Recommending Publication Venue in Context Using Abstract

Tian-mu MA, Wei FANG


One of the key steps of composing a scientific work is selecting the most appropriate publication venue. All a researcher can rely on are experiences from previous study or retrieving similar topics provided by publication venues. A reliable submission recommender is thus in great demand. Unlike the existing works with the same endeavor, we propose a topic modeling based recommending system leveraging briefly abstracts which avoids the somehow unappeasable demand of full texts, references, etc. In this paper, we set up a model with carefully estimated parameters for fine-grained representation of the corpus. Combining with several schemes of generating candidate venues, a context filter based on publication dates is included to optimize the results. Our proposal was evaluated and compared with several baselines on a dataset of 23,766 papers of 18,862 authors from 373 venues. Extensive experiments illustrate that our recommending system can product more satisfactory accuracy among other methods while requiring far more less information.


Topic modeling, Information retrieval, Recommending system, Topic coherence


Full Text:



  • There are currently no refbacks.