TR2023019940A2

TR2023019940A2 - A SYSTEM THAT ESTIMATES THE FAILURE END TIME WITH BAYESIAN NON-NEGATIVE MATRIX FACTORIZATION (BNMF) AND TRANSMITS IT TO THE USER THROUGH THE CHAT ROBOT

Info

Publication number: TR2023019940A2
Application number: TR2023/019940
Authority: TR
Inventors: Bi̇ngol Ezgi̇
Original assignee: Turkcell Teknoloji̇ Araştirma Ve Geli̇şti̇rme Anoni̇m Şi̇rketi̇
Filing date: 2023-12-30
Publication date: 2024-01-22

Abstract

Bu buluş, internet bağlantı cihazlarının tarihsel alarm sonlanma verilerinin Bayesian Non-Negative Matrix Factorization (BNMF) algoritmalarıyla analiz edilmesiyle güncel alarmların sonlanma sürelerinin tahminlenmesini, internet bağlantı cihazını kullanan abonenin bir arayüz üzerinden bağlantısında problem olduğunu girmesiyle sonucunda yapılan işlemlerle arıza alarmın sonlanma tahmin süresinin kullanıcıya ilgili arayüz üzerinden gösterilmesini sağlamak amacıyla geliştirilen bir sistem (1) ile ilgilidir.This invention aims to estimate the termination times of current alarms by analyzing the historical alarm termination data of internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms, and to determine the termination time of the fault alarm according to the user by the transactions made as a result of the subscriber using the internet connection device entering that there is a problem in his connection through an interface. It is related to a system (1) developed to display it over the interface.

Description

TARIFNAME ARIZA SONLANMA SÜRESINI BAYESIAN NON-NEGATIVE MATRIX FACTORIZATION (BNMF) ILE TAHMINLEYEN VE SOHBET ROBOTU ÜZERINDEN KULLANICIYA ILETEN BIR SISTEM Teknik Alan Bu bulus, internet baglanti cihazlarinin tarihsel alarm sonlanma verilerinin Bayesian Non-Negative Matrix Factorization (BNMF) algoritmalariyla analiz edilmesiyle güncel alarmlarin sonlanma sürelerinin tahminlenmesini, internet baglanti cihazini kullanan abonenin bir sohbet robotu (chatbot) arayüzüne baglantisinda problem oldugunu girmesiyle sonucunda yapilan islemlerle ariza alarmin sonlanma tahmin süresinin kullaniciya ilgili sohbet robotu üzerinden gösterilmesini saglayan bir sistem ile ilgilidir. Önceki Teknik Günümüzde internet saglayici firmalar müsterilerine kaliteli internet hizmeti verebilmek için baglanti cihazlarini sürekli aktif ve çalisir halde tutmak istemektedirler. Internet saglayici firmalarin sahip olduklari cihazlar çesitli durumlarda ariza verebilmektedir. Söz konusu arizalarin en kisa sürede çözülmesi ve kesintisiz bir hizmet saglanmasi gerekmektedir. Bu sebeple internet baglanti cihazlarinin tarihsel alarm sonlanma verilerinin Bayesian Non-Negative Matrix Factorization (BNMF) algoritmalariyla analiz edilmesiyle güncel alarmlarin sonlanma sürelerinin tahminlenmesini, internet baglanti cihazini kullanan abonenin bir sohbet robotu (chatbot) arayüzüne baglantisinda problem oldugunu girmesiyle sonucunda yapilan islemlerle ariza alarmin sonlanma tahmin süresinin kullaniciya ilgili sohbet robotu üzerinden gösterilmesini saglayan bir sisteme ihtiyaç duyuldugu anlasilmaktadir. dokümaninda kablosuz 5G agi gibi bir kablosuz haberlesme aginda hücresel haberlesme kesintisi sirasinda kullanici deneyiminin yönetilmesi için sistemler ve yöntemler açiklanmaktadir. Örnek düzenlemeler ag haberlesmesi hata testinin dahil oldugu hücresel haberlesme agi kesinti sirasinda kullanici deneyiminin yönetilmesi için sistemler ve yöntemler içermekte, belirli telekomünikasyon agi bilesenleri ve belirli ag baglantilari hatalari arasindaki iletisimdeki hatalarin tespit edilmesine yanit olarak farkli kurtarma aksiyonlarinin gerçeklestirilmesi, dagitik birimdeki radyo kaynak yöneticisini dahil eden hücresel telekomünikasyon agi kesintisi sirasinda kullanici denetiminin yönetilmesi, dagitik birim (DU) ile iletisim halindeki gelecek haberlesme arizalariyla ilgili tahminleri yapan akilli katman modülünü içeren hücresel telekomünikasyon agi arizalari/kesinti sirasinda kullanici deneyiminin yönetilmesi, yedek veri merkezinden faydalanilarak hücresel telekomünikasyon agi kesintisi durumunda kullanici deneyiminin yönetilmesi, hücresel telekomünikasyon agi kesintisi sirasinda kullanici deneyiminin yönetilmesi için yeni eklenecek merkezi