CC BY
202Анализ использования чат-ботов в энергетической индустрии
Косова Екатерина Михайловна
студент, кафедра «Реклама и связи с общественностью», Московский государственный институт международных отношений (университет) МИД Российской Федерации, ekatherine.kosova@protonmail.com
Сегодня способность компаний установить двустороннюю связь с целевыми аудиториями постоянно оставаться в контакте с ними играет жизненно важную роль. В этих условиях на помощь бизнесу приходят чат-боты. В данной статье представлены основные преимущества использования чат-ботов как инструмента взаимодействия компании и целевых аудиторий на примере энергетической отрасли как одной из наиболее бюрократизированных отраслей промышленности, слабо подверженной мгновенным изменениям в области коммуникаций. Также в статье рассматриваются основные технологии, которые используются при создании чат-ботов. Ключевые слова: чат-бот; энергетическая индустрия; цифровой маркетинг; искусственный интеллект, машинное обучение.
1. Introduction
Total penetration of computer technologies into all spheres of society turned information into the main resource of modern economy. The key factor in the development of this phenomenon, called "network transformation", is the formation of global information space, which has the character of dynamic autopoietic system.
A striking example of autopoiesis is the development of new media, i.e. a special format of media that has evolved in the digital age. The core feature of the new media is the interactivity of work with information, which includes not only the ability to generate information independently, but also participation in its distribution. Thus, the consumer of information can simultaneously act as its creator, thereby conditioning the self-reproduction of the information field.
In this circumstance, the ability of companies interested in digital presence and digital promotion to continuously stay in touch with the user by developing two-way communication, begins to play a vital role. At the same time, it is necessary to take into account that human communication resources are limited by at least two parameters: time (communication is possible during working hours and with a certain delay) and the so-called human factor, i.e. the inability to fully control the process of communication. In this situation, the technology of chatbots comes to the aid, allowing to organize a fully controlled communication in 24/7 mode.
In this paper, we consider the main advantages of chatbots as a means of interaction between the company and the client on the example of the energy industry as one of the most vertical and bureaucratic industries, weakly subject to instantaneous changes in the field of communications, as well as the basic technologies that are used in the creation of chat-bots.
2. Chatbot market overview
Chatbots can be described as a group of machine conversation systems interacting with human users via natural conversational language [1; p. 489]. Current chatbot market size varies significantly according to different researches, however, the numbers are still impressive: it is estimated to grow from $2.6-17.17 billion in 2019 to $9.4-102.29 billion in years 2024-2025 showing consequently a compound annual growth rate (CAGR) of 29.7- 34.75% [2, 3, 4, 5]. So why are chat-bots that popular, especially in B2C segment?
X X
о
го А с.
X
го m
о
to о to о
o
CN
o
CN
O HI
m
X
3
<
m o x
X
Source: Mordof Intelligence
Figure 1. Chatbot market concentration [5]
Center, by 2020, 70% of all interactions between company and clients will be based on emerging tools, including chatbots and mobile messaging [6].
Being a trustful intermediary between the company and its clients, chatbots enables business growth and especially marketing development in several dimensions:
• enhancing user experience via constant 24/7 technical support;
• strengthening market and customer analysis via chatbot query analysis;
• retaining the customer on the company's website and/or application by providing more thorough and interactive user experience;
• expanding the scope of services, e.g. by integrating the ability to make online payments within the chatbot;
• etc.
However, while the previous few years were a real paradise on earth for the development of chatbots and their universal integration, today there is a decline in public interest in this tool. About 7 respondents of 10 declared being tired of chatbots [7], however, we suppose, such results are not enough reason for burring the technology. Actually, this "chatbot fatigue" seems to originate from habituation, as chatbots in recent years have become an integral part of many services, sometimes being used without real need.
Figure 3. News coverage of chatbots over time [2]
Figure2. Reasons companies offer messaging [7]
First and foremost, the economy of instant gratification has trained customers that all their requests, especially requests for information, are met and fulfilled immediately. We lose the sight of the value of information and the idea that information is possible to get after a while, we want answers to our questions here and now, otherwise our attention weakens and we go looking for a more "user-friendly" service. In particular, this is why we are moving away from e-mail to social networks and messengers that allow us to instantly ensure that our message has been delivered and received without leaving a painful information vacuum. In this circumstance, chatbots can actually save the day as they accomplish two functions simultaneously: immediately react to client inquires and even entertain the client if needed, e.g. providing humorous responses or chatting within the interface. According to 2019 Gartner Magic Quadrant for the CRM Customer Engagement
Figure 4. Percentage of businesses using messaging [7]
Presumably, the leading sector in terms of chatbot integration is retail industry, as it needs to react rapidly the most and has a relatively free business structure focused on immediate adaptation to the market and rapid implementation of new solutions. However, even stricter and more bureaucratic industries, albeit with some delay, are beginning to implement chat bots to improve communication with the consumer. In this paper, we will consider several chat bots used in the B2C segment of the energy market.
3. Rule-based chat-bots
For the purposes of our work, we will highlight two main types of chat-bots: rule-based and AI-based [8]. Rule-based chatbots are generally created on the basis of a linguistic model and formal grammars, such as proposed by Noam Chomsky [9; p. 299-300].
The idea underlying the formalization of language as a communication tool is quite simple: in the process of communication, subjects use a finite set of characters that are spoken or written out in a strict temporal order, in other words, form linear sequences, such as words, sentences, etc. If you write down and examine the resulting sequence, you can highlight a number of rules describing the language, i.e. its grammar.
These bots are relatively rigid and slow to develop, as they are based on stative language structures. Hence, most commonly they are used to answer standard questions, such as the FAQ, and less commonly used to build more complex architecture, such as calculators, which help the user calculate the approximate cost of services based on standardized answers. Due to the relative cheapness in creation and maintenance, these bots are actively embedded in both company websites and messengers (Telegram, Facebook, Viber, etc.), allowing the user to interact with the company without leaving the favorite application. Thus, the main advantage of this chatbot type is that they do not require any special traffic for downloading and installation time, do not take up space in memory or on the smartphone screen, which becomes crucial for today's user, for whom a smartphone becomes a tool for work and life.
Most often, energy industry resort to the use of rule-based chat bots, integrating them into messengers (Telegram, Viber) rather than website. In 2018, the Russian company Chelyabenergosbyt, which until 1 July 2018 was the main power supply company and a guaranteed supplier of electricity in the Chelyabinsk region of the Russian Federation, became the first Russian B2C energy company to introduce the use of the chatbot. The chat-bot was created on the basis of the Viber messenger and assumed a purely functional interaction with the user: transmitting meter readings, checking account balance and exploring the information on the meter.
In 2020, against the background of the COVID-19 epidemic and the ensuing policy of widespread self-isolation, a similar and similar in functionality chat-bot was also created on the basis of Viber by another Russian energy sales company, Kamchatskenergo. The bot's functionality is also highly practical and does not allow for full communication between the company and the customer, but this bot, unlike the previous one, provides the possibility of feedback. In order to leave a comment about the bot or company's operation, the user needs to enter the text of the request and send it to the chatbot,
which in return will send a message that the application is registered and accepted for consideration.
Alternative and complex solution offers another new Russian development: Energobot for Telegram. This chatbot is positioned as an online consultant for clients of energy supply companies. The service traditionally allows to transmit meter readings, leave requests for repairs, and learn about power outages in advance. According to the unique commercial offer of the technology, it allows companies to save from 500 000 rubles (~6,840 USD) on the development of mobile applications and from 500 000 rubles on the purchase of telephony due to the functionality of IVR, as well as investments in their subsequent development and maintenance.
4. AI-based chatbots
Another huge class of chatbots is based on using machine learning (ML) and artificial intelligence (AI) technologies. Main concepts used while designing such a bot include parsing, pattern matching, AIML, chat scripts, SQL and relational databases, Markov Chain and different linguistic language tricks such as canned responses, model of personal history of interaction etc [10; p. 73]. Obviously, AI-based chatbots are more complex and sophisticated and are oriented towards a more thorough communication with the user. As all the other machine learning technologies, chatbots require vast amount of training data which is the main obstacle of using them widely.
General logic of client request processing of such a bot may be divided into nine main steps [11]:
1. The system receives a client request to the dialogue management module (DialogManager).
2. DialogManager loads a dialog context from a database.
3. The client request (together with the context) is sent for processing to the natural language understanding (NLU) module, resulting in the determination of the client's intention and its parameters. In case of processing non-text events (buttons, etc.) this step is skipped.
4. Based on the dialog script and the extracted data, DialogManager determines the following most suitable state (block, screen, dialog page) that best corresponds to the client statement.
5. Executing business logic (scripts) according to the specified chat-bot scenario.
6. Call for external information systems, if any, programmed in business logic.
7. Generate a text answer using macro substitution and natural language word matching functions.
8. Saving the context and dialog parameters in Dialog State DB for processing subsequent requests.
9. Sending a reply to the client.
AI-based chatbots are used in energy industry even less than rule-based chatbots, probably due to lack of relevant training data and quite a narrow range of tasks, mostly standard, that are transferred to the chat-bot within the B2C energy services. In 2018, Göteborg Energi, a Sweden-based energy company selling and distributing electricity, district heating, district cooling, natural gas etc, turned to GetJenny, a company specializing in AI-based chatbots. The main needs the chatbot was to meet were [12]:
• providing instant automated support and answers to commonly asked questions;
X X
o
OD >
c.
X
OD m
o
ho o lo o
• freeing up staff from routine tasks to switch to other types of work;
• forwarding of interested customers to the company's website
о
CS
о
CS
О Ш
m
X
3
<
m О X X
Figure 5. Client request processing cycle [11]
The chatbot called Ellis was launched in April 2019 and by this time it has already been able to recognize 75% of clients' queries. As the results of the experiment show, main goals have been achieved, and now companies seek to increase chatbot capacity to serve customers by expanding working ours to 24/7.
What Changed
Chai Opening Hours
Number oí Chars
Customer Service Automation
Chat conversations with a human agent
Change per Month
Unchanged Chatbot supported Irvcagents during office hours
1 increase on average
Before Chatbot 0% Now. montWy average ~604
Before Chalbot 100V With Chatbot -40%
Explanation
Goteborg Energi increased their capacity to answer messages during live chit opening hotrs with a chatbot
Customer messages handled rrwe dtring off« hatrs thanks to increased team capacity
Percentage of the completed conversations by chatbot EHis
Customer service agents are involved in 40% of chat conversations on average
Figure 6. Results of integrating Ellis into Goteborg Energi customer relations [12]
However, creating dialogue applications using only linguistic or machine learning methods is complex, resource demanding and often prohibitively expensive. Hybrid chatbots provide the necessary flexibility and speed to develop an application. We are not going to explore them in this paper due to the lack of practical cases in the energy industry, but we believe it is necessary to note the existence of this option.
5. Conclusion
Despite the fact that the media hype around the use of chatbots is gradually subsiding, their actual popularity is still growing. The latter is greatly aided by the CoVID-19 epidemic, which has challenged even the toughest and most bureaucratic industries to move to remote operation.
In practice, there are two main types of chatbots depending on the underlying technology: rule-based and AI-based chatbots. Rule-based chatbots are mainly aimed at solving standardized and simple tasks and are based on formalizing the natural language and searching for the most obvious and important patterns in it. Al-based chatbots are created for more complex interaction with the consumer and are constantly being automatically
developed to achieve a more "humane" format of communication.
B2C energy sector mainly uses rule-based chatbots. Probably, this phenomenon is connected with a limited range of company and user interaction options, as well as with the specific nature of this interaction. It is mainly of a long-term nature (usually the energy supplier does not change over a long period of time) and boils down to a limited set of routine activities, automation of which is achieved through the use of chatbots.
Chatbot integration: analysis of energy industry cases Kosova E.M.
Moscow State Institute of International Relations Today, the ability of companies to continuously stay in touch with the user by developing two-way communication begins to play a vital role. Under these conditions, chatbots come to save the day. This article depicts the main advantages of chatbots as a means of interaction between the company and the client on the example of the energy industry as one of the most vertical and bureaucratic industries, weakly subject to instantaneous changes in the field of communications, as well as the basic technologies that are used in the creation of chatbots. Key words: chatbot; energy industry; digital marketing; AI, ML. Reference
1. Shawar, B. A., & Atwell, E. (2005). Using corpora in machine-
learning chatbot systems. International Journal of Corpus Linguistics, 10(4), 489-516.
2. CB Insights (2019). Lessons From The Failed Chatbot Revolution And 5 Industries Where The Tech Is Making A Comeback. URL: https://www.cbinsights.com/reports/CB-Insights_Chatbot-Revolution.pdf (access from: 23.04.2020)
3. Chatbot Report 2019: Global Trends and Analysis. URL: https://chatbotsmagazine.com/chatbot-report-2019-global-trends-and-analysis-a487afec05b (access from 21.04.2020)
4. Mai-Hanh Nguyen, (2020). The latest market research, trends,
and landscape in the growing AI chatbot industry. Business Insider. URL: https://www.businessinsider.com/chatbot-market-stats-trends (access from: 22.04.2020)
5. Mordor Intelligence (2020). CHATBOT MARKET - GROWTH,
TRENDS, AND FORECAST (2020 - 2025). URL: https://www.mordorintelligence.com/industry-reports/chatbot-market (access from: 22.04.2020)
6. Zendesk (2019). 2019 Gartner Magic Quadrant for the CRM
Customer Engagement Center. URL:
https://www.zendesk.com/resources/gartner-magic-quadrant-crm/ (access from: 22.04.2020)
7. Zendesk (2020). State of Messaging 2020. URL: https://www.zendesk.com/message/state-of-messaging-2020/#chapter-1 (access from: 22.04.2020)
8. Artificial Solutions. Chatbots: The Definitive Guide (2020).
9. Noam Chomsky & George A. Miller (1968). Introduction to the
Formal Analysis of Natural Languages. Journal of Symbolic Logic 33 (2):299-300
10. Sameera A. Abdul-Kader, Dr. John Woods (2015). Survey on Chatbot Design Techniques in Speech Conversation Systems. (IJACSA) International Journal of Advanced Computer Science and Applications, Vol. 6, No. 7.
11. Кирилл Петров (2018). Разговорный AI: как работают чат-боты и кто их делает. URL: https://habr.com/ru/company/just_ai/blog/364149/ (access from: 25.04.2020)
12. Jenny. Göteborg Energi Enhances Customer Experience with Chatbot Ellis. URL: https://www.getjenny.com/customer-service-chatbot-enhances-customer-service (access from: 27.04.2020)
13. Камчатскэнерго. Инструкция пользователя по работе в Чат-боте. URL: https://kamenergo.ru/fizicheskim-litsam/servisy/chat-bot/ (access from: 27.04.2020)
14. ПАО "Челябэнергосбыт". Чат-боты ПАО "Челябэнерго-сбыт". URL: http://esbt74.ru/internet-priemnaya/boty/ (access from: 27.04.2020)
15. ЭНЕРГОБОТ. URL: http://www.energybot.icl-solutions.ru/ (access from: 27.04.2020)
