Please use this identifier to cite or link to this item: http://hdl.handle.net/10125/64055

Follow-back Recommendations for Sports Bettors: A Twitter-based Approach

File Size Format  
0253.pdf 790.46 kB Adobe PDF View/Open

Item Summary

Title:Follow-back Recommendations for Sports Bettors: A Twitter-based Approach
Authors:Wandabwa, Herman
Naeem, Muhammad Asif
Mirza, Farhaan
Pears, Russel
Keywords:Decision Making in Online Social Networks
data mining
information retrieval
social networks
twitter
show 1 moreuser profiling
show less
Date Issued:07 Jan 2020
Abstract:Social network based recommender systems are powered by a complex web of social discussions and user connections. Short text microblogs e.g. Twitter present powerful frameworks for information consumption, due to their real-time nature in content throughput as well as user connections. Therefore, users on such platforms consume the disseminated content to a greater or lesser extent based on their interests. Quantifying this degree of interest is a difficult task based on the amount of information that such platforms generate at any given time. Thus, the generation of personalized profiles based on the Degree of Interest (DoI) that users have towards certain topics in such short texts presents a research problem. We address this challenge by following a two-step process in generation of personalized sports betting related user profiles in tweets as a case study. We (i) compute the Degree of Interest in Sports Betting (DoiSB) of tweeters and (ii) affirm this DoiSB by correlating it with their friendship network. This is an integral process in the design of a short text based recommender systems for users to follow i.e follow-back recommendations as well as content-based recommendations relying on the interests of users on such platforms. In this paper, we described the DoiSB computation and follow-back recommendation process by building a vector representation model for tweets. We then use this model to profile users interested in sports betting. Experiments using real Twitter dataset geolocated to Kenya shows the effectiveness of our approach in the identification of tweeter's DoiSBs as well as their correlation with their friendship network.
Pages/Duration:10 pages
URI:http://hdl.handle.net/10125/64055
ISBN:978-0-9981331-3-3
DOI:10.24251/HICSS.2020.313
Rights:Attribution-NonCommercial-NoDerivatives 4.0 International
https://creativecommons.org/licenses/by-nc-nd/4.0/
Appears in Collections: Decision Making in Online Social Networks


Please email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.

This item is licensed under a Creative Commons License Creative Commons