birimlere izin verilmesi için yeniden boyutlandirilacak olan merkezi birimlere (CU) karsilik gelen bir mikro servis grubunun kapsanmasi seklindedir. Bulusun Kisa Açiklamasi Bu bulusun amaci, internet baglanti cihazlarinin tarihsel alarm sonlanma verilerinin Bayesian Non-Negative Matrix Factorization (BNMF) algoritmalariyla analiz edilmesiyle güncel alarmlarin sonlanma sürelerinin tahminlenmesini, internet baglanti cihazini kullanan abonenin bir sohbet robotu (chatbot) arayüzüne baglantisinda problem oldugunu girmesiyle sonucunda yapilan islemlerle ariza alarmin sonlanma tahmin süresinin kullaniciya ilgili sohbet robotu üzerinden gösterilmesini saglamak amaciyla gelistirilen bir sistem gerçeklestirmektir. Bulusun Ayrintili Açiklamasi Bu bulusun amacina ulasmak için gelistirilen "Ariza Sonlanma Süresini Bayesian Non- Negative Matrix Factorization (BNMF) ile Tahminleyen ve Sohbet Robotu Üzerinden Kullaniciya Ileten Bir Sistem" ekli sekilde gösterilmis olup, bu sekil; Sekil-1; Bulus konusu sistemin sematik görünüsüdür. Sekilde yer alan parçalar tek tek numaralandirilmis olup, bu numaralarin karsiliklari asagida verilmistir. 1. Sistem 2. Sohbet Robotu (Chatbot) Uygulamasi 3. Veri Tabani 4. Sunucu Internet baglanti cihazlarinin tarihsel alarm sonlanma verilerinin Bayesian Non- Negative Matrix Factorization (BNMF) algoritmalariyla analiz edilmesiyle güncel alarmlarin sonlanma sürelerinin tahminlenmesini, internet baglanti cihazini kullanan abonenin bir sohbet robotu (chatbot) arayüzüne baglantisinda problem oldugunu girmesiyle sonucunda yapilan islemlerle ariza alarmin sonlanma tahmin süresinin kullaniciya ilgili sohbet robotu üzerinden gösterilmesini saglamak amaciyla gelistirilen bulus konusu sistem (1); -internet hizmeti kullanicilarinin ariza bilgilendirmelerine erismelerini saglamak üzere yapilandirilan en az bir sohbet robotu uygulamasi (2), -internet erisim hizmetini saglayan firma ile baglanti kurarak kullanici verilerini kayit altinda tutmak üzere yapilandirilan en az bir veri tabani (3), -alarm bilgilerini kontrol ederek alarmin planli bir kesinti ya da bilinen bir ariza olup olmadigini sorgulamak, sorgulama sayesinde planli kesintinin ve/Veya bilinen arizanin oldugu tespit edildiginde veri tabani (2) üzerinde yer alan ekiplerin kesinti ve/Veya ariza için bildirdigi tahmini sonlanma süresi bilgisine erisip sohbet robotu uygulamasi (2) üzerinden kullanicinin bilgilendirilmesini saglamak üzere yapilandirilan en az bir sunucu (4) içermektedir. Bulus konusu sistemde (1) yer alan sohbet robotu uygulamasi (2), herhangi bir iletisim protokolü kullanarak sunucu (4) ile iletisim kurmak üzere yapilandirilmaktadir. Sohbet robotu uygulamasi (2), herhangi bir iletisim protokolü kullanarak sunucu (4) ile veri alisverisi gerçeklestirmek üzere yapilandirilmaktadir. Sohbet robotu uygulamasi (2), kullanicinin bir arayüz üzerinden giris islemlerini gerçeklestirerek ariza sonlanma bilgisi talep edebilmesini saglamak üzere yapilandirilmaktadir. Sohbet robotu uygulamasi (2), internet hizmet kullanicilarinin bir arayüz üzerinde ariza bilgilendirmesi almalarini saglamak üzere yapilandirilmaktadir. Sohbet robotu uygulamasi (2), kullanicilarin uygulamaya herhangi bir elektronik cihaz üzerinden erisebilmelerini saglamak üzere yapilandirilmaktadir. Sohbet robotu uygulamasi (2), kullanicinin bilgileri görüntülemesini ve alarm sorgulama islemine baslanildigi bilgisini alabilmesini saglamak üzere yapilandirilmaktadir. Sohbet robotu uygulamasi (2), dügüm üzerindeki cihazda aktif bir ariza olup olmadigi sorgusunun sunucu (4) üzerinde yapilmasinin ardindan, aktif bir alarm olmadigi durumda kayitli bir alarm olmadigi bilgilendirmesinin kullaniciya iletilmesini ve aktif bir alarm oldugu durumda akistan dönen ariza detay bilgisi ve tahmini sonlanma süresi bilgilerinin kullaniciya bir arayüz üzerinden iletilmesini saglamak üzere yapilandirilmaktadir. Bulus konusu sistemde (1) yer alan veri tabani (3), herhangi bir iletisim herhangi bir iletisim protokolü kullanarak sunucu (4) ile iletisim kurmak üzere yapilandirilmaktadir. Veri tabani (3), herhangi bir iletisim protokolü kullanarak sunucu (4) ile veri alisverisi gerçeklestirmek üzere yapilandirilmaktadir. Veri tabani (3), internet saglayici firma müsterileri verilerini kayit altinda tutmak üzere yapilandirilmaktadir. Veri tabani (3), müsteri modeminin bagli oldugu ilk dügüm bilgisi (node info) verilerini müsteri numarasi bazli olarak kayit altinda tutmak üzere yapilandirilmaktadir. Veri tabani (3), internet saglayici firma müsterilerinin bagli oldugu dügüm üzerindeki cihazin yolladigi son aktif alarm bilgisi verilerini kayit altinda tutmak üzere yapilandirilmaktadir. Veri tabani (3), ekiplerin kesinti veya ariza için verdikleri tahmini sonlanma süresi verilerini kayit altinda tutmak üzere yapilandirilmaktadir. Bulus konusu sistemde (1) yer alan sunucu (4), herhangi bir iletisim herhangi bir iletisim protokolü kullanarak sohbet robotu (2) ve veri tabani (3) ile iletisim kurmak üzere yapilandirilmaktadir. Sunucu (4), herhangi bir iletisim protokolü kullanarak sohbet robotu (2) ve veri tabani (3) ile veri alisverisi gerçeklestirmek üzere yapilandirilmaktadir. Sunucu (4), alarm bilgilerinin kontrol edilerek alarmin planli bir kesintiye veya bilinen bir arizaya mi ait oldugunu sorgulamak üzere yapilandirilmaktadir. Sunucu (4), yapilan sorgulama sonucunda söz konusu alarm planli kesinti veya bilinen ariza ile olusmus ise veri tabanindan (3) ekiplerin kesinti veya ariza için verdikleri tahmini sonlanma süresini çekmek ve kullaniciya cevap olarak sürenin dönülmesini saglamak üzere yapilandirilmaktadir. Sunucu (4), yapilan sorgulama sonucunda söz konusu alarm planli kesinti veya bilinen ariza ile olusmamis ise müsterinin bagli oldugu dügüm üzerindeki cihazda son bir yil içinde olusan alarmlarin listesini veri tabanindan çekmek ve alarm sonlanma süresini hesaplamak ve kullaniciya cevap olarak hesaplanan sürenin dönülmesini saglamak üzere yapilandirilmaktadir. Sunucu (4), sorgulama sonucunda planli kesintinin ve/veya bilinen arizanin olmadigi tespit edildiginde, veri tabanindan (3) ilgili abonenin bagli oldugu dügüm üzerindeki modemde son bir yil içindeki olusan alarmlarin listesine erismek ve söz konusu son bir yil içindeki olusan alarmlarin verilerini veri-zaman bazli modellemek üzere yapilandirilmaktadir. Sunucu (4), veri zaman bazli alarm modelle verilerinde normalizasyon yaparak tercihen 1 saatlik periyotlar olusturmak üzere yapilandirilmaktadir. Sunucu (4), veri bulunmayan 1 saatlik periyotlardaki degerlerin tespit edilmesinde enterpolasyon yapmak üzere yapilandirilmaktadir. Sunucu (4), saatlik olarak kaydedilen modem alarm degerleri üzerinde iteratif sekilde BNMF algoritmasi çalistirarak veri zaman matrisinin 2 ayri matrise ayrilmasini saglamak ve veri desenini olusturmak, t zaman noktasina kadar alinan veri setinde t+l zaman noktasini 0 olacak sekilde genisletmek, bu sekilde baslangiç matrisinden elde edilen 2 ayri matrisin çarpimiyla t+l noktasina denk gelen satirin ilgili abonenin modeminin tahmini alarm sonlanma süresi degerini içermesini saglamak üzere yapilandirilmaktadir. Sunucu (4), BNMF algoritmasiyla üretilen tahmini alarm sonlanma süresinin bilgilendirme uygulamasi üzerinden sohbet robotu (2) araciligi ile kullaniciya bildirilmesini saglamak üzere yapilandirilmaktadir. Sanayiye Uygulanabilirlik Bulus konusu sistem (1) sayesinde internet baglanti cihazlarinin tarihsel alarm sonlanma verilerinin Bayesian Non-Negative Matrix Factorization (BNMF) algoritmalariyla analiz edilmesiyle güncel alarmlarin sonlanma sürelerinin tahminlenmesi, internet baglanti cihazini kullanan abonenin bir arayüz üzerinden baglantisinda problem oldugunu girmesiyle sonucunda yapilan islemlerle ariza alarmin sonlanma tahmin süresinin kullaniciya ilgili arayüz üzerinden gösterilmesi saglanmaktadir. Bu temel kavramlar etrafinda, bulus konusu "Ariza Sonlanma Süresini Bayesian Non- Negative Matrix Factorization (BNMF) ile Tahminleyen ve Sohbet Robotu Üzerinden Kullaniciya Ileten Bir Sistem (1)" ile ilgili çok çesitli uygulamalarin gelistirilmesi mümkün olup, bulus burada açiklanan örneklerle sinirlandirilamaz, esas olarak istemlerde belirtildigi gibidir. TR DESCRIPTION A SYSTEM THAT ESTIMATES THE FAILURE TERMINATION TIME WITH BAYESIAN NON-NEGATIVE MATRIX FACTORIZATION (BNMF) AND TRANSMITS IT TO THE USER THROUGH THE CHAT ROBOT Technical Field This invention aims to determine the termination of current alarms by analyzing the historical alarm termination data of internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms. It is related to a system that enables estimating the durations, and showing the estimated termination time of the fault alarm to the user via the relevant chatbot, as a result of the actions taken as a result of the subscriber using the internet connection device entering a chatbot interface that there is a problem in connecting to the chatbot. Prior Art Nowadays, internet provider companies want to keep their connection devices always active and working in order to provide quality internet service to their customers. Devices owned by internet provider companies may malfunction in various situations. These problems must be resolved as soon as possible and an uninterrupted service must be provided. For this reason, by analyzing the historical alarm termination data of internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms, the termination times of current alarms are estimated, and the termination of the malfunction alarm is predicted by the transactions made as a result of the subscriber using the internet connection device entering a chatbot interface that there is a problem in connecting. It is understood that a system is needed that allows the duration to be displayed to the user via the relevant chat robot. The document describes systems and methods for managing user experience during cellular communications outages in a wireless communications network, such as a wireless 5G network. Exemplary embodiments include systems and methods for managing the user experience during cellular communications network outages, including network communications fault testing, performing different recovery actions in response to the detection of errors in communications between certain telecommunications network components and certain network links faults, involving the radio resource manager in the distributed unit. managing user control during cellular telecommunications network outage, managing user experience during cellular telecommunications network failures/outages, which includes a smart layer module that makes predictions about future communication failures in communication with the distributed unit (DU), managing user experience in case of cellular telecommunications network outages by utilizing the backup data center It encapsulates a group of microservices corresponding to central units (CUs) that will be resized to allow for newly added central units to manage the user experience during cellular telecommunications network outage. Brief Description of the Invention The purpose of this invention is to estimate the termination times of current alarms by analyzing historical alarm termination data of internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms, as a result of which the subscriber using the internet connection device enters a chatbot interface that there is a problem in connecting to it. It is to realize a system developed to ensure that the estimated termination time of the fault alarm is displayed to the user via the relevant chat robot. Detailed Description of the Invention "A System that Estimates the Failure End Time with Bayesian Non-Negative Matrix Factorization (BNMF) and Communicates it to the User via the Chat Robot", developed to achieve the purpose of this invention, is shown in the figure below; Figure 1; The subject of the invention is the schematic view of the system. The parts in the figure are numbered one by one, and the equivalents of these numbers are given below. 