A Survey on Various Types of Chatbots

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

A Survey on Various Types of Chatbots

Supreetha H V1

1M. Tech Student, Dept. Of CS&E, RVCE Bengaluru, Karnataka, India ***

Abstract - Chatbots are software applications which are mainly used to perform tasks asked by the users. These are just like virtual assistants. With the increase in conversational medias, chatbot’s application increased drastically. Many social media, e commerce, banking sectors also use these chatbots for easy access to the customers. These chabtots can communicate with users anytime and they won’t require more cost for maintenance. Thus many websites and conversational platforms are choosing chatbots as face of their products. The aim of this paper is to get to know in details about various kinds of chatbots available in market and their applications in real life scenarios.

Key Words: chatbots, conversational media, ML, NLP, virtual assistants

1.INTRODUCTION

Machinelearningandartificialintelligencearehottopicsin recenttimes.Thegrowthinthesedomainshavemadehuge impactinmanyares.Oneofthemajorapplicationsofthese domains can be seen in chatbots. The chatbots just try to imitate human behaviour and tries to help users. The increase in chatbot can decrease work load of many industries.Sayinan e commercewebsite,thechatbotcan interact with user and learn user’s personal interests and suggestsomeoftheproduct.Thiswillmakecustomerhappy andwillalsohelpsellerstoincreasetheirprofit.Someofthe chatbots can also be used in health field also to boost the confidenceoftheusers.Sayifapersonisindepressionand isnotreadytosharehisdifficultieswithfamilyandfriends, hecanconversewithchatbottocomeoutfromthestateof mind. Similarly chatbot’s application can be seen in many areas.

1.1 Natural Language Processing

NLPisthereasonwhichcausedthedevelopmentofchatbots. Withnaturallanguageunderstanding,thechatbotswilltryto understandthecontextoftheusergiveninputandprovide suitableresponsesfortheusers.TheNLUcanbedividedinto further stages namely, lexical analysis, syntax analysis, semanticanalysis,disclosureintegration,pragmaticanalysis. Lexical analysis include splitting of user given inputs to paragraphs,sentencesandwords.Thesyntaxanalysischecks for logical meaning of the user given inputs. Checks if all words are correctly placed in order or not. “The good is flower” is rejected by English syntactic analyzer. The sentence is not placed in meaningful way hence it will be rejected.Itwilltakecareofgrammaticalpart.Thesemantic

analysis will try to understand context and tries to understandifthereisanymeaningforuserssentence. Sayif a userasksifthemilkisblack?Thebotshouldrejectsuch logic lessquestionsasitdon’thaveanymeaning.Disclosure integrationstagemakeschatbotmoreconversationalenough forusers.Thiswilltracktheprogressofchatandtriestolink currentandpreviousstatestoprovidemeaningfulresponses for users. The pragmatic analysis do real time analysis for understandingqueriesaskedbytheusers.Thediagrammatic representationofstagesinNLUisgiveninFigure1.

Figure 1. Stages in NLU

Theorganizationofthereportisasfollowsthesection2will discuss about the various kinds of chatbots and its applicationsandsection3willgiveconclusionfollowedby referencesection.

2. TYPES OF CHATBOTS

Chatbotscanbebroadlyclassifiedintotwotypes,theyare: AIchatbotsandRulebasedchatbots.AIchatbotsarerecent developmentswhichareusingintelligentsystemsforgiving outputstouses.Itrequiresmoretrainingforthesechatbots tobehavelikehumans.Sometimesthesebotsaretrainedin such away that it would become much difficult even for userstounderstandwhetheraotherendifitisuserorabot they are conversing with. Rule based bots don’t require much training, they will be having set of questions and answerstheycaninteractonlyforthosesetofqueries.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Thissectiongivesindetaileddescriptionaboutvariouskinds of chatbots with its applications. Menu based chatbots, Linguisticchatbots,MLchatbots,Hybridchatbots,keyword recognitionchatbotsandVoicebots.Thediagrammaticview ofthisisshowninFigure2.

