Metoda samonadziranog učenja za sustave preporuke temeljena na slučajnim šetnjama u grafovima

undergraduate thesis

• Title (english): Self-supervised learning method for recommender systems based on random walks in graphs
• Author: Andrea Rakocija
• Mentor: Mario Brčić (mentor)
• Committee member: Mario Brčić (predsjednik povjerenstva)
• Committee member: Agneza Krajna (član povjerenstva)
• Committee member: Juraj Dončević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-15, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Zadaća algoritama preporuke je rangiranje i selekcija sadržaja koji su zanimljivi pojedinom korisniku. U sustavima preporuke temeljenim na učenju postoji problem hladnog početka (engl. cold-start) do kojeg dolazi kada ne postoje produkcijski, stvarni podatci za korisnike. Cilj ovog rada je konstruirati pomoćni zadatak (engl. artificial task) temeljen na slučajnim šetnjama u svrhu rješavanja problema hladnog početka. Pomoćni zadatak će omogućiti samonadzirano učenje za stvaranje dobrih reprezentacija na autorskim materijalima. Na temelju dobivenog skupa podataka stvorit će se model pojačanih stabala (engl. boosting trees). Potom će se analizirati pretpostavke i mane konstruiranog problema za učenje za potrebe inicijalnog korištenja modela u produkcijskom sustavu.

Abstract (english)

The task od recommendation algorithms is to rank and select content that is interesting to an individual user. In learning-based recommendation system, there is cold-start problem that occurs when there is no production, actual data for users. The aim of this paper is to construct a pretext task based on random walks in order to solve the problem of cold start. The pretext task will enable self-supervised learning to create good representations on the author's materials. A model of the boosting trees will be created based on the obtained dataset. Then, the assumptions and shortcomings of the constructed learning problem will be analyzed for the purposes of the initial use of the model in the production system.

• Keywords: sustav preporuke; samonadzirano učenje; hladni početak; pomoćni zadatak; pojačana stabala
• Keywords (english): recommender system; self-supervised learning; cold-start; pretext task; boosting trees
• Language: croatian
• URN:NBN: urn:nbn:hr:168:975088
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 21:31:46