Date

2021-3-05

Author

Gürlek, Mesut

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

351
views

1061
downloads

Cite This

People are using social media platforms more and more every day; hence, they are be-coming suitable for research studies by their rich content. Twitter is one of the biggestand most widely used social media platforms, and many studies focus on Twitter forsocial media research. In this thesis, we propose methodologies for determining usertypes of Twitter accounts by their metadata, content, and structure. Our first problemis classifying organization vs. individual account types using only metadata. After wegive details about the data collection and analysis process, we clean text features andnormalize number features. Proposed model contains LSTM for textual features andMLP for numeric features. Language independent word embeddings make the pro-posed model useful for accounts in different languages and multi-language accounts.Experiments show that our model requires less data and performs better results thanprevious studies about account type classification. The second problem is to person-ality type prediction of individual accounts. Ground truth data comes from Big FivePersonality tests and Twitter accounts of a group of people. This time model also con-tains tweets and graph features of the account. Personalities are classified in terms oftheir Big Five Test scores, and for each OCEAN feature, we train different models.Experiment results are promising even though we study with a small set of users.

Subject Keywords

Twitter, Account classification, Organization vs. individual, Account meta-data, Big five personality, Personality prediction

URI

https://hdl.handle.net/11511/89834

Collections

Graduate School of Natural and Applied Sciences, Thesis