A temporal expert finding methodology based on united author-document-topic graphs

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Enformatik Enstitüsü, Bilişim Sistemleri Anabilim Dalı, Türkiye

Tezin Onay Tarihi: 2014

Öğrenci: AHMET EMRE KILINÇ

Danışman: TUĞBA TAŞKAYA TEMİZEL

Özet:

Expert finding is a challenging research topic due to fast paced technological development resulting in changes in people’s expertise areas in time. However, the majority of the studies in the literature about expert finding systems do not take into account such temporal changes. For example, probabilistic models, which are widely used in this domain, are based on word or term associations between queries and documents. On the other hand, separated author-document graphs, which are used as baseline approach in this thesis, are based on topic modeling techniques. This approach does not take into consideration both queries and documents in the same topic modeling process, but it considers only documents in topic modeling process. As a result, it impairs relations between topic queries and documents. In this thesis, a novel expert finding system which uses domain limited Latent Dirichlet Allocation (LDA) based topic modeling and dynamic, united author-document-topic graphs is proposed. The proposed method is tested with ArnetMiner and UVT datasets and outperforms the baseline separated author-document-topic approach.