Intelligent Library Assistant (ILA) Using Artificial Intelligence and Natural Language Processing

International

of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

e ISSN:2395 0056

p ISSN:2395 0072

Intelligent Library Assistant (ILA) Using Artificial Intelligence and Natural Language Processing

1,2,3B.E. Student, Department of Computer Engineering, 4Project Guide, Smt. Indira Gandhi College of Engineering, Navi Mumbai, Maharashtra, India ***

Abstract With the outbreak of the recent pandemic coronavirus (formally known as COVID 19 or 2019nCov), we are embracing the "new normal" and using and relying more on virtual space than physical space ever before. In the library context, moving the user assistance program from physical to virtual causes problems with this migration. This project is based on a possible solution using a practical implementation of conversational AI. a "bot" that meets the needs of users 24 hours a day, 7 days a week without human intervention. A bot is a computer program that can simulate conversations and interactwithhumans(spoken,written,orboth).Itgetsasetofpre programmedcommandsandcontinueslearningbasedon theinputitreceives.ThisprojectgivesanoverviewofconversationalAI,bots,andtheirmultitaskingcapabilities,aswellasa practical overview of their implementation. In addition, the project described the requirements of the current situation and the long term benefits of bots in the library. Building customer support is one of the most important workflows for any business in the world. Technological growth is driving customer expectations, including the need for 24/7 service customer support, quick response, and answers to simple questions that require vast amounts of human resources and knowledge. Chatbots are solutions based on adapting AI technology to redesign customer service spaces. Imagine for a moment that chatbotshelpservecustomersratherthanhumansprovidingcustomersupport.Inthisproject,achatbotcalled"ILA"isbuilt to explore library resources. This makes it easy to access the university library and find books available to students and facultywithouthavingtogotothelibraryinperson.Studentscanexplorelibraryresourceswiththehelpofthischatbot. This allowsstudentstoconnecttothelibraryanytime,anywhere.ThechatbotisbuiltinPythonandintegratedwithFlask,sousers canaccessthechatbotfromtheirmobilephones,laptops,andotherhandhelddevices.

1. INTRODUCTION

Librariesactwiththesocialmindsetofprovidingvaluableinformationresourcestothosewhocannotaffordahugecollection ofbooksforprofessionalsupport.Thisisoneoftheuniversity'smostimportantassets.Therefore,the universityisfocusedon maintaining and managing the details of library data as efficiently as possible. The value of a source can be judged not by money, but by its knowledge and usefulness. Therefore, institutions need to ensure that teachers and students receive the supporttheyneedtoeffectivelyuselibraryresources.Themainobjectivesofthisprojectare,toexploretheLibraryresources to the users by adapting an Artificial Intelligence Technology to easy communication between the user and the chatbot anytime and anywhere. In recent times, chatbots have reduced the need for human intervention. Artificial intelligence and machine learning have accelerated the evolution of chatbots to blur the line between human and bot styles of interaction. Chatbotsaregenerallyusedinsystemsthatinvolvechattingwithuserstoeitheracquireinformationorprovidetheserviceas information to them. Although certain chatbots use natural language processing, many chatbots store their responses in a database and retrieve them when they receive input from the user. Advanced AI technology creates a chatbot self learning system.Chatbotscanlearntorespondtoeachconversationwithoutascript.Chabothelpspeoplebyreducingthetimespent navigatingthewebsimplybyprovidingtheinformationtheyarelookingfor.

2. RELATED WORK

Nowadays, many applications require chatbot intervention rather than human interaction with the user requesting the service. For example, The university management system model chatbot is a simple system with pre programmed data. The methodsusedarepatternmatching,naturallanguageprocessing,anddatamining.Thespeakeroruserinputtextismatched bytheChabotwiththetemplatethatexistsinthedatabase.Anotherweb basedCollegeInquirychatbot canmarkaresponse from a chatbot as invalid if the response is not related to a query. In addition, some bots are trained with applications that respondtomultipleanalyzedrequestsfromusers.Education,business,andsocialmediaaresomeoftheareas wherechatbots

Factor value:

International

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

Technology (IRJET)

e ISSN:2395 0056

p ISSN:2395 0072

areused.Itcanbeusedasalearningtool.Currently,chatbotlanguagesupportislimited.Therefore,thereisplentyofroomfor futurechatbotstoremovesuchlanguagebarriers.

