Search

The task of assessing the similarity of research papers is of interest in a variety of application contexts. It is a challenging task, however, as the full text of the papers is often not available, and similarity needs to be determined based on the papers' abstract, and some additional features such as their authors, keywords, and the journals in which they were published. Our work explores several methods to exploit this information, first by using methods based on the vector space model and then by adapting language modeling techniques to this end. In the first case, in addition to a number of standard approaches we experiment with the use of a form of explicit semantic analysis. In the second case, the basic strategy we pursue is to augment the information contained in the abstract by interpolating the corresponding language model with language models for the authors, keywords and journal of the paper. This strategy is then extended by revealing the latent topic structure of the collection using an adaptation of Latent Dirichlet Allocation, in which the keywords that were provided by the authors are used to guide the process. Experimental analysis shows that a well-considered use of these techniques significantly improves the results of the standard vector space model approach.

Due to the increasing number of conferences, researchers need to spend more and more time browsing through the respective calls for papers (CFPs) to identify those conferences which might be of interest to them. In this paper we study several content-based techniques to filter CFPs retrieved from the web. To this end, we explore how to exploit the information available in a typical CFP: a short introductory text, topics in the scope of the conference, and the names of the people in the program committee. While the introductory text and the topics can be directly used to model the document (e.g. to derive a tf-idf weighted vector), the names of the members of the program committee can be used in several indirect ways. One strategy we pursue in particular is to take into account the papers that these people have recently written. Along similar lines, to find out the research interests of the users, and thus to decide which CFPs to select, we look at the abstracts of the papers that they have recently written. We compare and contrast a number of approaches based on the vector space model and on generative language models.

While collaborative filtering and citation analysis have been well studied for research paper recommender systems, content-based approaches typically restrict themselves to straightforward application of the vector space model. However, various types of metadata containing potentially useful information are usually available as well. Our work explores several methods to exploit this information in combination with different similarity measures.