RU2726259C1 - System for providing expert advice in real time - Google Patents

Abstract

FIELD: data processing.SUBSTANCE: invention relates to information and analytical systems, which enable to provide users with help services in real time, and can be used to provide experts in real-time consulting and consulting services to users. System comprises a database of experts registered in the system, for each expert description of expert skill, where each description contains a tag cloud suitable for determining conformity of subjects of issues of an expert industry, and a rating index of an expert, quantitatively describing an expert's skill level relative to other experts registered in the system; database of registered users, containing identification data and bank requisites of users; means for obtaining user questions in natural language format; a question parsing tool which enables parsing each of the received questions for essential features suitable for analysis on compliance with an industry expert question defined by a tag cloud.EFFECT: technical result consists in improvement of quality and speed of user service, high probability of obtaining correct answers to questions.7 cl, 9 dwg, 5 tbl

Description

FIELD OF TECHNOLOGY

The invention relates to information and analytical systems that allow users to provide reference services in real time and can be used to provide users with reference and consulting services of experts - individuals 24 hours a day in real time using the Internet.

BACKGROUND

A prerequisite for creating a system is the need to create scalable help systems to solve problems or answer questions that often do not have a clear answer. To solve such issues, human participation is necessary. For example, artificial intelligence cannot assess the danger of the consequences of an incorrect answer to the user in the event that the correct answer is obviously impossible. Also, artificial intelligence does not reliably minimize the risks of an incorrect answer.

One of the analogues of the proposed invention is a typical technical support service, which provides for the call of the operator. In typical systems, the essence of the question is determined by the user himself, and the structure of the use of specialists, the payment of which is carried out regardless of the time spent on work and the quality of answers, does not provide users with quick and high-quality service. For example, in a typical technical support structure (Bosch AXT), a user addresses a question to a support service, the operator of the first support line registers a call, helps the user independently if possible, or transfers the question to the second support line. The second line of support receives requests from the first line, works on them, if necessary, attracting specialists from related departments to solve the problem (for example, system administrators, support for special software, support for special equipment, billing system administrators, etc.). Such an organization of a reference and information system does not meet the needs of many users, according to which it is expected to receive answers in a minimum time, for example, on a commercial basis.

To address the above disadvantages of the prior art, the technical solution is described in US patent US9992342, published 05.06. 2018. A well-known solution provides the fastest user access to the services of professional consultants. In this decision, there is no thorough analysis of user questions when choosing consultants, in some cases the choice of an expert (consultant) is carried out by the user. Such an approach to solving a technical problem does not ensure the high quality of the provision of consultant services, both in terms of the reliability of the advice provided, and in terms of timeliness and speed of service.

ESSENCE OF TECHNICAL SOLUTION

The technical result achieved by using the invention is to improve the quality and speed of user service, increasing the likelihood of receiving the right answers to the questions posed.

To achieve the technical result, a system is proposed for providing expert advice in real time, containing: a database of experts registered in the system, containing, for each expert, a description of the expert’s qualifications, where each description contains a tag cloud suitable for determining the conformity of the subject of industry issues expert and expert rating index, which quantitatively describes the expert’s qualifications relative to other experts registered in the system; a database of registered users containing identification data and bank details of users; a tool for receiving user questions in a natural language format; a tool for parsing questions, providing a parsing of each of the received questions for essential features suitable for analysis on the correspondence of a question of an industry of an expert defined by a tag cloud; means of selecting an expert whose industry is as close as possible to the subject matter of the issue; means for managing a data transmission channel from an expert to a user; a processing center, which ensures blocking of funds in the account of the user’s account before activating the data transfer channel and debiting the funds according to the result of receiving a response from the expert by the user approved by the user; means for setting a response timer configured to determine a time limit for an expert response not exceeding a predetermined time limit; means for determining the cost of the response, made determining the cost of the response in accordance with a predetermined time limit and rating index of the expert; feedback means made by the user receiving an assessment of the quality of the response; a means of refining expert rating indices made by qualifying expert rating indices taking into account assessments of the quality of answers and complexity of questions and a means of accounting active system experts made by forming a list of available experts from experts registered in the system; the system is designed to provide: receiving a question from a user registered in the system using identification data; user selection of expert rating index; parsing the user’s received question for significant features with sufficient detail to determine whether the industry’s industry question is determined by the tag cloud for essential features, selecting experts from among the available experts, tag clouds and rating indices that most closely match the selected rating and the essential features of the question; the selection of one expert whose tag cloud according to morphological characteristics matches the text of the question as much as possible; transfer of the conditions for providing a response in the form of information about the selected expert, time limit and cost of the response to the user; after obtaining consent from the user with the terms of the response, blocking the funds in the user’s account in the amount of the proposed cost of the response; transfer of the question to the selected expert; starting a timer set for a given response time when starting to provide the user with an expert response; request from the user to evaluate the expert’s response after a specified response time; write-off of funds from the user’s account in case of a positive assessment of the answer and taking into account the assessment of the answer in the expert rating index.

In the particular case, the system is designed to ensure that funds are unblocked on the user’s account and that the conditions for providing the answer are repeated if, after the time limit for providing the answer has expired, the completeness of the expert’s response to the user has not been ensured.

In a particular case, the system contains automated arbitration tools that provide formal comparisons of the expert’s question and answer, unblock the funds in the user’s account when the answer does not correspond to the user's question, and form an expert’s answer assessment, ensuring that the expert’s answer is taken into account in the expert rating index.

In a particular case, the expert tag cloud in the system is divided into segments, each of which has an expert rating index corresponding to a particular industry.

In the particular case, the system contains means for expanding the expert tag cloud.

In the particular case, the system provides for planning consultations for the appointed time.

In a particular case, the system contains automated arbitration tools that provide formal comparisons of the expert’s question and answer, unblock the funds in the user’s account when the answer does not correspond to the user's question, and form an expert’s answer assessment, ensuring that the expert’s answer is taken into account in the expert rating index.

BRIEF DESCRIPTION OF THE DRAWINGS

The description below is illustrated by the images in which:

FIG. 1 illustrates the general architecture of the system indicating the processes of interaction between elements;

FIG. 2 illustrates a functional diagram of the interaction of elements and blocks of the system;

FIG. 3 illustrates an example of the dynamic use of system functions during operation;

FIG. 4 illustrates an exemplary embodiment of an algorithm for generating a schedule of available slots in an expert profile;

FIG. 5 illustrates an exemplary embodiment of a personified pricing algorithm and a quantum calendar;

FIG. 6 illustrates an exemplary algorithm for adding topics to the expert profile;

FIG. 7 illustrates yet another embodiment of a personified pricing algorithm and a quantum calendar;

FIG. 8 illustrates an example embodiment of an algorithm for determining a relevant expert for a given request;

DESCRIPTION OF EMBODIMENTS OF THE INVENTION

In FIG. 1 shows the general architecture of the system with an indication of the interaction processes between the elements.

