Chatbot-Intelligent Conversational Agents T.Mounika, Akshita Sangam, Shivani Lakhane, Mr.Rajasekhar Sastry, Dr.B V Ramana Murthy and Mr.C Kishor Kumar Reddy. Stanley College of Engineering and Technology for Women,Hyderabad. [email protected],[email protected],[email protected], [email protected],[email protected] and [email protected]Abstract Chatbots are one class of intelligence, conversational software agents activated by natural language input . They provide conversational output in response and can also execute tasks if commanded. Although chatbot technologies have existed since the 1960’s and have influenced user interface development in games since the early 1980’s, chat bots are now easier to train and implement. This is due to plentiful open source code, that is widely available for developing platforms, and implementing options via Software as a Service (SaaS). In addition to enhancing customer experiences and supporting learning, chatbots can also be used to spread rumours , or attack people for posting their thoughts and opinions online. This paper presents secondary sources for quality issues and attributes as they relate to the contemporary issue of chatbot development and implementation. Finally, quality assessment approaches are reviewed and method based on these attributes and the Analytic Hierarchy Process (AHP) is proposed and examined. Keywords: Chatbot, artificial intelligence, intelligent agents, usability, quality attributes, Analytic Hierarchy Process (AHP), conversational agents. 1. Introduction “The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.” - Weiser (1991) It is not always that the “people” you interact with online are not all people. Customer service chat and commercial social media interactions are managed by intelligent agents. Many of them have been developed with human identities and even personalities. Even though the technology itself is not new, reliable linguistic functionality and availability through Software as a Service (SaaS),addition of intelligence through machine learning has increased its popularity. Between 2007 and 2015, chatbots were participating in one third of all online interactions (Tsvetkova et al., 2016) and the rate at which new chatbots are being deployed has increased since then. Social conversational bots can be used to provide benefits to companies, which are used to reduce time, provide enhanced customer service, increase satisfaction and increase engagement. Sometimes , chatbots are specifically designed to be harmful. For example, networks of fake users (called “sybils” on JASC: Journal of Applied Science and Computations Volume VI, Issue I, January/2019 ISSN NO: 1076-5131 Page No:464
13
Embed
Chatbot-Intelligent Conversational Agents · Chatbot-Intelligent Conversational Agents T.Mounika, Akshita Sangam, Shivani Lakhane, Mr.Rajasekhar Sastry, Dr.B V Ramana Murthy and Mr.C
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Twitter) have been implemented to artificially inflate “follower” counts to increase social status for users
who purchase them and to spread fake news or rumours, and even to intimidate users who express certain
political beliefs.In the 2016 US Presidential elections, up to a fifth of the comments and responses on
Twitter were driven by fully or partially autonomous Twitter accounts.
Due to their flexibility and ease of use , speculations have been made that conversational agents may
be universal user interfaces and may replace apps. In addition ,the chatbots and conversational agents
are anticipated to be important interfaces in Virtual Reality (VR) environments .Hence, it is important to
understand the issues and quality attributes associated with developing , implementing high-quality
chatbots and conversational agents and identifying a mechanism for quality assurance across these
factors.
The rest of the file is arranged in the following manner. First,presentation of the historic context of
chatbots , chatbots background and conversational agents , which are two types of dialogue systems .
Second, outline the methodology for literature rewiew along two topics. a)Quality attributes for chatbots
and conversational agents ,b)Quality assesment approaches . Third and finally , results are produced and
an approach for evaluating the quality of these technologies in terms of key attributes is presented.
2.Background
Chatbot systems that originated with programs like ELIZA were intended to demonstrate the natural
language conversation with a computer. An early stated goal of such systems was to pass the Turing Test ,
in which a human interrogator is considered a computer to pass a test as a human. Primitive systems
like ELIZA used keyword matching and minimal context identification , lacked the ability to keep a
conversation going. Through interactions with program, it was easy to guess that ELIZA was a
computer.Researchers continued to develop the demonstration systems with natural language
capabilities, but none of them were capable of passing the Turing Test. In the 1980’s, ALICE was
created, becoming significant not for its conversational capabilities but because it led the development of
Artificial Intelligence Markup Language . AIML is an eXtensible Markup Language based which
supports most chatbot platforms and services of today’s use.
