Bayesian Networks-based personal data synthesis
| Autor: | Irina Deeva, Alexander V. Boukhanovsky, Petr Andriushchenko, Anna V. Kalyuzhnaya |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Social network
business.industry Computer science media_common.quotation_subject Probabilistic logic Inference Bayesian network Machine learning computer.software_genre Synthetic data Joint probability distribution Quality (business) Social media Artificial intelligence business computer media_common |
| Zdroj: | GOODTECHS |
| Popis: | Often, confidentiality problems and a lack of original data, make it challenging to analyze user data carefully. In such situations, synthetic data can be used that is more suitable for testing and training marketing strategies, personalized assistants, or behavior analysis systems than the original data. In this paper, the approach for generating synthetic social media profiles data based on Bayesian networks was analyzed. The personal data synthesis problem was considered as the inference of a joint probability distribution from the oriented probabilistic models like Bayesian networks. The quality of this approach in generating VKontakte (VK is the Russian analog of Facebook) social network data was demonstrated and assessed. The Bayesian network approach has shown itself well in the tasks of deriving joint and marginal data distributions, which has led to the production of high-quality synthetic personal data. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|