In the proposed system, the general architecture of which with an indication of the interaction processes between the elements is shown in FIG. 1, independently used external means of payment. As an example, in FIG. 1 are shown:

Payment acceptance service Yandex. Cash desk 105 associated with the bank (107) accepts payments from users and disburses funds at the request of the Expert.

CloudPayments 110 service associated with the bank (107), not only carries out electronic cash circulation, but also provides the distribution of checks related to electronic payments.

SMS-Gate 115 is an optional system unit used to send messages using the short message service (SMS) to subscribers of cellular networks and control the system using received messages. For example, after registering a new user in the system (on the platform) using the identification data of the mobile phone, the access password and notifications are sent to the user's mobile phone using SMS.

If the user has chosen the option of SMS reminders about the consultation, an SMS message is generated to SMS notify the user and / or expert to remind about the upcoming consultation, for example, 1 day before the start of the consultation and 15 minutes before the start of the consultation.

WebRTC 125 technology is used to transmit streaming video and audio data (120), which provides streaming audio, video and other content from the browser and to the browser in real time without installing plug-ins or other extensions. The technology is open source. With this technology, integration of the server and client part of the platform is required.

A set of interconnected programming interfaces and API protocols (130 from the English. Application programming interface - 130, are used in WebRTC 125 technology for data transmission and processing.

The system content management subsystem (CMS) 135, presented by the server part, is shown as an example of the commercial 1C-Bitrix product. The server used in the proposed invention consists of the following functional elements:

- main module;

- structure management;

- information blocks;

- Search;

- translation;

- social network;

- post office;

- calendars;

- universal lists;

- integration with CRM;

- Transactional blocks;

- system of accrual \ debit of the platform’s internal currency;

- currencies;

- web forms;

- web analytics;

- scaling;

- database;

- mobile platform.

The software interface (REST API) 140 shown in the image is used for software and hardware integration of the server and client parts of the platform.

The system stores information necessary to pay for the consultation, and the system collects and transfers the information necessary to pay for the consultation to client devices (user devices) connected to the system using REST API 140.

The system database is represented by the Database 145 block stored on a server managed using the MySQL 150 database management subsystem.

The optional Crash log system 155 subsystem is used to log errors to analyze them later and prevent recurrence.

The server for sending e-mail notifications is represented by the SMTP server (SMTP-Server) 160 block; the module for sending instant notifications and processing responses sent, for example, using client equipment, is represented by the Push and Pull 165 block.

The client part is represented by a Mobile Application IOS user device 170 using a system-compatible mobile application 180 for the IOS 175 operating system and a 185 Mobile Application user device device using a system-compatible mobile application 195 for the Android 190 operating system. The number of mobile or stationary users devices and the possibility the use of other operating systems is determined by the bandwidth of the communication channels of the central server and the data exchange formats supported by the server. In the particular case, an Internet browser supporting the secure data transfer protocol (https) can be used as a compatible application. Personal devices of users connected to the system can use separate communication channels for transmitting audio and video in the process of receiving services.

As shown in FIG. 1, the proposed system uses a central server and a server for processing audio and video data to provide services. The central server contains an application server (127), a database (145), payment processing tools (110, 105), as well as functional interfaces used to connect user devices to the system.

The system with the proposed architecture provides the user with the opportunity to receive the necessary and relevant information from the Expert, improves the quality of services by using the Expert rating system, provides flexibility by providing a choice of consultation format, and also eliminates geographical restrictions.

Regardless of the number of users and experts using the system, the system contains tools that enable users and experts to communicate with high quality work. The user can ask a question on the platform with text or audio input, the platform will select an Expert, providing one of the consultation formats selected by the user: text, audio, video chat.

In particular, the system contains a means for receiving user questions in a natural language format. Due to the fact that user questions are received and processed by automated means, the preferred means of receiving questions are input and reception of textual information, or means of converting live speech into a text format. Additionally, means of recognition and conversion of graphic information can be used, which ensure the receipt of handwritten text or images of documents, as part of user questions or for exchanging information with an expert.

Means of controlling the data transmission channel from the expert to the user ensure the interaction of the expert and the user in the selected communication format and can use the connection between the user and the expert using a central server. Using a server is a prerequisite for the operation of the system, since the organization of the work of arbitration means requires registration of all stages of the interaction of the expert and the user.

Textual, and in particular cases graphic information, is recognized or divided into elements by means of parsing questions, which provides analysis of each of the received questions into essential features suitable for analysis on the correspondence of a question of an expert industry defined by a tag cloud. The technology for parsing questions can be determined by the search technology using artificial intelligence, machine learning, the technology of semantic analysis, gradient boosting, etc.

Expert selection tools are automated search tools that use tools compatible with parsing technology that provide expert selection. When choosing an expert, the industry (set of competencies) of the expert is taken into account. Which should be as relevant as possible to the subject matter. Search tools can additionally use the possible rejections of users from the experts proposed by the system, as well as evaluations of the results of consultations put forth by users, and refusals of experts from the proposed questions. In particular, with the effective functioning of the search system, the search results are accepted by both users and experts, and with stable dissatisfaction of users with the answers of experts or with the refusal of experts from the proposed questions, the search technology can be refined. In the particular case, retraining of the system is used or the search technology is being replaced.

The processing center used in the system is designed to lock the funds in the account of the user account before activating the data transfer channel and write off the funds according to the user-approved result of receiving a response from the expert by the user.

Information about the account and account details of the Expert and the Customer (user) is stored in the database of the system, and all transactions on the accounts are automatically taken into account by the system.

Due to the fact that it is essential for the proposed system to set the limit for the response time of the expert, the system contains means for setting the response timer, which determines the time limit for the response of the expert, not exceeding a predetermined time limit.

The user can also select a time slot for consultation.

Means for determining the cost of the response, made determining the cost of the response in accordance with a predetermined time limit and rating index of the expert. In the particular case of the invention, the cost of the response can be determined by writing in the expert profile, and the task of the means of determining the value can be to extract the specified information from the expert profile.

A feedback tool is made that receives an assessment of the quality of the response from the user, for example, by suggesting alternatives to the user, one of which should correspond to the user's opinion regarding the quality of the response.

A means of refining expert rating indices, made by refining expert rating indices taking into account assessments of the quality of answers and complexity of questions as shown below.

A tool for accounting active system experts made by forming a list of available experts from experts registered in the system.

To ensure the operation of the system in the system, a database of experts registered in the system is created, containing, for each expert, an expert qualification description, where each description contains a tag cloud suitable for determining whether the topics of the industry are relevant to the expert and an expert rating index that quantitatively describes the expert qualification relative to other experts registered in the system.