Chatbots get natural language input which are interpreted through speech recognition
software,sometimes. To get engaged in goal-directed behaviour , they choose to execute one or more
related commands. Chatbots are usually autonomous, reactive, proactive, and social and these factors
make it an intelligent agent. So, in order to adapt to new information or new requests ,most advanced
systems employ machine learning. Ai chatbots or conversational agents are used to automate interaction
between company and server. Chatbots are computer programs that use natural language to communicate.
Now chatbots are shifting towards mobile messenger interface, they are platform independent and use
messenger infrastructure and even download apps. The goal is different from what we think and deal
with. Chatbots and many other kinds of digital bots like Apple-Siri or Amazon-Alexa are aimed to
accomplish on turning on music, checking weather. Bots are popping up every where from facebook and
even to home personal assistants. Advances in natural language processing, machine learning and other
AI technologies. The AI bots are limited to natural language processing and their basic functionality is to
JASC: Journal of Applied Science and Computations
Volume VI, Issue I, January/2019
ISSN NO: 1076-5131
Page No:465
fix data. Bots have ability to become smarter and more proactive to enable more meaningful
communication.
One category of conversational agents are the chatbots .They are software systems that mimic
interactions with real people. They are not given specific shapes or forms of animals, avatars, humans, or
humanoid robots. Conversational agents are a class of dialog systems that have a subject of research
in communications for decades. Interactive Voice Response (IVR) systems are also dialog systems.
Figure 1: Relationships between classes of software-based dialog systems.
One category of conversational agents are the chatbots .They are software systems that mimic
interactions with real people. They are not given specific shapes or forms of animals, avatars, humans, or
humanoid robots. Conversational agents are a class of dialog systems that have a subject of research in
communications for decades. Interactive Voice Response (IVR) systems are also dialog systems. They
are not usually considered conversational mediators as they implement decision trees. AI chatbots or
conversational agents can be used to automate communication between company and server. Chatbots are
computer programs that communicate with its users using natural language. Chatbots are now shifting to
mobile messenger interface, they are platform independent as they use messenger infrastructure and
downloading apps.
The goal is different from what we think when we deal with chatbots and other kinds of digital
helpers like Apple’s Siri or Amazon’s Alexa,where the aim is to get something accomplished like turning
on some music or checking weather. AI gets smarter the more you interact with it. Bots are popping up
everywhere from Facebook to home personal assistants. Advances in natural language processing
,machine learning and other AI technologies created the substance for bots. The AI part of these bots is
JASC: Journal of Applied Science and Computations
Volume VI, Issue I, January/2019
ISSN NO: 1076-5131
Page No:466
limited to natural language processing and basic functionality tied to fixed data. Bots need the ability to
become smarter and more proactive to enable more meaningful communication. For example, they learn
user preferences and behaviour to offer the right services at the right time. Without this, two way
communication can’t happen.
3. Chatbot Creation
3.1. Design
The Chatbot design, it is the process that defines the communication between the user and the Chatbot.
The Chatbot designer will classify the chatbot personality, the questions that will be points to the users,
and the overall communication.
It can also be viewed as a division of the conversational design. To regulate the speed up process, user can
use chat design tools which helps for immediate preview, team collaboration and video export. An
important part of the chatbot design is also centered around user testing. User testing can be performed
following the same principles that guide the user testing of graphical interfaces.
3.1.1 Predetermined links and buttons
These saved users from typing. People appreciated having these options and even expected them for common inputs. These were usually displayed in a carousel and could include images.
3.1.2 Text
Allowed users some flexibility in choosing the types of questions they sought to ask and enable them to diverge from the script of the chatbot.
Changing from Choice based to Conversationbased : Here we you remind the user that they can ask many different type of questions, in many different ways. Now-a-days people are used to clicking things on a digital screen, without any thought. They are used or habituated to not interacting to websites. * It’s not supposed to be a menu. It’s a suggestions screen. This change may look drastic, but this changes user behavior at a fundamental level as we have seen.
JASC: Journal of Applied Science and Computations
Volume VI, Issue I, January/2019
ISSN NO: 1076-5131
Page No:467
Figure 2: home screen designing
3.2 Building The process of building a Chatbot can be divided into two main tasks a) Understanding the user's intent and b) Producing the correct answer.
3.2.1Understanding the user’s intent
The first task involves understanding the user input. In regulate to properly understanding of a user input
in a at no cost text form, a Natural Language processing Engine can be used.
