WO2023033684A1

WO2023033684A1 - Method for the mobile messaging of mobile subscribers

Info

Publication number: WO2023033684A1
Application number: PCT/RU2022/050276
Authority: WO
Inventors: Евгений НЕКРАСОВ
Original assignee: Акционерное Общество "Квантум А Рус"
Priority date: 2021-09-04
Filing date: 2022-09-02
Publication date: 2023-03-09
Also published as: RU2021126096A

Abstract

A method for the mobile messaging of mobile subscribers involves user software for determining an SMS template and type, and server software for classifying SMS traffic. The server software for classifying SMS traffic operates in two modes: a primary (templating) mode used for SMS tariffing, and a secondary (heuristic) mode used for the business needs of the mobile network operator. An SMS template comprises up to six variables.

Description

A method of mobile informing cellular subscribers.

The present invention relates to the field of telecommunications, in particular to a2p services of automatic CMC (SMS/MMS/USSD/PUSH) mailing lists of cellular operators intended for mobile informing subscribers. The invention can be used in cellular communication networks of GSM, IN, IMS standards.

In the context of the application, the SMS mailing service, the a2p service and mobile informing are identical.

SMS mailing services are known and widely used for mobile informing cellular subscribers about events generated by their behavior. For example, an order in an online store, payment with a bank card, a one-time password for entering a social network, etc. Bulk SMS mailings are carried out for advertising purposes.

SMS mailing services are made according to the web service model, which provides for the payment by the user of the service to the service provider of a subscription to the web service and / or post-payment for the SMS traffic generated by him. Users of SMS mailing services are heterogeneous business entities (online stores, banks, social networks, etc.). SMS service providers are mobile operators (operator). Tariffication of SMS mailing for the user is different and depends on the type of SMS. Cellular operators charge different SMS prices depending on its type - transactional (account balance notification), service (one-time password), authorization (authorization code on the site), informational (order delivery message), advertising (commercial offer), and etc. The syntax rules of SMS messages according to belonging to a certain type are determined by an agreement between the user and the operator.

For the operator, the economic feasibility of different tariffing by type of SMS is the ability to flexibly apply marketing tools (price, discounts, service packages, etc.). They help attract diverse users - from small online stores to large banks. For the user, the economic advantage of different tariffs by SMS type is the ability to choose a more economical type of SMS distribution for their business tasks.

Different tariffication of SMS types encourages users, in order to reduce their costs, to mask more expensive advertising mailings within less expensive ones. For example, an agreement between the operator and the user of the service determines that the SMS structure “Dear XXXX, you have received XXX XXX XXX XXX credited to your account” corresponds to the transaction type at the rate of 1 rub. for SMS. And the user generates CMC mailing "Dear customer, 100 bonuses for purchases in our store have been credited to your account." The SMS syntax formally corresponds to the transactional type, but semantically obviously refers to the advertising type at the rate of 2 rubles. for SMS. The classification server will assign a transactional type to this SMS, and the billing will charge the SMS as transactional. Subscribers will receive an advertising mailing, and the operator will receive less than 1 rub. for SMS due to the substitution of the type of SMS. The operator, in order to prevent the loss of part of the income due to the substitution of the type of SMS traffic, is forced to use complex and expensive technical means (semantic analysis and neuro classification) on the classification server to correctly recognize the type of SMS traffic online.

The proposed method for mobile informing cellular subscribers is aimed at solving the problem of reliably determining and recognizing the template and type of SMS in order to properly charge SMS traffic without losing classification accuracy. The method also provides an increase in the performance of technical means of the SMS mailing service and an expansion of the arsenal of tasks to be solved by the SMS mailing service.

The applicant's patent 2697648 "Traffic classification system" is known, which is taken as a prototype, according to which SMS is received through the network interface, its scheme is determined and divided into semantic fields. On the server, the administrator defines an informal template and a probability factor for assigning SMS to a certain category. Next, the informal pattern is converted into a regular expression regex. Apply the regular expression regex to the meaningful fields of the message and submit the message to the input of the neural network. And the result of the classification of the message by the neural network is corrected in accordance with a certain probability coefficient by the system administrator.

