NUST Institutions Library Catalogue catalog › Details for: Cross-Domain Recommendation using Deep Learning /

Normal view MARC view ISBD view

Cross-Domain Recommendation using Deep Learning / Rabia Khan

By: Khan, Rabia Contributor(s): Supervised by Dr. Naima Iltaf Material type: Text

TextPublisher: Rawalpindi, MCS (NUST), 2025Description: xix, 170 pSubject(s): PhD Computer Software Engineering Thesis | PhD CSE ThesisDDC classification: 005.1,KHA

Contents:

Recommender systemsplayacrucialroleinfilteringvastinformation,buttraditionalsingle- domain modelsstrugglewithdatasparsity,cold-startissues,andlimitedpersonalization. These challengesareamplifiedasuserpreferencesoftenspanmultipledomains,which single-domain recommendersfailtocapture.Toaddresstheseissues,thisthesisexplores cross-domain recommendation using Multi-Layer Perceptron's (MLPs)to model non-linear dependencies andaligndomain-specificfeatures,improvingpersonalizationandscalability. However,MLPsoverlookthegraph-structurednatureofuser-iteminteractions,limiting their abilitytomodelhigher-orderrelationships.Toovercomethis,aGraphConvolutional Network-basedframeworkisproposedtolearnbothdomain-specificandsharedembed- dings throughcollaborativegraphsignals.Additionally,tocapturefine-grainedpreferences, another modelisproposedthatintegratesmetapath-basedgraphswithtransformer-based reviewencoding.Thiscapturesopiniondynamicsfromreviewsandenhancesknowledge transfer insparsesettings. Finally,recognizingthelimitationsofratings-onlymodelsandbiasedsharedembeddings, a sentiment-awarecontrastivelearningframeworkisintroduced.Bydynamicallyadapt- ing userrepresentationsusingreviewsentiment,themodelachievesimprovedcross-domain personalization, mitigatesnegativetransfer,andsupportsaccuraterecommendationsacross heterogeneous domains like movies and books.

Tags from this library: No tags from this library for this title. Log in to add tags.

Holdings ( 1 )
Title notes ( 1 )
Comments ( 0 )

Item type	Current location	Home library	Shelving location	Call number	Status	Date due	Barcode	Item holds
Thesis	Military College of Signals (MCS)	Military College of Signals (MCS)	Thesis	005.1,KHA (Browse shelf)	Available		MCSPhD CSE-26

Total holds: 0

Browsing Military College of Signals (MCS) shelves, Shelving location: Thesis Close shelf browser

Previous	No cover image available	No cover image available	No cover image available	No cover image available	No cover image available	No cover image available	No cover image available	Next
Previous	005.1,JAV A Web of Things Architecture Focusing on Trust /	005.1,JAV Development of Crowd Behavior Analysis Techniques /	005.1,KAN Development of Improved Saliency Detection Techniques /	005.1,KHA Cross-Domain Recommendation using Deep Learning /	005.1,NAS The Impact of Business Analytics in Information Security Risk Management /	005.1,NAW Adaptive Trust Calculation in Fog Computing /	005.1,NAW Mining Crowd Sourcing Repositories for Open Innovation in Software /	Next

Recommender systemsplayacrucialroleinfilteringvastinformation,buttraditionalsingle-
domain modelsstrugglewithdatasparsity,cold-startissues,andlimitedpersonalization.
These challengesareamplifiedasuserpreferencesoftenspanmultipledomains,which
single-domain recommendersfailtocapture.Toaddresstheseissues,thisthesisexplores
cross-domain recommendation using Multi-Layer Perceptron's (MLPs)to model non-linear
dependencies andaligndomain-specificfeatures,improvingpersonalizationandscalability.
However,MLPsoverlookthegraph-structurednatureofuser-iteminteractions,limiting
their abilitytomodelhigher-orderrelationships.Toovercomethis,aGraphConvolutional
Network-basedframeworkisproposedtolearnbothdomain-specificandsharedembed-
dings throughcollaborativegraphsignals.Additionally,tocapturefine-grainedpreferences,
another modelisproposedthatintegratesmetapath-basedgraphswithtransformer-based
reviewencoding.Thiscapturesopiniondynamicsfromreviewsandenhancesknowledge
transfer insparsesettings.
Finally,recognizingthelimitationsofratings-onlymodelsandbiasedsharedembeddings,
a sentiment-awarecontrastivelearningframeworkisintroduced.Bydynamicallyadapt-
ing userrepresentationsusingreviewsentiment,themodelachievesimprovedcross-domain
personalization, mitigatesnegativetransfer,andsupportsaccuraterecommendationsacross
heterogeneous domains like movies and books.

There are no comments on this title.

to post a comment.