In a system with a consistently low percentage of consultations, the results of which use arbitration tools, a recommendation mechanism may be used, according to which a recommendation expert or user with a certain rating can increase the starting rating of new experts. Experts with the status of “recommender” can determine the starting rating of new experts and specify the rating of other experts, for example, at the request of experts who want to increase the starting rating. In addition, recommendations affecting the rating index of an expert can be obtained both from experts without the status of “recommender” and from customers.

To start working in the system, the expert creates a profile that minimally contains a unique username and password for entering the system.

The basic structure of the expert profile fields can be as follows:

a. FULL NAME;

b. Education and competency level;

c. A set of topics;

d. Selected strategies (presence on the platform) by topics;

e. rating index;

f. the ability to give and receive recommendations;

g. CRM system (request history, customer register, consultation register and consultation record);

h. “Black list” (unwanted customers for an expert).

After registering an expert in the system, and filling out profile profile data, the expert selects the topics (industries) of consultations and describes or indicates the level of ownership of the industry material. It is assumed that the expert selects topics in which he is competent from the topics proposed by the system. An expert can also indicate one or several topics that are not in the system. The choice of topics can be carried out either by choosing from a complete list, or by using keywords or phrases, followed by the use of relevant topics presented in the tag cloud. A tag cloud can be formed by summarizing information from information sources related to user requests, or formed by summarizing user questions asked earlier, and also determined by experts' suggestions.

If, after selecting tags from the tag cloud, the system can determine the expert’s specialization from those specified in the system, the system offers the expert a name or description of the industry. An expert can refine the industry by continuing to select tags from the cloud. Also, when registering, an expert can set his own assessment of the level of subject ownership using the scale provided for in the system. For example, a level can change discretely from a beginner level to a professional level. In the process of forming a profile. Prior to the final completion of registration, the expert may be provided with background information reflecting the cost of the consultation quantum, as well as the parameters of the time spent on the consultation. The expert can agree with the recommended amount, or determine your tariff or tariff per time slot, in relation to the format of the consultation.

An important property of the proposed invention is the quantization of time spent on interactive consultations, with the exception of time spent on textual consultations. When implementing the invention experimentally, it was found that it is advisable to quantize the time spent by quanta with a length of 100 to 130 seconds. Quantization by quanta with a length of less than 100 sec leads to a loss of distinguishability of quanta in pricing, due to the excessively large number of quanta spent on 1 consultation. Quantization with quanta longer than 130 sec leads to significant underutilization of the paid time, since most of the consultations are conducted within 5-10 minutes. Preferred to use is quantization by quanta with a length of 121 sec, which symbolizes the direct connection of the expert with the user (1: 1).

To create a professional profile in the selected topics, an expert, before connecting to the system, passes qualification consultations, usually two, which are conducted by experts with the status of "Recommender". Based on the results of the consultations, the expert is assigned a rating index. Based on the results of such qualification consultations, the starting rating index is determined based on the processing of the information provided by the expert himself. The updated rating index can be used in the system for pricing purposes and to search for experts for consultations.

In the particular case of implementation, the expert rating index is formed in several stages:

1. The starting rating index is formed when an expert is registered on the platform (in the system). The platform informs the expert about what a starting rating index is, how it is formed;

2. The expert rating index is assigned for the first time after conducting qualification consultations and evaluating recommenders. The platform informs the Expert 1) by what algorithm Recommenders evaluate the expert’s test consultations and 2) why such a rating index is formed and 3) provides recommendations on the possibility of increasing the rating index.

Rating Index Calculation Algorithm:

The rating index is formed depending on the factors listed below:

1. starting rating index, determined for all topics declared by the expert;

2. ratings of recommenders affixed based on the results of test consultations;

3. The number of ratings and their values affixed by users-customers based on the results of consultations. In a particular case, the arithmetic mean of the values given by users is used as an estimate;

4. frequency of personal inquiries to the Expert;

5. the relative number of conversions of requests for consultations in the consultations;

6. number of user recommendations;

7. the number of recommendations of other experts in the same specialty;

8. set weights used as shown in table 1 below.

Table 1.

the criteria assessment criterion weight final starting rating index on all topics 8 five% 0.4 evaluation by test consultants on all topics 7 five% 0.35 quantity and estimates (by the Customer) of the consultations (arithmetic mean) 6.7 40% 2.68 frequency of personal inquiries to the Expert (personal / all inquiries) 0.1 ten% 0.01 conversion of requests into consultations (consultations / all requests)
* not taken into account in the ranking of pros 0.8 ten% 0.08 Number of recommendations from customers (number of recommendations / number of consultations) 0.2 15% 0,03 Number of recommendations from experts in this specialty 0.3 15% 0,045 rating index one hundred% 3,595

Consultation formats can be selected from the list:

a. Text format

b. Audio format;

c. Video conference.

For consultations in various formats, it is possible to use different time slots or standards of time slices (121 sec) spent on a consultation in order to maximize the effectiveness of the expert’s interaction with users in the format of the platform, as well as for the expert’s convenience. For example, an audio format can be selected for the morning hours, text for the daytime hours, and video chat for the evening hours.

The semantic and morphological profile of an expert or a tag cloud can be represented by a set of semantic, stylistic and morphological characteristics characterizing the industry of the expert and his preferred areas of consultation, formed on the basis of the nature of the consultations he made, the users’ estimates, the number and results of using the arbitration procedure regarding the consultations .

According to the results of the expert tag cloud analysis and statistics processing of consultations conducted by other experts, the platform (system) calculates and informs the Expert about the median cost of a quantum on the selected topic at the level indicated by the expert and the average number of quanta that was used as part of one consultation on this topic . Also, an expert can be informed about the average number of queries per month on topics corresponding to the expert tag cloud. Based on this information, the Expert can choose a strategy for participation in the system. For example, the following strategies can be selected:

1. Beginner (recommended to all new experts on the platform).

The choice of this strategy provides:

- constant information support of the Expert on the platform;

- the search algorithm limits queries:

- only from Customers with a rating level above average;

- the level of coincidence of the expert profile with the request should be higher than the median;

- only the specified consultation formats and only in the specified time slots;

- booking a time for future consultations with a personal request;

- The price recommended by the platform for a quantum and not exceeding the average number of quanta by subjects;

- the number of topics is limited;

- there is no way to refuse consultation without lowering RI

2. Business (for experienced experts).

The choice of this strategy provides:

- if desired (option to be disabled), expert information support

- the search algorithm limits queries:

- only the specified consultation formats and only in the specified time slots. An expert whose slot is approved by the user is removed from the list of available experts;

- booking a time for future consultations with a personal request indicating the types of consultations (audio \ chat \ text);

- the opportunity to propose a replacement type of consultation on a submitted application;

- an expert can change the platform’s recommended price for a quantum and set its number of quanta by topics depending on the chosen strategy:

- an increase in the rating index due to the price per quantum below the median and an increase in the request-consultation conversion;

- monetization of a high rating index due to the price per quantum above the median;

- the number of topics depends on the chosen strategy.