3.2.1.1 Natural language processing (NLP)
It is a subfield of computer science and artificial intelligence worried with the communications between
computers and users (natural) languages, in particular how to program computers to process and analyze
large amounts of natural language data.
Natural language processing systems are based on difficult sets of hand-written rules.Later it was given
by the machine learning systems. It is a static revolution.
The syntaxes are Grammar induction, Lemmatization, Morphological segmentation, Part-of-speech
tagging, parsing, Sentence breaking, Word segmentation, Terminology extraction.
This is even having the speech recognition, given a voice clip of a user or people speaking, resolve the
textual demonstration of the speech. This is the contradictory of text to speech and is one of the
tremendously difficult problems colloquially termed as "AL-Complete". In natural speech there are hardly
many pauses between successive words, and thus speech segmentation is a necessary subtask of speech
recognition. Note that in almost many spoken languages, the sounds representing in a row letters blend
into each other in a process termed co articulation, so the adaptation of the analog signal to discrete
characters would become a very difficult process. Also, given that words in the similar language are
verbal by people with different accent, the speech recognition software must be able to be familiar with
the wide variety of input(user) as being identical to each other in terms of its textual equivalent.
JASC: Journal of Applied Science and Computations
Volume VI, Issue I, January/2019
ISSN NO: 1076-5131
Page No:468
Natural language processing involves breaking down of sentences and other parts of language, into
components. It processes the semantics of content to identify things like the entities and intents of the
user. This has been a hard problem for computer science to tackle, but in recent years it had advanced in
the field and have made this feasible.
Basic steps ofNatural processing:
Sound
Waves words
Figure 3: basic steps of natural processing
3.2.2 .Producing the correct answer
The second task may involve different approaches depending on the type of the response that the chatbot
will generate.
3.3Analytics
The usage of the chatbot can be monitored in order to spot potential flaws or problems. It can also provide
useful insights that can improve the final user experience.Analysing the data effectively is the key to
finding thevaluable insights into crafting growth stratagies, discoverig the new bussiness opportunities,
determinig effective employee recruiment and retaining stratagies.
Analytics is an extension of Application Insights. Application Insights provides service-level and I
nstrumentation data like traffic, latency, and integrations. Analytics provides interaction-level reporting
on user, message, and channel data.
3.3.1 Specify channel
Choose which channels appear in the graphs below. Note that if a bot is not enabled on a channel, there
will be no data from that channel. 3.3.2 Specify time period
Analysis is available for the past 90 days only. Data collection began when Application Insights was
enabled.
Figure 4: specifying the time period
3.3.3 User
The Users graph tracks how many users accessed the bot using each channel during the specified time frame.
The percentage chart shows what percentage of users used each channel.
The line graph indicates how many users were accessing the bot at a certain time.
syntactic processing
Semitic processing
Pragmatic
processing
phonetics
JASC: Journal of Applied Science and Computations
Volume VI, Issue I, January/2019
ISSN NO: 1076-5131
Page No:469
The legend for the line graph indicates which color represents which channel and the includes the total number of users during the specified time period.
3.3.4 Activities
The Activities graph tracks how many activities were sent and received using which channel during the
specified time frame.
Figure 5: activity sheet
The percentage chart shows what percentage of activities were communicated over each channel.
The line graph indicates how many activities were sent and received over the specified time frame.
Analytics are not available until Application Insights has been enabled and configured. Application
Insights will begin collecting data as soon as it is enabled. For example, if Application Insights was
enabled a week ago for a six-month-Old bot, it will have collected one week of data.
Analytics requires both an Azure subscription and application insights resource.To access
Application Insights, open the Bot in the Azure Portal.
3.3.5.Self-service rate
This metric helps you identify the number of users who get what they want from the chatbot without any
human input. For example, if your chatbot goal was to sell a particular product, you will measure the
percentage of user interactions that achieved that goal.
3.3.6 Maintenance
To keep the chatbot up to the speed with varying company products and services. Traditional chatbot
improvement platforms require ongoing maintenance. This can either be in the form of an current service
supplier or for larger enterprise in the appearance of an in-house chatbot instruction team. To eliminate these
costs, some startups are experimenting with artificial Intelligence to develop self-learning chatbot, particularly