This solution has a number of disadvantages. Excessive complexity of the architecture as a whole due to the mandatory execution of time-consuming semantic analysis and CMC neuro classification online. The need for the system administrator to determine informal templates and coefficients of probability of referring them to a certain category. The nomenclature of business tasks of users is constantly increasing, the heterogeneity of the semantics of all kinds of SMS is increasing. As a result, the syntax of the informal regular expression language becomes more complicated. This, in turn, requires additional administrator training in regex issues, which violates the main requirement - system administration by users who are not trained in regex issues. The proposed method provides a reduction in SMS traffic processing delays, an increase in SMS traffic processing performance, a given classification accuracy and an expansion of the arsenal of tasks to be solved by technical means of an SMS traffic classification server due to: . Exceptions for assigning templates and the probability coefficient for a template to be assigned to a certain category by the service administrator. . Replacing the time-consuming semantic analysis of SMS traffic online with a preliminary check of the template, storing it in the memory of the SMS classification server and templating SMS traffic online. . Disabling the machine learning module in the main mode of checking and classifying SMS traffic. Performing additional verification and classification of SMS traffic (neuro classification) online for specific operator business scenarios. . Constant determination of the coefficient of probability of assigning a template to a certain category in the program code of a neural network.

The economic effect of the proposed method is to increase the reliability of SMS traffic classification and the possibility of performing additional operator business scenarios. The technical effect of the claimed method is to increase the automation of the administration of the classification system, simplify the technical appearance of the classification system and increase the performance of the classification server. The expansion of the arsenal of tasks to be solved consists in the allocation of two modes of operation of the classification server - the main one for classifying SMS and an additional one for special business scenarios without their mutual influence on the performance and results of each separately.

The technical objective of the invention is to improve the performance of the technical means of the SMS traffic classification system, improve the accuracy of the SMS traffic classification, as well as simplify the administration and architecture of the SMS traffic classification system.

The technical problem is solved due to the method of mobile informing cellular subscribers, according to which the cellular operator provides the user of the SMS distribution service with software that interacts with the network node that performs the classification of SMS traffic and is made with the ability to determine the template and type of SMS, by the technical means of the network node that performs the classification SMS traffic performs preliminary reception, CMC template checking and storing, online performs basic checking and classification of SMS by type using a pre-stored CMC template, and the SMS template includes up to six auto-substitution variables.

The technical objective of the invention is to expand the arsenal of tasks to be solved by technical means of the SMS traffic classification system.

The technical problem is solved due to the method of mobile informing cellular subscribers, according to which the cellular operator provides the user of the SMS distribution service with software that interacts with the network node that performs the classification of SMS traffic and is made with the ability to determine the template and type of SMS, by the technical means of the network node that performs classification of SMS traffic is performed by preliminary reception, verification and storage of the SMS template, in online mode it performs the main verification and classification of SMS by type using the SMS template previously stored in the memory and additional verification and classification of SMS by type by a neural network, the result of which is corrected in accordance with a constant probability coefficient for assigning a message to a particular type, and the SMS template includes up to six auto-substitution variables.

The software that interacts with the network node that performs the classification of SMS traffic can be implemented as a web, mobile or console application. The type of SMS is transactional, service, authorization, informational, advertising, as well as any other type of SMS in accordance with the business tasks of the user of the SMS distribution service and the mobile operator. And the six auto-substitution variables are denoted by symbols and take values:

%w - any continuous set of letters and/or special characters;

%w+ - sequence of words separated by spaces;

%w{ 1, n} - limited sequence of words from 1 to n separated by spaces;

%d - any continuous set of numbers and/or special characters;

%d+ - sequence of numbers separated by spaces;

%d{ 1, n} - limited sequence of numbers from 1 to n separated by spaces.

Figure 1 shows the generalized architecture of the a2p service.