2.1 Menu Based Chatbots

These chatbots are initially trained with yes or no option menus. The next question asked by the bot will be linked withtheanswergiveninthepresentstage.Thesebotswill bepre fedwithlistofquestionnaireswithitsanswerpath for two options also. The answer chosen by user will determinenextsetofquestions.Mainlydecisiontreemodels areusedtotrainthistypeofchatbots.

tools to develop these chatbots. Some of the very famous platformswhereonecandevelopthesekindsofchatbotsare DiaglogFlow,RASA,gupshupinterface,BotKitetc.Thesewill behavingtemplateforchatbotdevelopmentwhereonecan create bots according to their needs. They also provide integrationwithothermessagingplatformslikewhatsapp, facebook, telegram etc to deploy build chatbots. One can selectanyMLmodelsoftheirinteresttodevelopchatbots here.SomeofthemodelsthatcanbeusedareRNN,Neural networkmodes,DIETclassifiers,CNNmodelsetc.According to users comfort they can choose any model to train the chatbots.

2.4 Voice Bots

Themainadvantageandgrowthofchatbotwasseenonce voicebasedchatbotweredeveloped.Becauselargenumber ofaudiencedon’twishtotypeorsomepeopledon’tknow how to type. With the voice bots, all these challenges are solved. Thischatbotwillbehavinganadditionalfeatureof text tospeechorspeech to textconversions.ManyAPI’swill help to enable this feature. The main usecase of this chatbotscanbefoundforspeciallyvisualizedpersonsalso. Theycangiveinputtothebotsbyvoiceandthechatbotwill respondbackwithvoiceoutputs.Thustheycanbusedeven as assistants by specially visualized persons. Some of the additionalfunctionalitieslikesendinge mail,voicemessages canalsobepossibleusingthesebots.

2.5 Keyword Recognition based Chatbots

Figure 2. Types of chatbots

2.2 Linguistic Chatbots

Many linguistic chatbots can be seen now a days. A wide range of applications are based on this type of chatbots. Manysocialmediaplatformslikefacebook,messengerwill be having a number of linguistic chatbots. These bots will attractlargeaudiencebecausemanyuserswillnotbeaware ofEnglishlanguage.Addinglinguisticchatbotswillattract morecustomersfora given websiteandfinallyincreasing the profit of the organizations. Many times these bots are created for language learning and language translation purpose also. These will help to learn new languages by givingtranslationsinuseraskedlanguages.Inaplacewhere wedon’tknownativelanguage,Iwouldbebeneficiary

2.3 ML chatbots

TheMLchatbotswillbeproperlytrainedbots.Thesewillbe trained for intent entity selection and response giving methodologies. Many platforms will provide open source

Thesearerequirementspecificchatbots.Sayifauserasks “book a movie ticket”, the chatbot will try to retrieve keywordssuchasbookandmovieticket.Itwillfirstcollect thekeywordsfromthequeriesandthenprocessit.Ifabot can’tunderstandthekeywordthenproperoutputcan’tbe givenandsuddenstopornoresponsefromthebotcouldbe observed.Henceinthesecasespropertrainingofchatbots areverymuchessential.Morethetraininglesserwillbenon reorganizationofkeywords.

2.6 Hybrid Model Chatbots

These bots will be having both features of menu drive chatbots and AI enabled chatbots. These are growing chatbotsnow a days.Inrealtimescenariosmaypeoplewill belazyenoughtotypeandanswerallthequeriesaskedby theusers,Soifinanotherflow,asetofquestionswithlistof optionsareprovided,userscanswitchthewayofanswering accordingtotheirconvenience.Manyofthewelldeveloped e commerce websites will use this hybrid bots for their websites.Theywillprovidesetofproductsasmenudriven and sometimes they will also converse contextually with userstoengagethemmorewiththeirproducts.Hencethis botisamilestoneinchatbothistory.

Table1willgivecomparisonoftypesofchatbots.

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal |

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

Table 1 Comparison of types of chatbots

Chatbot AI/ Rule Based? Time for

Response giving and Training Advantages Limitations

Menu Based Chatbots

RuleBased  LessTraining.  LessResponse time.

Linguistic Chatbots AIBased  MoreTraining.  MoreResponse time.

 Quicklyresponsewill beobtained.  Lessambiguityas pathsarepredefined.

 Helpfulforlanguage learning.  Canbeintegrated withmultiple messagingplatforms

 Limitedusage.  Only used for Yes/No type of questions.

 Moretrainingis required.  Sometimesmore trainingmay leadto overfitting.