3. LITERATURE SURVEY

ArtificialIntelligenceChatbotsarefollowedinlotsofsectorsassociatedwithconsumerrelationships.Astudiescrewviaway of means of Liao et al. (2016) designed information enriched multimodal style chatbot that assists clients in looking for merchandiseandmatching extraordinarystyles. They mentioned the machineand offeredtheirstrategiesand stories whilst highlightingthedemandingsituationswithinsidethefield.Ko&Lin(2018)havedeliveredacardBot,achatbotforenterprise cardpopularityandcontrolinwhichanOCRmodulechangedintoapplied.Inthefitnesssector,Madhuetal.(2017)proposed a concept for growing AI chatbots for scientific treatments. This unique form of chatbot assists human beings in taking important medicine and treatment. Park and Jeong (2019) have delivered a brand new conversation chatbot that interacts with the clients nearby. In keeping with their belief, it additionally has the capability for different fields. Villegas Ch et al. (2020) mentioned the AI version chatbot responses to the desires of college students inside a clever campus. Artificial Intelligence(AI)canconvertthejobsandcapabilitiesoflibrariestoservethebrandneweraoflibrarycustomers.Guptaetal. (2020) articulated the software and capability effect of synthetic intelligence in instructional libraries. They diagnosed 4 domains:educational,informative,assistive,andsocialnetworkingforsyntheticintelligenceapplications.Accordingtothem, the libraries can undertake AI for numerous functions such as reference offerings, Journal of Management Information and DecisionSciencesVolume23,SpecialIssue,20204451532 5806 23 S1 213CitationInformation:Nawaz,N.,&Saldeen,M.A. (2020). Artificial intelligence chatbots for library reference offerings. Journal of Management Information and Decision Sciences, 23(S1), 442 449. and highlighted that the very last intention of the chatbots is streamlining the capabilities of the referenceofferingsunit(Vincze,2017).MckieandNarayan(2019)intheirexploratorypaperemphasizedthesignificanceof attractivelibrariansingrowingchatbotsinconjunctionwiththecollaborationofgenerationbuilderstofulfilltherequirement of conducive gaining knowledge of the environment. Cox et al. (2019) defined Artificial Intelligence changed into taken into consideration to be one of the outstanding regions of attention that have to acquire interest from all sectors. Tubachi and Tubachi(2017)mentionedthattheprimaryrecordsassociatedwithlibraryofferingsandcenterscanbeaddedthrueasychat, prolonged chat, or video conferencing, email, FAQs, guided tours, and asking a librarian, internet forms, and chatbots. The chatbotscanplaya powerful position toattain thecustomers 24/7.Astheymentionedthat digital referencesaretakeninto considerationasanessentialdeviceinlibraries.Chatbotintegrationinsidealibraryinternetsiteisstraightforwardand cost powerful for libraries to make bigger their records offerings. Ali (2019) has shared the stories of growing “evidence of concept” and “AskSmooSmoo`' chatbots and he mentioned 3 regions of concern: consumer experience, collaboration, and skillscontrolforpowerfulofferings.Allison(2012)statedinhisresearchpaperthatmostuserrequestsaredirectedorfactual requests. McNeal and Newyear (2013) explained that libraries can build bots using the available coding options. AIML (ProgramZorProgram

O)andchatscriptsareconsideredthebestoptions forthispurpose."Virtualreferences"areconsideredanimportanttoolfor improving information services. Chatbots can be easily and inexpensively integrated into websites to improve information services. A systematic review of the literature review found that the number of research papers in this direction was very smallinthisstudy.Therefore,thepurposeofthisstudywastocarryoutthestudytofillthegap.Inaddition,theauthorsstated that few studies were conducted in connection with Bahrain. Therefore, researchers explored the benefits of AI chatbots in Bahrain'sacademiclibraryreferenceservices.

4. METHODOLOGY

4.1. NLP Techniques and Research

ThethreeimportantNLPviewpointsconcerningchatbotsandotherdialoguesystemsare:

1.NaturalLanguageUnderstanding(NLU),

2.NaturalLanguageGeneration(NLG),

3.DialogueProcessing.

International Research Journal of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

4.1.1. Natural Language Understanding (NLU)

e ISSN:2395 0056

p ISSN:2395 0072

NLUinvolvestheprocessofextractingmeaningsfromtextinputs.Inthisdirection,thebasicdesignstepsinclude:

•Syntactic Parsing Determines the function of each word (part of speech), how the words interact, how they are grouped into phrases, and how they change words. Context free 2 grammar (CFG) definition and parser implementation is the usual NLPtechniquesusedinthisstep.