Figure 2 shows a generalized functional diagram of the a2p service. Application - user software of the SMS mailing service, as a rule, is made in the format of the web interface of the user's personal account (LC), which is available to the user through a standard web browser (browser) (figure 1). Application can be executed as a console (no GUI) or mobile application. The main functionality of the Personal Account is the user's definition of templates and types of SMS distribution, in which these templates will be applied in accordance with the agreement between the user of the service and the operator. The graphical (or console) application interface (Fig. 2) contains a list of template types: T-transactional, C-service, A-authorization, I-information, etc. As well as a list of user templates: “BANK; T; Card operation% d completed”, “BANK;С;One-time login password %w”, etc. By agreement with the operator, the user describes the template using a list of six auto-substitution variables and their possible values:

1. %w - any continuous set of letters and/or special characters;

2. %w+ - sequence of words separated by spaces;

3. %w{ l, n} - limited sequence of words from 1 to n separated by spaces;

4. %d - any continuous set of numbers and/or special characters;

5. %d+ - sequence of numbers separated by spaces;

6. %d{ l, n} - limited sequence of numbers from 1 to n separated by spaces.

A list of 6 auto-substitution variables has been analytically defined by the Applicant. The list is necessary for reasons of ease of use by untrained users and sufficient to describe their any meaningful patterns. The simplicity, necessity and sufficiency of the six notations for auto-substitution variables is due to:

• the inclusion of only two alphabetic characters "d" (digit) and "w" (word) with an intuitive definition of substitution - a sequence of "d" numbers and "w" letters;

• addition of two alphabetic characters with only two expansion operators "+" and "{ 1, n}" with an intuitive extension definition - an unlimited "+" and a limited "{ 1, n}"sequence; covering any combination of natural language symbol substitution to build a heterogeneous SMS text in accordance with a large range of business tasks.

The Application interacts with the a2p server via the HTTP protocol to transfer user templates for pre-checking and writing to the a2p server's memory (figure 2). In special cases, for small users who do not have software that performs ESME functions, the application can send HTTP GET, POST requests to the a2p server with data on the text of senders, recipients and SMS text to perform SMS mailings directly by the a2p server. In this case, the a2p server initiates and participates in SMS distribution as an ESME sender (ESME in transmitter mode). This proprietary implementation (without ESME on the user's side) can be used for small users who are costly to purchase the ESME software client.

ESME is a software client in the architecture and definitions of the SMPP protocol that acts as an element of the external short message system (External Short Message Entity, SMPP protocol specification). In fact, it is a software client for initiating connections with the SMSC server (SMS center) via the SMPP protocol for sending SMS (figure 1). On the user side, it can interact with the application via consistent APIs (FIG. 2).

CRM - user software that performs the functions of customer relationship management (figure 2). In the general case, this is a standard system for processing user requests from clients, containing a database of MSISDN clients and the history of their requests.

A2p - SMS mailing server or mobile informing platform - a hardware and software system that classifies SMS traffic. The hardware architecture of the a2p server is implemented on the X86 architecture running a Unix-like operating system.

The software architecture of the a2p server generally performs (figure 2):

• in the mode of pre-checking and writing templates - receiving templates via HTTP protocol, their preliminary checking, approval and saving to memory; • in the main mode of operation performs the basic check and classification of SMS traffic for the correspondence of meaningful SMS text data to the mask of a previously saved SMS template;

• in an additional mode of operation performs the classification of meaningful text data of SMS messages by a neural network,

• performs initiation of connections with SMSC via SMPP protocol to perform SMS distribution via HTTP application request (ESME in transmitter mode).

A2p server performs ESME functions, interacts with the operator's billing network node for subsequent SMS billing, uses standard software tools for distribution and load balancing to process a large amount of SMS traffic. The a2p server software architecture provides software interfaces and algorithms for processing USSD, SMTP, PUSH (HTTP) traffic. Processing of USSD, SMTP, PUSH (HTTP) traffic is carried out similarly to processing of SMS traffic, and also includes preliminary checking and recording of templates, basic and additional checking and classification of USSD, SMTP, PUSH (HTTP) messages.

MS (Mobile Station) or person - a standard mobile phone for a cellular network subscriber.