1. System 2. Chatbot Application 3. Database 4. Server Analyzing the historical alarm termination data of Internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms, estimating the termination times of current alarms and enabling the subscriber using the Internet connection device to enter a chat. The system (1) that is the subject of the invention was developed to ensure that the estimated termination time of the malfunction alarm is displayed to the user via the relevant chatbot, with the actions taken as a result of entering that there is a problem in connecting to the chatbot interface; -at least one chatbot application (2) configured to enable internet service users to access fault notifications, -at least one database (3) configured to record user data by connecting with the company providing the internet access service, -by checking alarm information To query whether the alarm is a planned outage or a known malfunction, and when it is determined that there is a planned outage and/or known malfunction thanks to the query, the chatbot application ( It contains at least one server (4) configured to ensure that the user is informed via 2). The chatbot application (2) included in the system (1) of the invention is configured to communicate with the server (4) using any communication protocol. The chatbot application (2) is configured to exchange data with the server (4) using any communication protocol. The chatbot application (2) is configured to enable the user to request fault termination information by performing login operations through an interface. The chatbot application (2) is configured to enable internet service users to receive fault notifications on an interface. The chatbot application (2) is configured to enable users to access the application via any electronic device. The chatbot application (2) is configured to enable the user to view the information and receive information that the alarm inquiry process has started. After the query is made on the server (4) whether there is an active malfunction in the device on the node, the chat robot application (2) informs the user that there is no recorded alarm in case there is no active alarm, and in case of an active alarm, the malfunction detail information and estimated termination returned from the flow. It is configured to ensure that time information is transmitted to the user via an interface. The database (3) in the system (1) subject to the invention is configured to communicate with the server (4) using any communication protocol. The database (3) is configured to exchange data with the server (4) using any communication protocol. The database (3) is configured to keep the internet provider company's customer data under record. The database (3) is configured to record the first node information (node info) data to which the customer modem is connected, based on customer number. The database (3) is configured to record the last active alarm information data sent by the device on the node to which the customers of the internet provider company are connected. The database (3) is configured to record the estimated termination time data given by the teams for outages or malfunctions. The server (4) in the system (1) subject to the invention is configured to communicate with the chat robot (2) and database (3) using any communication protocol. The server (4) is configured to exchange data with the chatbot (2) and database (3) using any communication protocol. The server (4) is configured to check the alarm information and query whether the alarm belongs to a planned outage or a known malfunction. The server (4) is configured to retrieve the estimated termination time given by the teams for the outage or malfunction from the database (3) if, as a result of the query, the alarm in question is caused by a planned outage or known malfunction and to return the time as a response to the user. The server (4) is configured to retrieve from the database the list of alarms that have occurred in the last year on the device on the node to which the customer is connected, if the alarm in question has not occurred due to a planned outage or a known malfunction as a result of the query, and to calculate the alarm termination time and return the calculated time as a response to the user. . When it is determined that there is no planned outage and/or known malfunction as a result of the query, the server (4) accesses the list of alarms that occurred in the last year on the modem on the node to which the relevant subscriber is connected from the database (3) and provides the data of the alarms that occurred in the last year in question. It is configured for time-based modeling. The server (4) is configured to create periods of preferably 1 hour by normalizing the data with a time-based alarm model. The server (4) is configured to perform interpolation in determining the values in 1-hour periods for which no data is available. The server (4) runs the BNMF algorithm iteratively on the hourly recorded modem alarm values to ensure that the data time matrix is divided into 2 separate matrices and to create the data pattern, to expand the t + l time point to be 0 in the data set received until the t time point, thus starting the It is configured to ensure that the line corresponding to the t+l point by multiplying 2 separate matrices obtained from the matrix contains the estimated alarm termination time value of the relevant subscriber's modem. The server (4) is configured to ensure that the estimated alarm termination time generated by the BNMF algorithm is notified to the user via the chat robot (2) via the notification application. Industrial Applicability Thanks to the system (1) of the invention, the termination times of current alarms are estimated by analyzing the historical alarm termination data of the internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms, and the malfunction is detected by the actions taken as a result of the subscriber using the internet connection device entering that there is a problem in the connection via an interface. The estimated termination time of the alarm is displayed to the user via the relevant interface. Around these basic concepts, it is possible to develop a wide variety of applications related to the subject of the invention, "A System (1) that Estimates the Failure Termination Time with Bayesian Non-Negative Matrix Factorization (BNMF) and Communicates it to the User via the Chat Robot", and the invention cannot be limited to the examples explained here. as stated in the claims. TR

Claims

1.STEMS. By analyzing the historical alarm termination data of Internet connection devices with Bayesian Non-Negative Matrix Factorization (BNMF) algorithms, the termination times of current alarms are estimated, and the estimated termination time of the fault alarm is displayed to the user via the relevant interface, as a result of the operations performed when the subscriber using the Internet connection device enters that there is a problem in his connection through an interface. providing; -containing at least one chatbot application (2) configured to enable internet service users to access fault notifications, and -at least one database (3) configured to keep user data under record by connecting with the company providing the internet access service, -controlling alarm information to query whether the alarm is a planned outage or a known malfunction, and when it is determined that there is a planned outage and/or known malfunction thanks to the query, the chatbot application can access the estimated termination time information reported by the teams on the database (2) for the outage and/or malfunction. A . A system (1) as in claim 1, characterized by a chatbot application (2) configured to communicate with the server (4) using any communication protocol. . A system (1) as in Request 1 or 2, characterized by a chatbot application (2) configured to exchange data with the server (4) using any communication protocol. . Any of the above claims, characterized by a chatbot application (2) configured to enable the user to request fault termination information by performing login operations through an interface. A system (1) as in any of the above claims, characterized by a chatbot application (2) configured to enable Internet service users to receive fault notifications on an interface. . A system (1) as in any of the above claims, characterized by a chatbot application (2) configured to enable users to access the application through any electronic device. . An . After the query is made on the server (4) whether there is an active malfunction in the device on the node, in case there is no active alarm, the user is informed that there is no recorded alarm, and in case there is an active alarm, the malfunction detail information and estimated termination time information returned from the stream is transmitted to the user via an interface. A system (1) as