diplomsko delo
Jan Šušteršič (Author), Zoran Bosnić (Mentor), Domen Košir (Co-mentor)

Abstract

V zadnjih dveh desetletjih se je zaradi velike porasti uporabe interneta zelo razširilo spletno oglaševanje, kjer je eden izmed največkrat uporabljenih pokazateljev uspešnosti stopnja interakcije oglasov, ki je odvisna od številnih dejavnikov. V diplomski nalogi se osredotočamo na napovedovanje stopnje interakcije z oglasi z uporabo priporočilnih sistemov, ki temeljijo na matrični faktorizaciji. Pri tem poskušamo napovedi osnovne matrične faktorizacije izboljšati z upoštevanjem časovnih podatkov o uri in datumu, iz katerih lahko razberemo starost zapisa in kontekst, v katerem je bil ta zabeležen. Zgrajen časovno pogojeni priporočilni sistem primerjamo s statičnimi modeli in analiziramo stopnjo izboljšanja. Rezultati so pokazali, da z uporabo podatkov o starosti zapisov na dani množici podatkov konsistentno izboljšamo napovedi, medtem ko uporaba kontekstnih podatkov pripelje do slabših rezultatov in zahteva nadaljnje raziskave.

Keywords

oglaševanje;priporočilni sistemi;matrična faktorizacija;čas;računalništvo in informatika;univerzitetni študij;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [J. Šušteršič]
UDC: 004:659.1(043.2)
COBISS: 1538539459 Link will open in a new window
Views: 997
Downloads: 233
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Time aware recommender system for advertising
Secondary abstract: Due to the vast increase in the usage of internet in the past two decades there has been a major increase in the popularity of online advertising where one of the most widely used success indicators is ad interaction rate which depends upon multiple factors. The thesis focuses on predicting ad interaction rate with the use of recommender systems that are based on matrix factorization. We try to improve the basic matrix factorization by incorporating time data such as age and context into our recommendations. We then compare the time-aware recommender system with its more basic counterpart and analyse performance improvements. Results show that the use of information such as age can consistently improve recommendations on our dataset. Context information however, produces worse results and requires further research.
Secondary keywords: advertisement;recommender systems;matrix factorization;time;computer and information science;diploma;
Type (COBISS): Bachelor thesis/paper
Study programme: 1000468
Embargo end date (OpenAIRE): 1970-01-01
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 55 str.
ID: 11417797