call connection processing data;
(c) determining parameters of an artificial intelligence system by performing machine learning using, as input values, at least one of information included in the basic call connection processing data in the step (a) and the call quality profile generated in the step (b);
(d) a step of judging an obstacle call by applying at least one of real-time call connection data and CDR (Call Detail record) data to the artificial intelligence system;
before the step (c), further comprising:
(e) a step of normalizing at least one information included in the basic call connection processing data of the step (a) and the call quality distribution map generated from the step (b) according to a preset algorithm so as to have the same length,
in the step (c), machine learning is performed using the normalized data as an input value to determine each parameter of the artificial intelligence system;
the step (b) comprises:
(b1) a step of selecting a preset call quality interval for the call quality of each unit time included in the basic call connection processing data;
(b2) generating a distribution map of the call quality by using the ratio of each call quality interval for the entire corresponding call connection;
the basic call connection processing data of the step (a) includes customer request information corresponding to the corresponding call connection, and the barrier call determination required when performing the machine learning of the step (c) is based on the customer request information included in the basic call connection processing data and a ratio of the customer request information to the call quality per unit time which is equal to or lower than a preset reference quality.
2. The method for controlling a barrier call detection system according to claim 1,
and (c) forming the data normalized in the step (e) into one-dimensional image data, and then performing machine learning according to a Convolutional Neural Network (CNN) for the corresponding one-dimensional image data to determine each parameter of the CNN.
3. The method for controlling a barrier call detection system according to claim 1,
the basic call connection processing data in the step (a) includes a caller identification number, a call start time, a call end time, and call quality at each time.
4. An obstructed call detection system, comprising:
a data collection unit that collects, for each call connection, basic call connection processing data including call quality per unit time and client request information corresponding to the call connection;
a statistical data generating unit that generates a distribution map of call quality based on the call quality per unit time included in the basic call connection processing data; the statistical data generating unit selects preset call quality sections for each call quality per unit time included in the basic call connection processing data, and generates a distribution map of the call quality by using a ratio of each call quality section for the entire corresponding call connection;
a machine learning processing unit that determines each parameter of the artificial intelligence system by performing machine learning using, as input values, at least one piece of information included in basic call connection processing data collected from the basic call connection processing data and the distribution map of call quality generated by the statistical data generation unit;
a determination unit that applies at least one of real-time call connection data and CDR data to the artificial intelligence system to determine an obstacle call, the obstacle call determination by the determination unit being based on client request information included in the basic call connection processing data and a ratio of a preset reference quality or less of the call quality per unit time;
further comprises a normalization processing unit: which normalizes at least one information included in the basic call connection processing data collected from the data collection unit and the distribution map of the call quality generated from the statistical data generation unit so as to have the same length according to a preset algorithm,
the machine learning processing unit determines each parameter of the artificial intelligence system by performing machine learning using the normalized data as an input value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011039833.0A CN112188004B (en) | 2020-09-28 | 2020-09-28 | Obstacle call detection system based on machine learning and control method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011039833.0A CN112188004B (en) | 2020-09-28 | 2020-09-28 | Obstacle call detection system based on machine learning and control method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112188004A CN112188004A (en) | 2021-01-05 |
| CN112188004B true CN112188004B (en) | 2022-04-05 |
Family
ID=73944480
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011039833.0A Expired - Fee Related CN112188004B (en) | 2020-09-28 | 2020-09-28 | Obstacle call detection system based on machine learning and control method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112188004B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115002743A (en) * | 2022-06-28 | 2022-09-02 | 广西东信易通科技有限公司 | Median scene call enhancement system based on machine learning algorithm |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1414729A (en) * | 2001-10-24 | 2003-04-30 | 興和株式会社 | Network speech quality valuator device |