Pre-checking and writing the template to the memory of the a2p server is performed by the program logic of the a2p server after the user defines a new template (figure 2). The program logic of the client (application) sends newly defined user-defined templates (BANK;T; “Operation on card %d completed”) and (BANK;C; “One-time password for entering %w”) to the address of the a2p server via HTTP protocol. Template syntax through ";" defines: sender (BANK), template type (T-transactional, C-service), text (Card transaction, One-time login password) and a metacharacter (%d; %w) that takes values in accordance with the definition of autosubstitution variables. A preliminary check may include manual moderation of the template and its type and/or programmatic checking of the syntax and semantics of the template for belonging to the declared type and the correctness of the content of the semantic fields by the technical means of the a2p server. After the a2p check, the server makes a new entry about the new template and its type in the list of user templates and writes the template and its type into memory. The main mode of checking and classifying SMS traffic (figure 2) is performed by stenciling SMS arriving at the network interface and provides for:

• receiving CMC (pdu submit_sm SMPP) on the negotiated network interface,

• extracting meaningful data from the short_message field of the submit_sm message,

• division of meaningful data of the short message field into semantic fields field,

• applying masks of previously saved user templates (source_addr) to the semantic fields field,

• definition of a templated SMS in case of matching one of the user's templates (source_addr) and classification of the SMS according to the type of this template,

• definition of SMS untemplated in case of non-compliance with any user template (source_addr) and classification of SMS as advertising,

• writing to a log file (.log) of CMC classification data (pdu submit_sm),

• sending SMS to the network to the address of the recipient (field destination_addr, pdu submit_sm),

• transfer of a registration file (.log) to the address of the billing network node for SMS billing.

If the SMS does not match any of the user's templates, the a2p program logic defines the SMS as not templated. For non-template SMS, the program logic assigns an advertising type, which is recorded in a .log file. Further, the billing of the operator charges this SMS as advertising. This logic was chosen due to the fact that the semantics of advertising messages is quite complex and requires a significant increase in the list of auto-substitution variables. This in turn would violate the requirement that auto-substitution variables be easy to use by untrained users.

An additional mode of checking and classifying SMS traffic performs neuro classification of SMS (figure 2). The additional mode is used to execute special business scenarios of the operator, within which SMS types are allowed and used for recognition, which may not be defined and not used in the main traffic inspection and classification mode. For example, the neural network may recognize an H-obscene or M-fraudulent type of SMS, or recognize covert advertising semantics in a templated SMS (FIG. 2).

One of the special business scenarios of the operator, which performs the mode of additional verification and classification of traffic, is the implementation of the adBlock ad blocking service via the SMS switching channel with the subscriber. In this scenario, the operator can be based on the result determined by the template logic in the main mode of the a2p server. For SMS billing, such logic (recognition of an untemplated advertising SMS) may be acceptable under the terms of the agreement between the operator and the user. To fulfill the requirements of a subscriber's subscription to adBlock, the operator must make sure that there are no SMS advertising semantics, since in the general case, non-compliance with any user template is not equivalent to the presence of SMS advertising semantics. A subscriber's subscription to ad blocking should not only block promotional SMS, but also ensure the delivery of useful non-promotional SMS, the syntactic structure of which, in general, may not correspond to any user template.

The operator can provide other business scenarios in the mode of additional verification and traffic classification - for example, recognition of M-fraudulent, N-obscene SMS, etc. At the same time, as shown in figure 2, in the execution of special business scenarios in the mode of additional verification and traffic classification, the operator is not bound by the types of SMS traffic defined in the mode of basic verification and traffic classification.

Another scenario for applying the mode of additional verification and traffic classification is rechecking the templated SMS in case of suspicion of traffic substitution within the approved template.

An additional mode of checking and classifying SMS traffic provides for the supply of semantic fields (field..) to the input of the trained neural network (figure 2). The result of the neuroclassification is taken from the output neurons, which assign to their traffic type the value of the probability PXsh with which the neural network classifies SMS by this type of traffic.

The program logic of the server a2p may provide for a constant, i.e., a coefficient const Pf predetermined by the program algorithm, of referring the SMS to a certain one or more template types (figure 2). The coefficient const Pf divides the values of the probability of attributing an SMS message to a certain type of RHsh, which are formed by the neural network (output neurons) into appropriate and improper. The function F(PX1, const Pf) is quite simple - if the value of the probability PXsh formed by the output neuron Xsh is less than the value of the coefficient PXm<const Pf, then this type is defined as improper and is not used. SMS is assigned a type whose probability value is greater than the value of the coefficient PXm> const Pf, for example, P advertising (figure 2). In general, the logic of applying the coefficient const Pf is identical to the logic of the traffic classification system according to the patent of the applicant 2697648, except for the constant value const Pf predetermined by the algorithm of the neural network software module and excluding its appointment by the system administrator.

