Unit:
Κατεύθυνση / ειδίκευση Τεχνολογίες Πληροφορικής και Επικοινωνιών (ΤΠΕ)Πληροφορική
Author:
Liosis Ploutarchos
Supervisors info:
Τσαλγατίδου Αφροδίτη, Αναπληρώτρια Καθηγήτρια, Τμήμα Πληροφορικής και Τηλεπικοινωνιών, Σχολή Θετικών Επιστημών, Εθνικό και Καποδιστριακό Πανεπιστημίο Αθηνών
Original Title:
Measuring influence of users and photos regarding a specific topic in Flickr
Translated title:
Measuring influence of users and photos regarding a specific topic in Flickr
Summary:
This M.Sc. dissertation presents a comprehensive approach to influence detection and suggestions regarding a specific topic, within the context of the photo-sharing platform Flickr.
After a quick review of three of the most used filtering techniques, specifically content-based filtering, collaborative filtering and hybrid filtering, we present a solution that consists of a MySQL database, a backend service layer - built on top of Flickr’s APIs and the database - implemented using the Spring Boot Framework, and an Angular web application that provides end-users with an interface to interact with the web servers.
A key part of the proposed system is the recommendation engine, which identifies related tags that will compose a specific topic, using similarity criteria such as Euclidean distance, Cosine similarity and Weighted Measure Similarity formulas.
Our ultimate goal is to provide actionable insights into the key drivers of influence within the platform, identifying trends and influential photos and users, while creating a scalable, data-driven system that can adapt to a wide range of topics.
Main subject category:
Technology - Computer science
Keywords:
recommendation systems, influence, social networks