| CN1797969A (en) * | 2004-12-29 | 2006-07-05 | 乐金电子(中国)研究开发中心有限公司 | Communication quality adjusting device and method of portable terminal |
| CN105794187A (en) * | 2013-11-15 | 2016-07-20 | 微软技术许可有限责任公司 | Predicting call quality |
| CN107027023A (en) * | 2017-04-24 | 2017-08-08 | 北京理工大学 | VoIP based on neutral net is without reference video communication quality method for objectively evaluating |
| CN107408384A (en) * | 2015-11-25 | 2017-11-28 | 百度（美国）有限责任公司 | The end-to-end speech recognition of deployment |
| CN107995370A (en) * | 2017-12-21 | 2018-05-04 | 广东欧珀移动通信有限公司 | Call control method, device and storage medium and mobile terminal |
| CN110135633A (en) * | 2019-04-29 | 2019-08-16 | 北京六捷科技有限公司 | A kind of railway service Call failure prediction technique and device |
| US10757256B1 (en) * | 2019-06-03 | 2020-08-25 | RingDNA, Inc. | Adaptive real-time conversational systems and methods |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10339456B2 (en) * | 2017-01-05 | 2019-07-02 | T-Mobile Usa, Inc. | Machine learning-based troubleshooting of VoLTE calls |
| CN109995924A (en) * | 2017-12-30 | 2019-07-09 | 中国移动通信集团贵州有限公司 | Cheat phone recognition methods, device, equipment and medium |
| WO2019177481A1 (en) * | 2018-03-12 | 2019-09-19 | Ringcentral, Inc., (A Delaware Corporation) | System and method for evaluating the quality of a communication session |
| CN109302262B (en) * | 2018-09-27 | 2020-07-10 | 电子科技大学 | Communication anti-interference method based on depth determination gradient reinforcement learning |
| CN109829729A (en) * | 2019-01-18 | 2019-05-31 | 青牛智胜(深圳)科技有限公司 | A kind of intelligence outgoing call system and method |
-
2020
- 2020-09-28 CN CN202011039833.0A patent/CN112188004B/en not_active Expired - Fee Related
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1414729A (en) * | 2001-10-24 | 2003-04-30 | 興和株式会社 | Network speech quality valuator device |
| CN1797969A (en) * | 2004-12-29 | 2006-07-05 | 乐金电子(中国)研究开发中心有限公司 | Communication quality adjusting device and method of portable terminal |
| CN105794187A (en) * | 2013-11-15 | 2016-07-20 | 微软技术许可有限责任公司 | Predicting call quality |
| CN107408384A (en) * | 2015-11-25 | 2017-11-28 | 百度（美国）有限责任公司 | The end-to-end speech recognition of deployment |
| CN107027023A (en) * | 2017-04-24 | 2017-08-08 | 北京理工大学 | VoIP based on neutral net is without reference video communication quality method for objectively evaluating |
| CN107995370A (en) * | 2017-12-21 | 2018-05-04 | 广东欧珀移动通信有限公司 | Call control method, device and storage medium and mobile terminal |
| CN110135633A (en) * | 2019-04-29 | 2019-08-16 | 北京六捷科技有限公司 | A kind of railway service Call failure prediction technique and device |
| US10757256B1 (en) * | 2019-06-03 | 2020-08-25 | RingDNA, Inc. | Adaptive real-time conversational systems and methods |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112188004A (en) | 2021-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110781433B (en) | Data type determining method and device, storage medium and electronic device | |
| US9837078B2 (en) | Methods and apparatus for identifying fraudulent callers | |
| US7920677B2 (en) | Automated DSL performance adjustment | |
| CN109905696B (en) | Method for identifying video service experience quality based on encrypted flow data | |
| US7272209B2 (en) | Automated DSL performance adjustment | |
| US20160125884A1 (en) | Utilizing Voice Biometrics | |
| CN110147803B (en) | User loss early warning processing method and device | |
| CN101771758A (en) | Dynamic determine method for normal fluctuation range of performance index value and device thereof | |
| CN112702224B (en) | Method and device for analyzing quality difference of home broadband user | |
| CN107481090A (en) | A kind of user's anomaly detection method, device and system | |
| CN111385420B (en) | User identification method and device, storage medium and electronic device | |
| US20080270225A1 (en) | Apparatus and Methods for Determining Availability and Performance of Entities Providing Services in a Distributed System Using Filtered Service Consumer Feedback | |
| CN112671573B (en) | Method and device for identifying potential off-network users in broadband service | |
| CN110417747B (en) | Method and device for detecting violent cracking behavior | |
| CN117076258A (en) | Remote monitoring method and system based on Internet cloud | |
| CN112188004B (en) | Obstacle call detection system based on machine learning and control method thereof | |
| CN112567720A (en) | Fraud call detection system based on machine learning and control method thereof | |
| CN108076197B (en) | Method and device for detecting terminal network performance degradation | |
| CN108990092B (en) | Communication abnormity positioning method and device | |
| CN108306846A (en) | A kind of network access exception detection method and system | |
| CN114841705B (en) | Anti-fraud monitoring method based on scene recognition | |
| Vega et al. | Cognitive no-reference video quality assessment for mobile streaming services | |
| CN115373868A (en) | Intelligent data processing system and method | |
| KR102363432B1 (en) | System for detecting error occurrence call using machine learning and control method thereof | |
| CN108521435B (en) | Method and system for user network behavior portrayal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220405 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |