MathorCup优秀论文B题.docx

MathorCup优秀论文B题.docx

《MathorCup优秀论文B题.docx》由会员分享，可在线阅读，更多相关《MathorCup优秀论文B题.docx（20页珍藏版）》请在冰豆网上搜索。

MathorCup优秀论文B题

Thejudgesscoring,note

Teamnumber：

20017

Thejudgesscoring,note

Thejudgesscoring,note

`

Problem：

B

Thejudgesscoring,note

Title：

BooksRecommendation

Abstract

WiththedevelopmentofinformationtechnologyandtheInternet,weareenteringgraduallyfromalackofinformationeraintotheeraofinformationoverload.Bothinformationconsumersandinformationproducershaveencounteredalotofchallenges,forwhichrecommendationisanimportanttooltoresolvethiscontradiction,soitiswidelyusedintheproductsandapplicationsintheInternet.

QuestionOne:

Thedataintheannexuser_book_score.txtreflectsthatthebookIDanduserIDaffectbookscores.ThedatainAnnexbook_tag.txtreflectsthatbooklabelsindicatestypesofbooksandaccordingtothebookID,wecanfindthecorrespondinglabel,whichaffectsuserratingsofbooks.InAnnexuser_read_history.txt,thehistoryusersviewedbooksreflectsthekindofbookstheuserslike,thusaffectingthescore.InAnnexuser_social.txt,concernuserstotheirfriendsreflectstherelationshipbetweentheuser'schoiceandlikingofbooks.Thuscontactbetweentheuserandthebooksaredrawn,whichaffectsthescore.Sothefivefactorsimpactingtheuserratingforthebooksare:

bookID,booklabels,bookhistoryusersbrowseing,usersconcernedfriendsandtheuserID.

QuestionTwo:

theuserscoreofbooksintheannexpredict.txtconsidersthefactorsinuserscoreofbooks.Firstofall,dataareinkmotestbyapplicationofsoftwarespss,andthetestsig<0.05,soprincipalcomponentanalysisofthedataisdone.Inordertoensurethereliabilityoftheresults,dataforbooksIDanduserIDarestandardized,andfivemainindicatorsareincomponentanalysis.Whenthreemaincomponentsareacquired,thecumulativecontributionratereaches81.371%.Thesethreemaincomponentsareusedtodofactoranalysis,andintegratedmodelforprincipalcomponentis:

.Predictionforuserratingsforbooksinpredict.txtisultimatelygot.

QuestionThree:

throughthebookhistorytheuserbrowsingandanalysisofthebookscores,ontheonehandthesimilarityofbooklabelsarefoundinbooksbrowsinghistorytofurtherinfertheuser'sfavoritekindofbooks,andontheotherhandreaders’favoritebookscanalsobejudgedbythesimilarityhighscorebooksontheratingscale.Consideringthetwoaspectsofinformationabove,collaborativefilteringrecommendationalgorithmisusedtocalculatethesimilaritybetweenuserIDtoelecttwoorthreebookswhichhavethemostcomprehensivesimilarityandrecommendtotheusers.

Keywords:

overlap、similarity、principalcomponentanalysis、、factoranalysis、collaborativefilteringalgorithm

1PROBLEMRESTATED

WiththedevelopmentofinformationtechnologyandInternet,peoplegraduallywentoutintoaninformationoverloadedworld.Atthispoint,whetherinformationconsumersorinformationproducersareexperiencinggreatchallenges:

Asforinformationconsumers,itisaverydifficulttasktofindtheinterestinginformationfromalargeamountsofinformation.Forinformationproducers,lettingtheirownproductionofinformationstandoutanddrawinguser'sattentionwouldalsobeaverytoughissue.

Recommendationisakeytooltoresolvethiscontradiction.TheproductsandapplicationsarewidelyusedontheInternet,includingtherelevantsearch,recommendedtopic,variousproductsinelectroniccommercerecommendation,datingandrecommendationonthesocialnetwork.

Weobtainedawell-knownuserbehaviorinformationonlinebookstores,includingratingstatisticsforbooks,taginformationandtheuser'ssocialrelationships.Pleasecompletethefollowingquestionsbasedonthedata.

1Analyzingthefactorsaffectingtheassessmentofbooks;

2Developamodel,predictingbookscoresintheattachmentofpredict.txt;

3Aimingpredict.txtAnnexusers,werecommendthreebookstouserswhodidnotreadthesebooksbefore.

2PROBLEMANALYSIS

QuestionOne:

Toanalyzetheimpactofthebookusersratingfactors,itintroducestheannexbook_tag.txt,user_book_score.txt,user_read_history.txt,user_social.txtdatatothesoftware,afterdataprocessing,dataattachmentscanbeseeninuser_book_score.txtbookIDandtheuserIDofthemainfactorsaffectingthescore,AfterAnnexbook_tag.txt,user_social.txtuser_read_history.txtthedataanalysis,thelabelrepresentsdifferenttypesofbooks,thusaffectingthescore.Userscanfindthelabelsofvariouskindsofbooks,lookingthroughthehistorybooksoftheIDaccordingtobookswithdifferentID,.ForusersconcerningaboutthenumberoffriendsID,youcananalyzetherelationshipbetweentheobtaineddegreeofusers’choicesandfavoritebooks,andfinallydrawcontactbetweentheusersandthebooksandaffectthescore.

QuestionTwo:

Torequireuserstopredictthescoreofbooksinpredict.txtannex,itmustbebasedonaknownproblemaffectingusersofbooksscorefactorsanduseprincipalcomponentanalysismodeledbyprincipalcomponentanalysis:

Consideringthefactorsintheuserassessmentofbooks.Firstofall,thedatausingsoftwarekmotest,thetestsig<0.05,principalcomponentanalysisofthedata.AsthebookID,userIDisjustasymbolinordertoensurereliabilityoftheresults,thentheneedtostandardizetherawindexdataprocessing,compositionanalysisoffiveindicatorsshots,whenacquiredthreemaincomponents,thecumulativecontributionrateof81.371%,andusingthesethreeprincipalcomponentsfactoranalysis,principalcomponentsolutionwaseventuallyintegratedmodelis:

ultimatelypredictpredict.txtuserratingsforbooks.

Questionthree:

Aftertheuserviewingthehistorybooksaswellasanalysisofthescoresofbooks,ontheonehand,wefindthesimilarityofbookstaggedbooksfrombrowsinghistoryandinferusers’favoritebooks.Onanotherhand,findingtheratingscalehighlevelsimilarityofbookscanbejudged.Consideringtheabovetwoaspectsofinformation,withtheusageofcollaborativefilteringrecommendationalgorithm,wecalculatethesimilaritybetweentheusers’IDtoelectthemostcomprehensivesimilarityamongthreebooksrecommendedtotheuser.

3.MODELASSUMPTIONS

1Assumesthatthedatafollowanormaldistribution.

2Variousfactorsareindependenteachother.

3Assumingcharacteristicvalues​​canbeseeninawayisaprincipalcomponentofimpactindicatorseffortssize.

4.Ignoretheoveralluserratingsforbookserrorsbasedonlimitedsampledata.

4SYMBOLDESCRIPYTION

Userdatarates

BooksIDstandardization

NumberofFriends

UserIDstandardization

Thenumberofbookslabels

Usersbrowsinghistory

Principalcomponent1

Principalcomponent2

Principalcomponent3

Principalcomponent

Commonusersandexcessivecollectionofitemscomment

Userratingsforbooks

Averageuserratingofbooks

Averageuserratingofbooks

（5）ESTABLISHANDSOVELTHEMODEL

5.1QuestionOne:

5.1.1Establishmentofamodel:

Accordingtothedatagiveninthetitlemeaningandaccessories,weanalyzesAnnexuser_book_score.txtuserID,booksID,listingthreecolumnsofdata,thefirststageisdividedintothreecontact:

Contacts

（1）theconnectionbetweenbooksandbooksID

（2）theconnectionbetweentheuserIDoftheuserID;

（3）theconnectionbetweentheuserIDandtheIDofbooks.

ThesecondlevelgiveninAnnexbook_tag.txtbookslabeldatafurtheranalysis:

（1）Foruserswhohavesamehobbies,selectedbookswillbesimilar,sothedegreeofoverlapwouldbehigh,booksoverlapsizecomparisonbetweenthelabelscandeterminethesimilaritybetweenthebookandthebooksize;

（2）ThenaccordingtoAnnexuser_read_history.txtuserviewedhistorybooksIDDataAnalysis:

Comparisonofthesimilaritybetweentheuserreadthebookcandeterminethesizeofcontactbetweentheusers.

（3）AnalyzingdataofAnnexuser_book_score.txt,accordingtouserbrowsingthroughthebooksofhistoryandscoringthelevelofsimilarity,itcanbebroughtbacktothesizeofthesimilaritybetweenthebooksandlabelanalysis,wecanlettheuserstoselectthedegreeandinterestsofbooksduetotherelationshipbetweentheobtainedcontactbetweentheusersandbooks.Conclusioncanbedrawnonthesecondstagebythefirststage,whichmeansanalysisoftheimpactoftheuseri.e.bookratingfactors.

Thefirststage

Thesecondstage

Figure5-1scoreaffectusersonbooksanalysischart

Figure5-1EffectcreateuserratinganalysisdiagramBooks,influencetherelationshipbetweenthebookIDlabeloverlapbooks,booksoverlapbooktagIDtotherelationshipbetweenthereaction,therelationshipbetweenthebookIDandtheinteractionscore;therelationshipbetweenbooksanduserIDaffecttheuser'sscorereadbooks,readbooksuserratingscounterproductivetotherelationshipbetweenbooksanduserID,userIDbetweenbooksandtherelationshipbetweenscoreinteraction;affecttherelationshipbetweentheuserIDusersreadbooks,readbooksontheuserreactiontotherelationshipbetweenauserID.

5.1.2Solvingamodel:

Figure4-1affectedbyuserratinganalysischartbooksanalyzetheimpactonusersofbooksscorefactors:

（1）IDbooks

Tagnumber

（2）books

（3）UserIDhistorybooksviewed

（4）thenumberofusersconcernedfriendsID

（5）theuserID

5.1.2Solvingamodel:

Figure4-1affectedbyuserratinganalysischartbooksanalyzetheimpactonusersofbooksscorefactors:

（1）IDbooks

Tagnumber

（2）books

（3）UserIDhistorybooksviewed

（4）thenumberofusersconcernedfriendsID

（5）theuserID

5.2