3. Pros (recommenders, celebrities, media personnel).

The choice of this strategy provides:

- if desired (option to be disabled), expert information support;

- the search algorithm limits queries:

- booking a time for future consultations both in search and in person;

- monetization of a high rating index due to the price per quantum above the median;

- the number of topics is not limited;

- an opportunity to refuse consultation without lowering RI.

Such a procedure for assessing competencies and choosing strategies is implemented for each topic that EXPERT decides to choose within its profile.

In addition to choosing a strategy for participation in the system, an expert can suggest and add topics that are not in the general tag cloud. The expansion of the tag cloud allows you to refine your search queries and improve the quality of service.

Upon completion of the formation of the Expert’s profile, primary indexation takes place, in which the expert’s profile is brought into line with the general topics used in the system, indicating the level of ownership of the topics and the selected strategies for the expert’s presence on the platform for each of the selected topics.

If necessary, the Expert Advisor can specify the data of his profile unlimited. Making changes to the profile affects the implementation by the platform of the algorithm for selecting an expert at the request of the customer (user).

Additionally, the Expert Advisor can indicate undesirable or conflicting topics, indicating which requests and from which customers he is not ready to accept. In the absence of such information, the cancellation of the consultation by the expert selected by the system is taken into account as a reducing factor for the expert rating index, and is also taken into account in the platform’s search algorithm.

In the above manner, the experts are connected to the system, a database of experts registered in the system is created and updated, containing, for each expert, a description of the expert’s qualifications, where each description contains a tag cloud suitable for determining the relevance of the topics of the expert’s industry and an expert rating index, quantitatively describing the expert’s qualification level relative to other experts registered in the system.

For the convenience of working on the platform, EXPERT can plan its activity and choose acceptable time slots for it for consultations in the selected formats (text, audio, video chat). At the same time, the platform algorithm takes into account free slots in total for all topics from the EXPERT profile. T.O. upon receipt of the request in the same format for any of the topics in the EXPERT profile, the platform automatically reserves 5 quanta for the subject of the request. T.O. these quanta become inaccessible to other topics within the framework of the EXPERT profile. EXPERT can plan one day and extend the same planning to all days, or plan each day individually. If there are no scheduled slots in the EXPERT calendar on the platform, his profile becomes inactive for search by request. Also, the profile is inactive when: 1) the free time slots and consultation formats do not match (if the request is for an audio format, and EXPERT has text in this slot, the platform algorithm excludes the profile of this EXPERT from the search list).

An additional feature of the platform’s analytics for EXPERT is statistics on the exclusions of his profile from the array as part of the request due to calendar restrictions, format mismatch and price factor. As part of the “analysis of missed opportunities.”

When new experts are added to an already working system, to determine their rating index, an additional proficiency test is carried out by interacting with experts with the status of “recommender”. An automated proficiency testing of experts can be preliminarily carried out by testing or receiving answers to previously asked questions of users, answers to which were given earlier by experts with a high rating. Comparison of the nature of the answers allows you to evaluate the expert rating.

In the process of the system’s work, the rating of the expert indexes is updated based on the results of the assessment of responses by users, made at the end of consultations.

The procedure for evaluating user consultation is mandatory and allows you to improve the quality of the advice provided or to maintain it at a consistently high level.

If the Customer evaluates the consultation unsatisfactorily, the platform launches the Arbitration algorithm. The assessment can relate to both the actions of the expert and the operation of the system as a whole.

Examples of rating scales are given below in table 2.

Table 2.

An expert • As far as the answer was:
• complete (1-10),
• concise (1-10)
• useful (1-10)
• For video chat / audio additional questions:
• The expert was friendly (1-10)
• The expert was correct (1-10) Platforms • Fast (1-10),
• Understood (1-10),
• Convenient (1-10),
• Wishes

Additionally, the user can submit informal proposals for changing the work of the expert and the system, which can be taken into account when improving the system.

The expert’s overall rating of the ratings set by the user based on the results of the consultations is available for review by the expert, which also improves the quality of the platform’s work compared to well-known counterparts.

The assessment of the expert’s activity and the rating of the expert is determined taking into account the weighting factors applied to the assessed marks, depending on the importance of the parameter being evaluated. In an advantageous embodiment of the system, the same rating scales are used for all ratings, for example, five-point or ten-point. For the parameters: completeness, conciseness, usefulness, friendliness and correctness of the answer, the number of voters, the arithmetic average of the ratings related to the parameters, as well as the total number of ratings can be taken into account. Expert rating can be formed as the arithmetic average of user ratings, using weighting factors. As shown in the table in the examples below, in table 3 and table 4.

Table 3.

consultation evaluation assessment criterion weight final full (1-10), five ten% 0.5 concise (1-10) 8 20% 1,6 useful (1-10) ten 40% 4 friendliness of an expert (1-10) five ten% 0.5 expert correctness (1-10) ten 20% 2 total for consultation one hundred% 8.6

The consultation score of 8.6 is above average, and since none of the parameters received a score below 5, the arbitration procedure does not start.

Table 4

consultation evaluation assessment weight criteria final full (1-10), five ten% 0.5 concise (1-10) 8 20% 1,6 useful (1-10) 3 45% 1.35 friendliness of an expert (1-10) five five% 0.25 expert correctness (1-10) five 20% 1 total for consultation one hundred% 4.7

The assessment of the consultation in table 4 is unsatisfactory, the assessment by the criterion of "usefulness" and the overall assessment of the consultation are below "5". The arbitration procedure is launched, and since the criterion of "utility" is examined, it is possible to study the nature of the response using speech or text recognition algorithms in order to identify the presence and sufficiency of information provided by the expert to disclose the nature of the answer to the question posed by the user. Morphological and syntactic markers characteristic for this topic can be used as supporting materials, the standards for which are formed on the basis of parsing and analysis of consultations of other experts on similar topics with the highest possible rating. The next stage of the analysis is an automatic comparison of the identified morphological and syntactic markers of consultation with the results of arbitration on similar topics, conducted by an expert with the status of “recommender”.

If you use the analysis of the parameter "friendliness" can also be used tools for the analysis of speech and text, revealing vocabulary or intonation, which can be considered as invalid. The essence of the answer may not be checked.

Analysis of the completeness of the consultation can be carried out by analogy with “usefulness” to check if the answer contains a minimum set of information related to the topic of the user's question, “laconicism and“ correctness ”are analyzed by analogy with“ friendliness ”. Based on the results of using the arbitration procedure, the system can formulate recommendations for the expert on changing the style of communication with users. For example, it may be recommended to exclude the words parasites from communication, add “confidence” when submitting recommendations. An analysis of professional aspects can also serve to improve the quality of the expert’s work.

In a particular case, ratings can be set on a 5-point scale, where 1 is poor; 2 - unsatisfactory; 3 - satisfactory; 4 - good; 5 is excellent.

If as a result of both stages the unsatisfactory level of consultation is confirmed, then the amount of payment for the consultation reserved on the Customer’s virtual account is unlocked and returned to the Customer. The expert receives a report on the arbitration procedure, shortcomings discovered and recommendations on the prevention of such excesses. An unsatisfactory assessment of the consultation, confirmed during the arbitration, reduces the rating index of the expert. If the expert rating index decreases below the limit set by the system, the expert is excluded from interaction with the system, for example, by blocking the profile.

An example of accounting for a negative (below average) rating for a consultation showing below in table 5.

The expert conducted 8 consultations and his index is 67 (i.e. the average score for a consultation is 8.37).

The rating for unsatisfactory consultation (9th) is shown in Table 5.

Table 5.

consultation evaluation assessment criterion weight final full (1-10), 1 ten% 0.1 concise (1-10) five 20% 1 useful (1-10) 1 45% 0.45 friendliness of an expert (1-10) five five% 0.25 expert correctness (1-10) five 20% 1 total for consultation one hundred% 2,8

After taking into account the consultation, the index will be: 67 + 2.8 = 69.8 and the average score for the consultation will be 7.7

The expert index is taken into account by the subject search algorithm when a request is received. If the index of the subject expert is significantly lower than the median value, the Expert Advisor is temporarily excluded by the platform algorithm from the active roster of experts on the subject, about which the expert is notified and advised of the platform: clarify the list / change of subjects, receive additional education, and work with the coach (i.e. hours on the platform), etc.

If, as a result of both stages of the analysis, the unsatisfactory level of consultation is not confirmed, then the amount for the consultation reserved on the Customer’s virtual account is unlocked and transferred to the Expert’s virtual account minus platform services, and the Customer receives a full report on the arbitration procedure. Arbitration in favor of the Expert lowers the rating index of the Customer.

If as a result of both stages of the analysis it is impossible to give an unambiguous assessment of the level of the consultation, then the algorithm involves at least 2 (two) platform experts on this topic with the status of “recommender”. Experts with the status of “recommender” evaluate the request and consultation on it, form their own opinion and give their opinion on the correctness of the unsatisfactory assessment of the consultation by the customer, and also give their recommendation to the expert. Each episode involving experts with the status of “recommender” for arbitration is taken into account in the “arbitration” algorithm for its enrichment and refinement.

The weights (significance) of the criteria (parameters) can be changed dynamically, for example, by determining the influence of weight on the number of users involved or depending on topics.

The database of registered users is formed in the usual manner for the prior art with the provision of indicating identification data, a password for entering the system, as well as bank details that allow for automated financial transactions.

The user can register in the system with a mobile phone number and a permanent password, and the user can log in to the system by entering a mobile phone number and a permanent password or by entering a mobile phone number and a new unique SMS password. To authorize the user in the system, the user is presented with a registration form in which the user enters a mobile phone number and a permanent password or a generated SMS password to confirm the mobile phone number.

In the personal account provided to the user and the expert by the system, the user and experts can supplement the data and edit such data, for example, personal data. Also, the user, for example, in his personal account, is provided with funds for entering card details to reserve the amount of internal currency and pay for the consultation.

The user's login, in particular, which is the user's mobile phone number, password, file with the user's avatar, name, surname, patronymic, user's mobile phone number (in the particular case, which is the username), user's email address (E-Mail), gender, date of birth, country, city, etc., are stored in the corresponding fields of user registration information and user personal data: “Login”, “Password”, “Photo”, “First Name”, “Last Name”, “Middle Name”, “Mobile” ”,“ E-Mail ”,“ Sex ”,“ Date of Birth ”,“ Country ”,“ City ”, etc., moreover, when changing the password, a copy of the new password is automatically saved in the“ Confirm New Password ”field of the user registration information .

For all participants in the system related to the essence of the proposed invention, their membership in groups is established in accordance with the roles in the system, for example: users (customers), experts, moderators, administrators.

Additionally, the amount of rights allowed for the group can be set. for example, moderators and experts can be given voting rights for an expert rating or for the authority of an expert. Due to the fact that experts can use the system as users, when using the system, warnings can be used that indicate the role of an individual performed at any time.

After connecting users and experts to the system, the system is ready to work.

To get started, the user logs in to the system using the user's credentials and ensures that the question is transmitted to the system in the format selected for communication.

In the particular case, depending on the importance of receiving the right or detailed answer, the user can select an expert by indicating the exact subject or value of the minimum rating index of the expert. The user can also select an expert according to the category of the question asked in order to get advice from the selected expert, including in the selected time slot.

After the Customer (or an unauthorized user) selects the expert and his free time slot, the Customer enters the card details to reserve the amount of domestic currency equal to the price of the expert’s consultation. The customer or an unauthorized user cannot reserve a time slot for a consultation until he pays, and an unauthorized user cannot get a consultation from an Expert without registering. Further, the system reserves the amount on the Customer’s virtual account and the Expert confirms the reservation of the time slot, after which it carries out a consultation in the specified time slot.

When a request is sent by the Customer (user), the platform recognizes the subject of the request by parsing the user’s received question for essential features with sufficient detail to determine whether the expert industry’s question is determined by the tag cloud. Further, the system determines a set of experts on this topic and informs the Customer about the price range for a quantum on the subject of the request. The customer selects the price range of the quantum value, the type of consultation (text \ audio \ video) and starts the search for an expert.

Within the chosen range of quantum cost, the platform algorithm searches for the three best matches according to the semantic and morphological profile of EXPERT, that is, it selects an expert whose tag cloud according to morphological characteristics matches the text of the question as much as possible.

The Expert Advisor with the maximum match index (RI by topic, consultation format, calendar, the Customer is not included in the Expert's unwanted list) will be sent to the Customer’s Avatar and subject of the request. After the expert clicks “yes”, the Customer is notified that the Expert is ready to contact and the Expert’s short business card is sent (photo, RI and a set of expert topics) the countdown is turned on before the consultation begins. The maximum time before the start of the consultation from the start of the countdown is 300 seconds.

If Expert 1 rejected the request, a similar request is sent to Expert 2: Customer’s profile picture and subject of the request. If Expert 2 rejects the request, a similar request is sent to Expert 3.

If Expert 3 rejected the request, the platform asks the Customer to clarify or reformulate his request.

After agreeing on the question, selecting an expert and obtaining consent from the user with the conditions for providing an answer, blocking the funds in the user’s account in the amount of the proposed cost of the answer, the timer is set, which is set for the specified response time at the beginning of the user providing the expert’s response, and the user asks the expert to evaluate the expert’s response to after a predetermined time, the answer, write-off of funds from the user’s account with a positive assessment of the answer and taking into account the assessment of the answer in the rating index of the expert.

The timer can be set discretely using the minimum response time - quantum, which is a predefined system parameter. The number of quanta that the EXPERT needs to answer the query can be determined by the system in accordance with the complexity of the subject of the request, or determined by the expert himself using the system interface, choosing from 1 to 5 quanta in the interface (5 is the maximum number of quanta for one consultation on platform).

In an alternative option for choosing an expert, when the Customer sends a personal request to the Expert, the platform informs the Customer about 1) the price per quantum according to the subject of the request from this expert and 2) the nearest available slots in the calendar.

When the Customer confirms the price for the quantum and the slot, the platform sends a request to the Expert: the Customer's Avatar and the subject of the request are sent. After the Expert clicks “yes”, the Customer is notified that the Expert is ready to contact and the Expert’s short business card is sent (photo, RI and a set of expert topics) the countdown is activated before the consultation. The maximum time before the start of the consultation from the start of the countdown is 300 seconds.

After the consultation, the Customer evaluates the Expert’s response according to the specified response criteria. If the total assessment for the consultation is more than 3.0, then the system transfers the reserved funds to the Expert’s virtual account, in particular, minus the commission (deductions for using the system), and the percentage of deductions is automatically deducted before the transfer of funds to the Expert. If the total assessment for the consultation is less than 3.0, the system offers 50% of the refund to the Customer, 50% of the transfer from the reserved funds to the Expert, minus the commission (deductions for using the system). If one of the parties does not agree, the consultation is automatically sent to the moderator and arbitration is carried out by the system moderator. In a particular case, if the Customer is not in the specified time slot, the system transfers the reserved funds to the Expert’s virtual account minus the commission (deductions for using the system) or in the absence of the Expert to the specified time slot, the system returns the reserved funds to the Customer’s virtual account

If the Expert Advisor rejected the request, the platform will clarify with the Expert the reason for the rejection of the request and offer the Customer if the Expert did not understand the request - clarify or reformulate his request for this Expert, if the Expert has entered the Customer on an undesirable list - use the search algorithm, if the Expert has made a mistake in creating the calendar - will offer a new slot to the Customer for approval. The calendar saves the questions of Customers and topics of consultation, as well as related data described in the framework of the present invention.

Alternatively, the use of morphological analysis tools is not required, but the interaction of the customer and the expert is carried out according to the rules of the system.

To increase the attractiveness of the system for experts, the system contains means of recommendations, as well as means of setting parameters for the interaction of the expert and the system.

An expert may request a service on the platform:

1) analysis of missed opportunities

2) recommend the optimal strategy for the presence of an expert on the platform based on predictive analysis: analysis of the dynamics of topics on the platform (query frequency, conversion, cost per quantum and number of quanta, number of experts, etc.), the dynamics of the expert himself in comparison with the sample experts on the platform and their desired / target indicators in future periods (income, frequency of consultations, price of a quantum, number of quanta by subject).

3) connect the "assistant" function - the platform will send situational pop - up messages to the expert in the application about what and how to change: the price of a topic quantum, number of quanta by theme, number and name of topics, conversion, speed, tonality and effective filling consultations, etc. to achieve the target results. The algorithm of this function uses an array of platform data, identifying implemented the most effective strategies for the presence of experts on the platform in the selected time period.

Using the platform for consultations, the expert generates an array of data, and the platform algorithm refines its semantic-morphological reference, thus ensuring continuous improvement of indexing algorithms.

For the White label: 0) one expert - one profile 1) the expert ALWAYS fills out the profile himself 2) RIE is charged regardless of the participation of E in WL. WL admin does NOT affect RI and tag cloud. 3) The WL admin affects a) inclusion of E in the box b) management of LC E criteria in terms of: quantum cost and calendar, customer restrictions, branding of selected LC E. zones. In terms of CRM: Customers and consultation materials are the property of WL upon exit E from the box are automatically deleted from the LC E, but stored in WL. An expert loses the history of only those consultations and contacts that were implemented in the course of work within the legal entity.

In FIG. 2 shows a functional diagram of the interaction of elements and blocks of the system.

As shown in FIG. 2, a user module (in this case, “participants”) 204 is a module for storing data about system participants, including. about users, experts, moderators. Using the module provides the ability to determine roles and access rights when interacting with the system.

The expert competency module 208 is a unit for storing and processing data sets that determine the expert’s competencies, including the tag cloud used for automated information retrieval.

The store module (billing system) 212 contains all the main functional elements of the trade (currencies, units, orders, order statuses, prices, accounts). The statuses of orders can include “Awaiting consultation” (Amount on the Customer’s account is reserved. Consultation is expected.), “Fulfilled” (Consultation completed. Reserved funds were transferred to a virtual account.), “Arbitrage” (Funds are blocked, Customer is dissatisfied. Arbitration is expected by the moderator .), “Arbitration settled” (Arbitration decided by the moderator. The reserved funds were transferred to the virtual accounts of the Expert and / or the Customer in accordance with the decision of the arbitration.). Information about Customers who paid for the Expert’s service is stored in the store’s buyers section. Information about payment for services is stored in the order payment section.

The catalog of services of 214 Consulting Experts contains descriptions of the services provided by the system and their formalized representations. The service catalog 214 provides means for supplementing and updating the list of services, including a description of the service; service name; competency to which the service relates; price per service quantum; base price.

Consulting 216 is a dynamically changing and replenishing database of transactions related to the activities of experts. The database contains information about each consultation, including the unique identifier of the consultation, the name of the consultation; user and expert identifiers, user question, consultation duration in quanta (calculated automatically as a period from the start date and time to the end date and time of the consultation); date and time of the start of the consultation; date and time of the end of the consultation; cost of consultation (calculated automatically as a product of duration by the price per unit of quantum / base price); service name (binding to an element of the information block "Catalog"); assessment for each criterion of the Expert’s response (completeness, conciseness, usefulness, friendliness, correctness); the total rating of the consultation (the arithmetic average of the ratings for each response criterion) and information about the possibility of starting the Arbitration process and the status of the Arbitration. Additionally, the transaction database contains links to graphic and media files related to consultations obtained using WebRTC.

Arbitration 218 is a means of recording data relating to arbitration processes. Identifiers of consultations (disputed consultations) are used as the information used, as a result of which the arbitration process was launched, as well as, for each such consultation, the completeness of the process and the results of the consideration of the user's claim. A consultation is considered controversial, after which one of the parties does not agree with the quality of the consultation provided by the user or the cost of payment. A situation is not ruled out in which the expert confirms the low quality of the services that were provided to him by the user.

Assessment of the platform 220 is a means of registering and processing assessments of participants in the system (platform), in particular, Customers and Experts. The module is used to register estimates of system performance from users and experts, as well as the results of processing estimates, for example, the number and arithmetic average of the ratings for each indicator used in the evaluation.

Social network (chat) 234 is a module for interacting with participants using instant messages, including for exchanging messages between a user and an expert in the process of providing text advice. The module provides for the possibility of exchanging attached files with archiving data used in the provision of counseling. In particular, it is possible to exchange files with the following extensions: doc; docx; xls; xlsx; ppt; pptx

Push and Pull 222 is a module for sending push notifications. One example of a functional module used to notify system participants using an instant alert system. It is mainly used for reminders of upcoming consultations, as well as about the passage of payment processes for consultations.

Mail 226 is a module for sending and receiving mail messages. Another example of a module designed to notify participants and collect information.

Event Calendar 230 is a module for planning and booking consultations. It is used in conjunction with warning and data collection modules. Participants can download from the calendar information about consultations in which they participate, choose the time for consultations. The calendar also displays information about the time intervals chosen by experts for consultations.

In FIG. Figure 3 shows an example of the dynamic use of system functions during operation.

Step 1 (303): The expert (client A) 313 wants to start communication with the customer (client B) 323. Client A 313 creates an offer SDP offer 333, which contains a random password for a peer-to-peer connection, and sends it to 1C-Bitrix 343 server. 1C-Bitrix 343 server sends “Offer SDP” 333 to client B 323.

Step 2 (353): Client B 323 sends a “Answer SDP” response 363 to a 1C-Bitrix 343 server. 1C-Bitrix 343 server sends a “Answer SDP” 363 to a client A 313.

Step 3 (373): Both clients (313, 323) now know about each other, but do not know their public IP addresses. The STUN server (383) at the WebRTC protocol level reports for each client (313, 323) its IP address and port for the public network.

Step 4 (393): Client A 313 and Client B 323 exchange information about public IP addresses through the server of the system, after which the server provides the organization of a communication channel.

In FIG. 4 shows an exemplary version of the algorithm for generating a schedule of available slots in the expert profile. The scheduling of available slots in the expert profile begins at step 404. Next, at step 414, authorization is performed, and then at step 424 it is checked whether the expert has added topics. If it is determined in step 424 that no topics have been added, then the topics are added in the topics adding section in step 434. If it is determined in topics 424 that the topics have been added, then in step 444 a schedule is created (determination of communication restrictions), then in step 454 by means of the system, a weekly timeline is formed, where the day is divided into slots by the duration of one quantum - 121 seconds. Thus, in a day - 714 quanta. Next, in step 464, the presence of inaccessible slots is checked, and if there are any, in step 474 a restriction is formed: the period during the day (slot), the type of communication, and the schedule for the available slots in the expert profile ends in step 494. If in step 464 if there are no inaccessible slots, then in step 484 the presence of slots with a limited type of communication is checked and, if there are any, the process of creating a schedule of available slots in the expert profile goes to step 474, and if they are missing, the process of creating a schedule of available slots in the expert profile ends in step 494.

In FIG. 5 and FIG. Figure 6 shows the associated personified pricing algorithm and the quantum calendar and the algorithm for adding topics to the expert profile, respectively.

So in FIG. Figure 5 shows one of the options for the personified pricing algorithm and the quantum calendar, which starts at step 505. Next, at step 510, registration, authorization, filling out the profile (field, determining geolocation) is performed and requests from which companies or individuals are not accepted, and then Step 515 verifies that the expert is a novice. If it is established in step 515 that the expert is a novice, then in step 525 a restriction of three subjects is established, after which, in step 540, the formation of subjects is carried out using the established restriction of three subjects. If it is determined in step 515 that the expert is not a novice, then in step 520, it is checked whether the expert is a professional (pro). If it is determined in step 520 that the expert is not a professional, then in step 535 a five-topic restriction is set, after which, in step 540, the formation of topics is carried out using the established five-subject limit. If it is established in step 520 that the expert is a professional, then in step 530 there are no restrictions on the number of topics and in step 540, the formation of topics is carried out without limiting their number. At step 545, it is checked whether the limit on the number of topics has been reached and, if so, as shown in FIG. 5, the personified pricing algorithm and the quantum calendar are completed in step 580 and go to step 636 (Fig. 6) of the related topics adding algorithm to the expert profile shown in Fig. 6. If in 545 it is established that the restriction on the number of topics has not been achieved, then in step 550 the selection or indication of the topic is carried out; the choice of one of the types of communications: text, audio, video (multiple choice); Assessment of topic ownership from zero to ten ([0, 10]). Whether the topic is new (for the platform) or not, is checked in step 555 and, if the topic is new, then in step 560 the expert offers his quantum price, after which, in step 565, the subject and price for one quantum are saved in the expert profile. Next, in step 570, the need to add a new topic is checked. If it was determined in step 570 that a new topic needs to be added, then go to step 540, and if you do not need to add a new topic, then go to step 580. If it was determined in step 555 that the topic is not new, then in step 585 obtaining all experts on a given topic, receiving costs for one quantum of each expert on the selected type of communication, calculating the median value of the cost of one quantum on the topic and type of communication, and proposing to the expert. Next, in step 590, it is checked: does the expert agree with the proposal for the cost of one quantum. If you do not agree, then go to step 560, and if you agree, then go to step 565.

In FIG. Figure 6 shows the algorithm for adding topics to the expert profile. The process of adding topics to the expert profile begins at step 606. Next, at step 616, authorization is performed, and then at step 626, the expert is added to add topics. If the expert has not added topics, then go to step 540 (FIG. 5) of the associated personified pricing algorithm and the quantum calendar shown in FIG. 5. If the expert has added topics, then go to step 636, in which a schedule is created (in particular, determining communication restrictions), after which, in step 646, a weekly timeline is formed, where the days are divided into slots by the duration of one quantum, in in particular, 121 seconds (thus, in the days of 714 quanta). Next, in step 656, a check is made for the presence of inaccessible slots. If inaccessible slots were detected in step 656, then in step 676 a restriction is formed: the period during the day (slot) and the type of communication, after which in step 686 the process of adding topics to the expert profile is completed. If inaccessible slots were not found in step 656, then in step 666 the presence of slots with a limited type of communication is checked. If the presence of slots with a limited type of communication is established in step 666, then go to step 676. If in step 666 it is established that there are no slots with a limited type of communication, then proceed to step 686.

