TY  - BOOK
AU  - Supapron Tantanasiriwong  
AU  - Guha,Sumanta
AU  - Janecek,Paul
AU  - Choochart Haruechaiyasak, 
AU  - Shyu,Mei-Ling
ED  - Ministry of Science and Technology National Science and Technology Development Agency (NSTDA), Thailand,
TI  - Cross-domain citation recommendation base on topic model and co-citation selection
T2  - Dissertation
PY  - 2016///
CY  - Pathum Thani, Thailand
PB  - Asian Institute of Technology
KW  - Knowledge management
KW  - Information and retrieval systems
KW  - Database management
N1  - A dissertation submitted in partial fulfillment of the requirements for the  degree of Doctor of Philosophy in Information Management; Thesis (Ph.D.) - Asian Institute of Technology, 2016
N2  - Finding publications for citation is a crucial task in the research community. Experienced researchers may indeed find it easy to determine relevant papers from a single source domain such as a publication database. However, the task is significantly more challenging when considering publications across different domain sources. In this dissertation, we propose a Cross-Domain Recommender System as a solution to the task. Our recommender system implements an algorithm which is a hybridization of two distinct approaches {u2013} the Topic Model and Co-Citation Selection approaches. Briefly, relevant terms from documents are first clustered into similar topics. The Co-Citation Selection technique then helps select citations based on a set of highly similar documents. To evaluate the performance, various comparison techniques are introduced to the CrossDomain Citation Recommendations (CDCR). We focus on the context of patent and publication as a primary and secondary domain, respectively. The outcome of our proposed cross-domain citation framework is confirmed through benchmarking with traditional baseline approaches using a corpus of patents collected from different technological fields, e.g., biotechnology, environmental technology, medical technology, and nanotechnology. Experimental results show that our hybrid algorithm approach yields better performance in predicting relevant publication citations than the known baseline approaches
UR  - http://203.159.5.9/ait-thesis/detail.php?q=B06999
ER  -