As shown above, the claimed technical result of the method provides due to:

- Pre-checking the template, its type, saving the SMS template on the a2p server and applying the mask of the saved template to the online SMS traffic, and the definition and preliminary check of the template includes the use of no more than six auto-substitution variables in the SMS template, which consist of only two letters characters and expansion operators and accept an intuitive set of values.

- Dividing the mode of online classification of SMS traffic into two modes of operation - the mode of the main check and traffic classification and the mode of additional check and traffic classification, and the result of the main check and traffic classification is used for billing, and the additional check and classification of traffic is used to solve problems in special business operator scripts.

The use of these organizational and technical measures simplifies the technical appearance of the SMS traffic classification system as a whole, provides the specified classification accuracy, simplifies the architecture and administration of the system, and ensures the performance of additional tasks.

Replacing resource-intensive semantic analysis with templating significantly increases the performance of the a2p server without loss of classification accuracy and allows you to free up the power of neuro classification to expand the arsenal of technical means - solving special business problems for the operator.

As can be seen from the description above, neural network classification is used only in the additional mode of operation of the a2p server, which is designed to perform special operator business scenarios on non-template SMS traffic or in some scenarios can be used to recheck templated SMS traffic. Nevertheless, in both cases, the use of neuro classification is performed outside the main mode of operation of the a2p server for its intended purpose - SMS classification for subsequent billing. This increases the performance of the a2p server and expands the arsenal of tasks it solves. The constant determination of the coefficient const Pf directly in the program code of the neural network module (predetermination of const Pf) for reasons of low variability in the range of types of SMS mailings does not reduce the flexibility of the system as a whole.

The proposed system and method for mobile informing cellular subscribers has been tested by the applicant. According to the test results, an increase in the speed of processing SMS traffic without reducing the accuracy of classification was confirmed.

Claims

Formula Method for mobile informing cellular subscribers, in accordance with which the cellular operator provides the user of the SMS distribution service with software that interacts with the network node that performs the classification of SMS traffic and is configured to determine the template and type of SMS, characterized in that the cellular operator by the technical means of the network node classifying SMS traffic performs preliminary reception, verification and storage of the SMS template, online performs basic verification and classification of SMS by type using a pre-recorded SMS template, and the SMS template includes up to six auto-substitution variables. A method for mobile informing cellular subscribers, in accordance with which the cellular operator provides the user of the SMS distribution service with software that interacts with the network node that performs the classification of SMS traffic and is configured to determine the template and type of SMS, characterized in that, the cellular operator by technical means of the network The node performing the classification of SMS traffic performs preliminary reception, verification and storage of the SMS template, online performs the main verification and classification of SMS by type using the SMS template previously stored in the memory and additional verification and classification of SMS by type by a neural network, the result of which corrects in in accordance with a constant probability coefficient for classifying a message as a particular type, and the SMS template includes up to six auto-substitution variables. The method of mobile informing cellular subscribers according to i.1,2, characterized in that the software interacting with the network node that performs the classification of SMS traffic and is configured to determine the template and type of SMS is made in the form of a web, mobile or console application. A method for mobile informing cellular subscribers according to i.1, 2, characterized in that the type of SMS is transactional, service, authorization, informational, advertising, as well as any other type of SMS in accordance with the business tasks of the user of the SMS distribution service and the mobile operator. The method of mobile informing cellular subscribers according to claim 1, 2, characterized in that six auto-substitution variables are indicated by symbols and take on the values:

%w - any continuous set of letters and/or special characters;

%w+ - sequence of words separated by spaces;

%w{ 1, n} - limited sequence of words from 1 to n separated by spaces; %d - any continuous set of numbers and/or special characters;

%d+ - sequence of numbers separated by spaces;

%d{ 1, n} - limited sequence of numbers from 1 to n separated by spaces.