ML Chatbots AIBased  MoreTraining  MoreResponse Time

Voice Bots AIBased  MoreTraining  MoreResponse Time

Keyword Recognition Based Chatbots

Hybrid Chatbots AI&RuleBased

Response Time

MoreTraining  MoreResponse Time

3. APPLICATIONS OF CHATBOTS

Theapplicationsofchatbotscanbeseeninmanyareas.Some oftheareaofapplicationsofchatbotsarediscussedinthis section. The diagrammatic representation of chatbot’s applicationisshowninFigure3.

Conversational bots justlikehumans.  Contextpreserved.



 Helpful for specially visualizedgroup.  Notypingisrequired.

 Moretrainingis neededtoget accurateresults.  Asitneeds processing responsetimeis alsomoreuser needstowaitto getresponse.

 Extrastepof text to speech orspeech to text.  Difficultto understandthe accentofusers correctly.

Easy for pre trained data  Fastresponse.



 Interactive  Gives personalized outputs

 Moretraining.  Limitedsetof queriescanbe answered.

 Much time for training.  Complex architecture.

3.1 CUSTOMER SUPPORT

Wheneverauserishavinganyissuesordoubtsheusually callstotollfreenumbersofcustomercare.Theyneedtowait forlongperiodtogettheirturn.Itwouldbetediousevenfor usersandalsoforthpersonwhoneedstoanswerthesecalls.

© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page690

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

If achatbotisgiventrainingforansweringthesequestionsit would be easier. Much of time is not wasted on repetitive tasks. Moreover chatbots wont get tired and they can be madeavailable24/7withlessmaintenancecost.

3.2 SOCIAL MEDIA

Thesebotswillbeincorporatedinsocialmediaforengaging people.Sometimesthechatbotscanbeharmfulalsoinsome socialmediaplatforms.SayinInstagrampeoplemaycreate fakebotstoincreasepopularityofaperson.Theymayfake likeandcommentusingchatbots.Similarcaseswerefoundin Twitter also where people will buy bots to change the mindsetofpeople.Somechatbotsarealsousedincreating fakeprofilesjusttogivefakefollowernumbersforusersin manysocialmediaplatforms.

3.5 SKILL ENHANCEMENT

Many chatbots will help in skill enhancements by giving personalized recommendation systems for users’ interest. They will collect users interests and give more training regardingskillssuchaspostermaking,digitaldrawingsand videoeditingandmakings.

3.6 BANKING SECTORS

In recent days all banking sectors have moved to online platforms. With this advancement they have also included chatbots in their website to guide users of all the benefits theyoffer.Thebotswillalsosuggestthetypeofaccountusers canopen,theinterestsindifferentaccounts,interestsoffered bythebankandloanfacilities.Itwillhelpusersastheyneed notwaitinlongqueuesinsidebanktogetallthedetails.In onlineonlyuserscanunderstandallthefacilitiesofbanking sectors.Itwillreducetimeofusersalsoandreduceburden onbankemployees.

4. CHALLENGES OF CHATBOTS

There are some challenges wile implementing of the chatbots.Someofthemare:

4.1 SECURITY

AIchatbotsneedstocollectinformationanddatawhich areneededtotransmitoverInternet.Thebotsmustmake surethattheconfidentialinformationabouttheusersare notsharedwithanyoneandhelp tomaintainprivacy of users.

4.2 UNDERSTANDING EMOTIONS

Figure 3. Applications of chatbots

3.3 TRAVEL AENCIES

Manytravelagenciesusechatbotsintheirwebsites.These botswillhelpinfindingcustomizedhotels,restaurantsalso. The bots will take care of finding the tickets in different modeslikeplanes,trainsandalsohelptogetbookingingood hotels and restaurants. This will make bookings easy for usersandincreasetheprofitsoftheagencies.

3.4 ASSITANT TO PERSONS

Thebotscanalsobeusedaspersonalassistants.Someofthe famouspersonalassistantsareSiri,Alexa,google’sassistant. Thesecanbeusedforentertainmentpurposealso.Theyhelp toengageusersbygivingsomeactivitiesorrecommending somepersonalizedmusic,comedies,newsetc.Theycanalso beusedtotracktheday to dayactivitiesofusersandcanalso beusedtogetalertreminders.

Theremaybecaseswhereusermygiveabbreviationsas inputs,andsometimesbotneedstoconsiderredundant words,andnegativewordstounderstandexactcontextof therequestaskedbytheusers.Propertrainingofchatbot isneededtoovercomethesechallenges.

