Keyword and Citation Network based Citation Recommendation

As research in different aspects of science is growing rapidly, there is no doubt that scientific resources such as papers and books are the main source of information and knowledge for scientists and researchers. They would therefore need to know the important resources and identify outstanding people in their fields of interest. This need becomes especially more important for novice researchers, as they tend to have less experience in their jobs. In particular, when researchers try to search in interdisciplinary fields, finding the best existing resources on the web is an exhausting operation. Some previous literature has focused on this special aim of information retrieval, yet as they tend to involve text-processing approaches, they are very time-consuming. This research proposes a paper recommender system based on input topic keywords and the structure of the citation networks. It then outputs a ranked list of the most important and relevant resources in the specified fields. The advantage of this method compared to previous approaches is in the search, extraction and ranking of the retrieved documents. Whereas previous methods use the text of the documents in order to determine their similarity and relevancy, this method greatly reduces the processing time by avoiding full text processing techniques. The results of the evaluation part show that the proposed method s results are more accurate and relevant to the users queries. The method is evaluated based on the coverage criteria. The proposed method’s coverage has been compared with existing methods and results show that this method outperforms the baseline methods considerably. In the ranking part, the evaluation has been done by calculating some criteria such as recall, NDCG and co-cited probability. This phase of the proposed method was also evaluated against other known method such as HITS, Cosine similarity and CRM and the outcomes showed improved results in all parameters.

Keywords:

Citation recommendation , recommender system , complex networks , collaboration network , keywords