Online social network risk management
Abstract/Contents
- Abstract
- As online social networks become more influential in society, a new form of cyber threat continues to impact its users: platform manipulation. This occurs when social network users attempt to exploit other users through their online behavior on a social network platform. Current methods for detecting and eliminating platform manipulation rely on the use of machine learning (ML) and artificial intelligence (AI) models. While these methods are able to parse through very large data sets efficiently, they are often trained to only recognize specific traits within user behavior (e.g., spam, malicious links, hate speech), and may not be able to incorporate other forms of evidence (e.g., human observations) in real time without being re-trained and validated, which can be costly and time-consuming in practice. To make an accurate risk assessment of a user's long-term behavior which can be updated over time, decision-makers need a method for integrating multiple forms of evidence (AI and human) across space (by malicious trait) and time into a probability distribution over possible scenarios, as well as knowledge of the potential consequences to the platform from each scenario. In this research, we use probabilistic risk analysis to combine both AI-generated and human-generated evidence in order to determine the risk of individual users to the platform. This allows decision-makers to not only consider the probability that users are malicious, but also their range of potential consequences. We also develop a decision analysis framework for platform referral decisions (whether to request human intervention) using the value of clairvoyance on human agent observations to determine when to involve human agents in the decision-making process (and when to rely on automated decisions to remove users from the platform).
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2022; ©2022 |
| Publication date | 2022; 2022 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Mogensen, Matthew David |
|---|---|
| Degree supervisor | Paté-Cornell, M. Elisabeth (Marie Elisabeth) |
| Thesis advisor | Paté-Cornell, M. Elisabeth (Marie Elisabeth) |
| Thesis advisor | Ashlagi, Itai |
| Thesis advisor | Shachter, Ross D |
| Degree committee member | Ashlagi, Itai |
| Degree committee member | Shachter, Ross D |
| Associated with | Stanford University, Department of Management Science and Engineering |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Matthew D. Mogensen. |
|---|---|
| Note | Submitted to the Department of Management Science and Engineering. |
| Thesis | Thesis Ph.D. Stanford University 2022. |
| Location | https://purl.stanford.edu/wn227pp6238 |
Access conditions
- Copyright
- © 2022 by Matthew David Mogensen
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...