•SemanticParsing Thepositionofasemanticparseristoextractthecontext unbiasedthatmeansofawrittensentence.The discriminative techniques together with support vector machines (SVM) and statistical techniques together with decision treesandclassificationandregressiontrees(CART)areusedtodiscoverthemaximuminalllikelihoodparsetreethatsuitsthe sentence.

•Contextual Interpretation Use discourse level information to refine your semantic interpretation by removing the remaining ambiguities such as anaphora, pronouns, and abbreviations. The list of discourse entities (DEs) contains a set of constantsthatreferencetheobjectcalledinthepreviousstatement.Theseobjectscanbeimplicitlyreferencedlater

4.1.2. Natural Language Generation (NLG)

NLGinvolvescreatingachatbotresponsebasedonwhatisdoneintheNLUstage.

•ArtificialIntelligenceMarkupLanguage(AIML) an XML compliantlanguagefordevelopingAIflowsindialogsystems.Its purposeistosimplifythetaskofconversationmodeling.

TheimportantelementsofAIMLare:

•Categories Abasicknowledgeunitconsistingofpatternsandtemplates.

• Recursive Usedtocollateothercategoriesrecursively.Thissimplifiescomplexgrammaticalforms,expressedasatag.

• Context Categorytagsusecontexttagstorefertotheuser'spreviousinput.

• Variables Usedtosupporttheretrievalandsettingofcommonlyusedtextwithtagscommonlyusedtostorenouns.

• PronounSwapping Usedtoreplacepronounssuchas"you"with"I"and"you"with"my"andsoon.

• Document Scheduling / Response Generation Breaks down high level communication goals into a structured representationofatomiccommunicationgoals.

•Microplanning Aphaseinwhichthenumberofphrasesgeneratedisdeterminedtoproduceamorenaturalvoice.Methods suchassemanticgrammarandinverseparsingareusedtogenerateunnaturalproto phrases.

• Surface realization The process of transforming the abstract structure obtained in the microplanning phase into a linguisticsurfacestructurebyaddingfunctionwords,conjugatingwords,determiningwordorder,andsoon.

•ChatScript TechniquestohelpifthereisnomatchAIML.Itfocusesonthebestsyntaxforcreatingavalidstandardanswer.

•Markov chain It is more probabilistically applicable and is used to build more accurate answers as a result. The idea of Markovchainsisthateachletterorwordinthesametextrecordhasafixedprobabilityofoccurring.

•Languagetricksincludesentences,phrases,andparagraphsavailableinchatbots,makingyourknowledgebasemorediverse andcompelling.Thetypesofvoicetricksarestereotypedresponses,typos,simulatedkeystrokes, personalstorymodels,and Non Sequitur (no logical reasoning). Each of these language tricks is used to serve a specific purpose and provide an alternativeanswertoaquestion.

• Ontologies/Semantic networks Consists of several relational and hierarchically related concepts. The purpose of using ontologyinchatbotsistocalculaterelationshipswithsynonyms,hyponyms,andotherrelationshipsthatarenaturallanguage

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

3765

International Research Journal of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

e ISSN:2395 0056

p ISSN:2395 0072

term names. The relationships between these concepts can be represented in graphs that allow computers to search using specificinferencerules.

4.1.3. Dialogue processing

The purpose is to build a man machine dialog based on the switching process. In this process, information is sent from one participant (the user and another) to the dialog manager (DM). The dialog management process involves implementing a dialoguestrategytoorganizea sequence of system dialogsto achieve the common goalsofusers and chatbots. Chatbot user satisfactionisstronglyinfluencedbytheconceptofdegreeofinitiative.

•System Led A system controlled initiative that asks users accurate questions and expects information or answers from them.

•UserLed User ledinitiative.Thesystemmustprovideinformationabouttheuser'srequestwithoutrequestingdetails.

•Mixed Initiative Aninitiativeinwhichbothparticipantsworktogethertoachieveconversiongoals.Forexample,chatbots share control between users and the system. With the relational database, chatbots can remember past conversations and build a knowledge base to make them more continuous and meaningful. You can use Structured Query Language (SQL) to generate queries and nest query blocks to store conversation history. This makes it especially easy to find word phrase matches.

5. DESIGN AND IMPLEMENTATION

TheLibraryChatbotisbasedonintelligencethatanalyzesuserrequestsandrespondsaccordingly.Thishelpstomakebooks available to students and faculties of different grades and faculties without having to go to the library. Flask is used as a platform/toolforbuildinguserqueriesandresponses.ItisintegratedwiththeFlask basedwebandtheresponseisdisplayed in the GUI. In the dialog flow, the user query is analyzed and the chatbot gets the response and responds to the user. The systemrespondswithaneffectivechatinterface,whichmeansthattherealpersonistalkingtotheuser.Withthehelpofthis bot,userscanrequestinformationaboutlibrary relatedactivitiesonline.ILAhelpsusersbysavingtimeandprovidingdataon libraryresourcesandresponsestosimplequeries.Usersdonothavetogodirectlytothelibrarytoensurebookavailability.

5.1. Agent and Intent creation

Thismoduleconsistsof1)Agentcreation&2)Intentcreation.

• Agent Creation: Agents are platform agnostic. Agent processes the request of the users which are in the form of natural language.Ittransformsauser’snaturallanguagerequestintoanactionablequeryandisusedtomanageconversationflowin aspecificway.ItmustbedesignedbyanagentonceandthenitcanbeintegratedwithavarietyofplatformsusingSDKsand integrations,ordownloadfilescompatiblewithFacebookMessenger,Slack,etc.

•Intentcreation:Itepitomizesthemappingbetweenwhatauserrequestsandwhatresponseshouldbetaken.Themaintask is to train the phrase by adding user expressions using multiple informal contexts, short formats, and so on. Actions and parameters allow you to match entities by specifying parameter names, entities, prompts, and values. The text response is rendered by the bot as output that can be rendered in two formats. The default response of the dialog flow and the conversationisperformedviatheFlaskGUI.Thedefaultreplyissomekindofplaintext.

5.2. Training the Bot and Database creation

Inthismodule,you'llcreateanentitythatmakesiteasytoreferenceyourintentandtrainyourbottolearnfromthatmistake. EachconversationinitiatedbyausernotspecifiedintheintentissavedintheUserSpeakscolumnofthetrainingblock.Select Click to Assign to approve the informal context used by the user. Finally, the informal context is approved and the bot is trained.Tostoreallthedetailsofabookintheuniversitylibrary,youneedtocreateadatabasetablewithrowsandcolumns. .CSV (Comma separated Values) or a data.json files are used to store and retrieve information such as book id, book name, authorname,subjectname,bookpublishing,bookpositionontheshelf,andmakebooksavailableinthelibrary.

2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008

3766

International

of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

FigNo.1.Architectureoftheproposedsystem

e ISSN:2395 0056

p ISSN:2395 0072

FigNo.2.Design

Theprocedurethatwasimplementedfortheworkingofthischatbotsystemisasfollows:

Step1:Start.

Step2:Trainthephrasesbyaddingtheuserexpressionintheintentse.g.Forauserexpression“ListthesubjectsforCSE first year”removalofstopwordswillbedonebyEntitymappinginthedata.jsonfile.

Step3:Addtherespectiveresponsesforthephrasestrainedduringtheintentcreation.

Step4:Gettheuserquery(INPUT).

Step5:InthequeryRemovestopwordsandfetchthekeywords.

©

IRJET

Impact Factor value: 7.529 | ISO 9001:2008

International

Journal of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

e ISSN:2395 0056

p ISSN:2395 0072

Step6:Matchthefetchedkeywordswiththekeywordsinintent,ifmatchedgotostep8,thebotwillprovideanappropriate response.Akeywordmatchingalgorithmwillbeappliedtoidentifythematching.Otherwise,gotostep7.

Step7:Ifnomatchedkeywordthenthenewphraseisaddedtothetrainingandmappedtoaspecificintent.

Step8:Returntheresponseasanoutputtotheuser.

Step9:Exit.

FigNo.3.ILAGUI

5.3. Assessment Results

•Scalability: Multiple users can access the bot at the same time. So that many users can talk to chatbots, and it's scalable. Integration using a Flask based GUI makes it very easy for the general public, including students and staff, to access the library.BotscanlinktomultiplesocialnetworkingsitessuchasTwitter,Slack,popularwebsites,anduniversitywebportals.

•Accessibility:Itwillbemucheasiertointegratewiththeuniversitywebportaltomakeitmorescalableforuseinuniversity bots.Eachstudentcanaccessusingausernameandpassword.

Usability:Wehavecreatedabottogetover100booksfromtheCSEdepartment.Youcaneasilyretrievemultipledepartment ledgerswithoutmakinganychangestotheprocess.Furtherprogresswillbe madesoonsothatthelibrarycan existwithout human resources, and simpler chatbots with more rule based development will dominate. In conclusion, chatbots are emerginginmarketing,largehospitals,andinstitutions,andmayevenreplacetheBPOindustryinthefuture.Thecombination ofchatbotsand robots automatesthe processindifferentareaswithfew staff or specific areaswith no employees. Thisalso minimizesoperatingcosts.

6. CONCLUSION

We have created a library chatbot for students, faculty, staff, and more. Library bots save time and replace human human interactions with human bot interactions. Chatbots are still in their infancy, but they are growing rapidly. The main goal of creatingalibrarychatbotistoallowend userstotalktothebotwithoutconsideringtask orientedscenariosandtosucceedin theproject.

7. FUTURE SCOPE

Theremaycomeatimeinthefuture,whenwecanreducetheamountoftimewestaffthereferencedesk,freeinglibrariansfor morecomplexdutiesthatrequiretheskillsthatareuniquetohumans.Itisdoubtfulthatmanyreferencelibrarianswillcheer

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

3768

International Research Journal of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

e ISSN:2395 0056

p ISSN:2395 0072

that day; librarians like most people are reluctant to see technology take over even a small part of their jobs. Blacksmiths mocked thefirstautomobiles,but weknowhowthat ended. Should referencelibrariansbeafraid ofchatbotslikeassistants, etc?Idoubtthatitmakesadifference;technologyconsumesjobsandtheraceisontobuildthebetterbot.

ACKNOWLEDGEMENT

We express our deepest gratitude towards Prof. K.T Patil (Computer Department, Smt.Indira Gandhi Collegeof Engineering, University of Mumbai) for his valuable guidance, moral support, and devotion bestowed on us throughout our work. If I can sayinwordsImustattheoutsetmyintimacyforreceiptofaffectionatecaretoSmt.IndiraGandhiCollegeofEngineeringfor providingsuchastimulatingatmosphereandwonderfulworkenvironment.

REFERENCES

[1] Turing,AlanM."ComputingMachineryandIntelligence."CreativeComputing6.1(1980):44 53.

[2] Dataset collection and information about the Cornell movie dialog corpus dataset are available at https://www.cs.cornell.edu/cristian/CornellMovieDialogsCorpus.htm

[3] NeuralMachineTranslationbyJointlyLearningtoAlignandTranslateDzmitryBahdanau,KyunghyunCho,YoshuaBengio (Submittedon1Sep2014(v1),lastrevised19May2016(thisversion,v7))

[4] Sequence to Sequence Chatbot build using TensorFlow update June 28, 201628,2016[Online], Available: http://complx.me/2016 06 28 easyseq2seq/

[5] AComparativeStudyonDecision MakingCapabilityBetweenHumanandArtificialIntelligence,SohamBanerjee,Pradeep KumarSingh,andJayaBajpai,Last

[6] Anjana Tiha(April 26, 2019), Intelligent Chatbot using Deep Learning,[Paper] Available: https://www.researchgate.net/publication/328582617_Intelligent_Chatbot_using_Deep_Learning

[7] User Adaptive Chatbot for Mitigating Depression, Pratik Kataria1, Kiran Rode2, Akshay Jain3, Prachi Dwivedi4, and SukhadaBhandarkar.DepartmentofComputerEngineering,MIT COEPuneIndia.

[8] Editorial Team “5 Enterprise Challenges Best Solved by Chatbots”, inside BIGDATA, 2 April 2018, https://insidebigdata.com/2018/04/02/5 enterprise challenges best solved chatbots/

[9] Takyar, Akash “EVERYTHING AROUND AI CHATBOTS CHALLENGES AND OPPORTUNITIES”, Leeway Hertz, https://www.leewayhertz.com/ai chatbots/ BIOGRAPHIES

NEERAJRAPELLI BEComputerEngineering,SIGCE

BHUSHANRAUT BEComputerEngineering,SIGCE

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

ARUNAROKADE BEComputerEngineering,SIGCE

Dr.KISHORTPATIL HODofComputerEngineering, SIGCE

3769