Politecnico di Torino (logo)

A financial approach for correlation with exogenous data and synergy detection in social networks

Fabio Bertone

A financial approach for correlation with exogenous data and synergy detection in social networks.

Rel. Luca Vassio, Martino Trevisan. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2022

PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (4MB) | Preview

This thesis studies the universe of Online Social Networks (OSNs) and their influencers, i.e., popular users, by applying instruments that typically belongs to the financial fields, technical analysis in particular. Two aspects of OSNs have been investigated. The first is the correlation between social network dynamics (e.g. fanbase evolution) and other exogenous dynamics (e.g. search engines queries). The second is the synergy between couples of influencers, meant as the highly correlated movement of dynamically normalized social network metrics of two influencers during a variable length time interval. First, this work provides a basic understanding of the fundamental financial concepts on top of which our reasoning is built. The first is the so called "Efficient Market Hypothesis" (EMH). A market is said to be efficient if at time t+eps it fully reflects the information available up to time t. The second one is the study of the Bollinger bands, an instrument belonging to the technical analysis of the financial markets. The relative position of the signal with respect to these bands allows to dynamically normalize signals that are of different scale and volatility. Given these two ingredients, in this thesis it has been developed the analogy between the OSNs world and the stock market. In this view, the fanbase cardinality of an influencer can be seen as the price of a stock, and the followers, seen as buyers, can purchase such stock by the act of following. The objective is to test if, and to which extent, followers’ dynamic is efficient in the sense of EMH. In order to assess that, the chosen source of exogenous information is the Google Trends Search Volume Index (SVI), measuring the amounts of queries submitted to the Google search engine (normalized on a scale from 0 to 100) having as a keyword the influencer. The endogenous variable, instead, is the fanbase cardinality of the given influencer. Fanbase cardinality and SVI are considered in the same time period. Once the data about these two measures over time has been collected, they are correlated through our efficiency measure. The latter is based on the average distance between the two dynamically normalized signals: the lower the distance, the higher the efficiency. In a perfectly efficient case, the SVI and the percentage increase of followers overlaps, while in the opposite case they are totally unrelated. This study shows that influencer with a relatively small (e.g. 1 million followers) are more efficient than the ones with a very big fanbase (e.g. 20 million followers). Taking constant the number of followers, the singer category is more efficient than VIP and athlete. The second aspect covered by this thesis is sinergy between influencers. To be clear, in this work when we use the word 'sinergy' the meaning is that two influencers are in sinergy in a certain time interval, when their dynamically normalized fanbase cardinalities (number of followers) change following trends that are highly (Pearson) correlated in such time interval. In he 10 scoring couples of influencers, 9 out of 10 have an actual reason to be correlated. For example, infuencers can be married, engaged or even YouTube partners. Differently from efficiency, sinergy applies equally both to small and big fanbase cardinalities. This work can be considered as a pioneeristic one, meant toshow that concepts from finance and instinstrument from technical analysis can successfully be employed in the study of OSNs macro dynamics, uncovering hidden behaviours.

Relators: Luca Vassio, Martino Trevisan
Academic year: 2022/23
Publication type: Electronic
Number of Pages: 61
Corso di laurea: Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering)
Classe di laurea: New organization > Master science > LM-32 - COMPUTER SYSTEMS ENGINEERING
Aziende collaboratrici: UNSPECIFIED
URI: http://webthesis.biblio.polito.it/id/eprint/24500
Modify record (reserved for operators) Modify record (reserved for operators)