in FIG. Figure 7 shows another variant of the personified pricing algorithm and the quantum calendar, which starts at step 705. Next, at step 710, registration, authorization, filling out a profile (field, determining geolocation) is performed and requests from which companies or individuals are not accepted, and then in step 715 verifies that the expert is a novice. If it is established in step 715 that the expert is a novice, then in step 725 a restriction of three subjects is established, after which, in step 740, the formation of subjects is carried out using the established restriction of three subjects. If it is determined in step 715 that the expert is not a novice, then in step 720 it is checked whether the expert is a professional (pro). If it is established in step 720 that the expert is not a professional, then in step 735 a five-topic restriction is set, after which, in step 740, the formation of topics is carried out using the established five-subject limit. If in step 720 it is established that the expert is a professional, then in step 730 there are no restrictions on the number of topics and in step 740 the formation of topics is carried out without limiting their number. At step 745, it is checked whether the limit on the number of topics has been reached and, if so, as shown in FIG. 7, the personified pricing algorithm and the quantum calendar are completed in step 780. If it is established in 745 that the limit on the number of topics has not been reached, then in step 750 a topic is selected or indicated; the choice of one of the types of communications: text, audio, video (multiple choice); Assessment of topic ownership from zero to ten ([0, 10]). The new topic (for the platform) or not, is checked in step 755 and, if the topic is new, then in step 760 the expert offers his quantum price, after which, in step 765, the topic and cost for one quantum are saved in the expert profile. Next, at step 770, the need to add a new topic is checked. If it is determined in step 770 that you want to add a new topic, then you go to step 740, and if you do not need to add a new topic, then you go to step 780. If in step 755 it is determined that the topic is not new, then in step 785 obtaining all experts on a given topic, receiving costs for one quantum of each expert on the selected type of communication, calculating the median value of the cost of one quantum on the topic and type of communication, and proposing to the expert. Next, in step 790, it is checked whether the expert agrees with the proposal for the cost of one quantum. If you do not agree, then go to step 760, and if you agree, then go to step 765.

In FIG. 8 and FIG. Figure 9 shows the associated algorithm for determining the relevant expert for a given request and the algorithm for searching an expert for a request.

So in FIG. Figure 8 shows an algorithm for determining the relevant expert for a given request.

The process of determining the relevant expert for a given request begins at step 810, after which at step 815 the request is entered (in text format or by voice). Next, in step 820, the request is transmitted to the search engine, in particular, to step 909 (FIG. 9) of the related query expert search algorithm shown in FIG. 9, wherein a voice input request can be converted to a text data format (into one of the known data formats). In step 825, a list of experts in subjects is obtained, after which, in step 830, the presence of an expert in the customer’s blacklist is checked. If it is determined in step 830 that the expert is on the customer’s blacklist, then return to step 825, otherwise, go to step 835. In step 835, it is checked whether the value of the expert’s quantum in the subject is within the range, and if not, then it proceeds to step 825. If it was determined in step 835 that the value of the expert’s quantum in the subject is within the range, then in step 840 the presence of the nearest available slots for the expert is checked. If it is determined in step 840 that the expert does not have the nearest available slots, then proceed to step 825. If the expert has the nearest available slots, then in step 845 it is checked whether the list of experts is empty. If the expert list is empty, then at step 850 the search parameters are changed and go to step 815. If it is determined at step 845 that the expert list is not empty, then at step 855 a list of ten relevant experts is formed. Next, at step 860, a consultation request is sent to all experts from the list, and at step 865, a check is made to see if the expert from the list confirmed the consultation within thirty seconds. If the expert from the list has not confirmed the consultation within a predetermined time, for example, thirty seconds, then proceed to step 860. If at step 865 it is established that the expert from the list has confirmed the consultation within a predetermined time, for example, thirty seconds, then at step 870, the nearest available slot of the selected expert is booked and a countdown is recorded, after which the process of determining the relevant expert for a given request ends at step 875.

In FIG. 9 shows an exemplary version of an algorithm for searching an expert on demand.

The search for an expert on request begins at step 909. At step 919, an incoming text request (question in text format) is received, and then at step 929, the line (in particular, each line) of the incoming text request is divided into tokens and received ( in particular, the formation) of an array of tokens of size N. In step 939, which is repeated N times, an N-search is performed on the token N (x) and obtaining (in particular, the formation) of the issuance (list) of experts in topics. Next, in step 949, the N-list of results for all N-searches is added up, and in step 959, the final list is sorted by the expert rating index, which is passed to step 825 (Fig. 8) and the search for the expert on request ends in step 969.

Claims (32)

1. A system for providing expert advice in real time, containing: a database of experts registered in the system, containing, for each expert, a description of the expert’s qualifications, where each description contains a tag cloud suitable for determining the relevance of the expert’s industry topics, and an expert rating index that quantifies the expert’s qualifications relative to other experts registered in system a database of registered users containing identification data and bank details of users; a tool for receiving user questions in a natural language format; a tool for parsing questions, providing a parsing of each of the received questions for essential features suitable for analysis on the correspondence of a question of an industry of an expert defined by a tag cloud; means of selecting an expert whose industry is as close as possible to the subject matter of the issue; means for managing a data transmission channel from an expert to a user; a processing center designed to lock funds on a user’s account before activating a data transfer channel and write off funds according to a user-approved result of receiving a response from an expert by a user; means for setting a response timer configured to determine a time limit for an expert response not exceeding a predetermined time limit; means for determining the cost of the response, made determining the cost of the response in accordance with a predetermined time limit and rating index of the expert; feedback means made by the user receiving an assessment of the quality of the response; a means of refining expert rating indices made by qualifying expert rating indices taking into account assessments of the quality of answers and complexity of questions, and a means of accounting for active experts of the system, made forming a list of available experts from experts registered in the system; the system is designed to provide: receiving a question from a user registered in the system using identification data; user selection of expert rating index; Parsing the received user’s question for significant features with sufficient detail to determine whether the expert’s industry’s question is determined by the tag cloud, for significant features, the choice of experts from among the available experts whose tag clouds and rating indices are most consistent with the selected rating and the essential features of the question; the selection of one expert whose tag cloud according to morphological characteristics matches the text of the question as much as possible; transfer of the conditions for providing a response in the form of information about the selected expert, the time limit and cost of the response to the user; after obtaining consent from the user with the terms of the response, blocking the funds in the user's account in the amount of the proposed cost of the response; transfer of the question to the selected expert; starting a timer set for a given response time when starting to provide the user with an expert response; request from the user to evaluate the expert’s response after a specified response time; debiting the user’s account with a positive assessment of the response and accounting for the assessment of the response in the rating index of the expert. 2. The system according to p. 1, characterized in that it is designed to ensure that funds are unblocked on the user's account and the terms of the answer are repeated to be determined if, after the time limit for the response has been reached, the completeness of providing the expert response to the user has not been ensured. 3. The system according to claim 1, characterized in that it contains automated arbitration tools that provide formal comparisons of the expert’s question and answer, unblock the funds in the user’s account when the answer doesn’t correspond to the user's question, and form an expert’s answer assessment, ensuring that the expert’s answer is rated in the rating expert index. 4. The system of claim 1, wherein the expert tag cloud is divided into segments, each of which has an expert rating index corresponding to a particular industry. 5. The system of claim 1, comprising means for expanding the expert tag cloud. 6. The system under item 1, characterized in that it provides for planning consultations for the appointed time. 7. The system according to p. 1, characterized in that it contains automated arbitration tools that provide formal comparison of the question and the expert’s answer, unlock the funds in the user’s account when the answer doesn’t correspond to the user's question and form the expert’s response assessment, ensuring that the expert’s answer is rated in the rating expert index.

Images