5. CONCLUSIONS

With growth in technology many of the platforms have includedchatbotsinsidetheirapplications.Alargenumberof chatbotsarealreadydevelopedinhealth,e commerce,social media,customerserviceandeducationalsectors.Themain aimofthispaperistounderstandvarioustypesofchatbots andtheareasinwhichtheycanbeusedfor.Fromtheanalysis it could be seen that almost every field is incorporating chatbotsbasedontheirrequirements.Thusincomingdays oncecansee major increasein numberofchatbotsandits applications,Oneneedtounderstandbothitsdrawbacksand applicationsandtrytousethischatbotasinfavourtohelp others.

REFERENCES

[1] M. M. Khan, "Development of An e commerce Sales Chatbot,"2020IEEE17thInternationalConferenceon SmartCommunities:ImprovingQualityofLifeUsingICT, IoT and AI (HONET), 2020, pp. 173 176, doi: 10.1109/HONET50430.2020.9322667.

[2] S. Thorpe and H. Scarlett, "Towards a Cyber Aware ChatbotService,"2021IEEEInternationalConferenceon Big Data (Big Data), 2021, pp. 6040 6042, doi: 10.1109/BigData52589.2021.9671775.

[3] W.Mahanan,J.Thanyaphongphat,S.SawadsitangandS. Sangamuang, "College Agent: The Machine Learning Chatbot for College Tasks," 2022 Joint International ConferenceonDigitalArts,MediaandTechnologywith ECTI Northern Section Conference on Electrical, Electronics, Computer and Telecommunications Engineering(ECTIDAMT&NCON),2022,pp.329 332, doi:10.1109/ECTIDAMTNCON53731.2022.9720420.

[4] P. Srivastava and N. Singh, "Automatized Medical Chatbot(Medibot),"2020InternationalConferenceon Power Electronics & IoT Applications in Renewable EnergyanditsControl(PARC),2020,pp.351 354,doi: 10.1109/PARC49193.2020.236624.

[5] N.Kanodia,K.AhmedandY.Miao,"QuestionAnswering ModelBasedConversationalChatbotusingBERTModel and Google Dialogflow," 2021 31st International Telecommunication Networks and Applications Conference (ITNAC), 2021, pp. 19 22, doi: 10.1109/ITNAC53136.2021.9652153.

[6] J. Doshi, "Chatbot User Interface for Customer Relationship Management using NLP models," 2021 International Conferenceon Artificial Intelligenceand Machine Vision (AIMV), 2021, pp. 1 4, doi: 10.1109/AIMV53313.2021.9670914.

[7] P. Voege and A. Ouda, "A Study on Natural Language Chatbot based Authentication Systems," 2021 InternationalSymposiumonNetworks,Computersand Communications (ISNCC), 2021, pp. 1 4, doi: 10.1109/ISNCC52172.2021.9615767.

[8] R. Garg et al., "NLP Based Chatbot for Multiple Restaurants," 2021 10th International Conference on System Modeling & Advancement in Research Trends (SMART), 2021, pp. 439 443, doi: 10.1109/SMART52563.2021.9676218.

[9] D.P.P.VillanuevaandI.Aguilar Alonso,"AChatbotasa SupportSystemforEducationalInstitutions,"202162nd International Scientific Conference on Information TechnologyandManagementScienceofRigaTechnical

University (ITMS), 2021, pp. 1 6, doi: 10.1109/ITMS52826.2021.9615271.

[10] H.Beattie,L.Watkins,W.H.Robinson,A.RubinandS. Watkins, "Measuring and Mitigating Bias in AI Chatbots," 2022 IEEE International Conference on Assured Autonomy (ICAA), 2022, pp. 117 123, doi: 10.1109/ICAA52185.2022.00023.

[11] C.J.LuoandD.E.Gonda,"CodeFreeBot:Aneasywayto jumpstart your chatbot!," 2019 IEEE International ConferenceonEngineering,TechnologyandEducation (TALE), 2019, pp. 1 3, doi: 10.1109/TALE48000.2019.9226016.

[12] S. Meshram, N. Naik, M. VR, T. More and S. Kharche, "Conversational AI: Chatbots," 2021 International ConferenceonIntelligentTechnologies(CONIT),2021, pp.1 6,doi:10.1109/CONIT51480.2021.9498508.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page692