in any of the above claims, characterized by a chatbot application (2) configured to provide . A system (1) as in any of the above claims, characterized by the database (3) configured to communicate with the server (4) using any communication protocol. A system (1) as in any of the above claims, characterized by a database (3) configured to exchange data with the server (4) using any communication protocol. A system (1) as in any of the above claims, characterized by the database (3) configured to record the data of the Internet provider company's customers. A system (1) as in any of the above claims, characterized by a database (3) configured to record the first node information (node info) data to which the customer modem is connected, based on customer number. A system (1) as in any of the above claims, characterized by a database (3) configured to record the last active alarm information data sent by the device on the node to which the customers of the Internet provider company are connected. A system (1) as in any of the above claims, characterized by a database (3) configured to record the estimated termination time data given by the teams for outages or malfunctions. A system (1) as in any of the above claims, characterized by the server (4) configured to communicate with the chatbot (2) and the database (3) using any communication protocol. A system (1) as in any of the above claims, characterized by the server (4) configured to exchange data with the chatbot (2) and the database (3) using any communication protocol. A system (1) as in any of the above claims, characterized by the server (4) configured to check the alarm information and query whether the alarm belongs to a planned interruption or a known malfunction. As a result of the query, if the alarm in question is caused by a planned outage or known malfunction, as in any of the above claims, characterized by the server (4) configured to retrieve the estimated termination time given by the teams for the outage or malfunction from the database (3) and to return the time as a response to the user. a system (1). As a result of the query, if the alarm in question is not caused by a planned outage or a known malfunction, the server (4) is configured to retrieve the list of alarms that have occurred in the last year on the device on the node to which the customer is connected from the database, calculate the alarm termination time and return the calculated time as a response to the user. A system characterized in any of the above claims (1). When it is determined that there is no planned outage and/or known malfunction as a result of the query, it is configured to access the list of alarms that occurred in the last year on the modem on the node to which the relevant subscriber is connected from the database (3) and to model the data of the alarms that occurred in the last year in question on a data-time basis. A system (1) as in any of the above claims, characterized by the server (4). A system (1) as in any of the above claims, characterized by the server (4) configured to create periods of 1 hour, preferably by normalizing the data with a time-based alarm model. A system (1) as in any of the above claims, characterized by the server (4) configured to perform interpolation in determining the values in 1-hour periods for which no data is available. By running the BNMF algorithm iteratively on the hourly recorded modem alarm values, ensuring that the data time matrix is divided into 2 separate matrices and creating the data pattern, expanding the t+l time point to be 0 in the data set taken up to time point t, thus creating 2 separate data obtained from the initial matrix. A system (1) as in any of the above claims, characterized by the server (4) configured to ensure that the line corresponding to the t + l point by multiplying the matrix contains the estimated alarm termination time value of the relevant subscriber's modem. A system (1) as in any of the above claims, characterized by the server (4) configured to ensure that the estimated alarm termination time produced by the BNMF algorithm is notified to the user via the chat robot (2) via the notification application. TR