Smart Book Recommender Evaluation Data
2018-04-05T16:53:03Z (GMT) by
This compressed zip file contains all the data regarding the evaluation study of the Smart Book Recommender (SBR). SBR is a semantically enhanced recommendation engine, developed in collaboration with Springer Nature (SN), for suggesting Springer books, journals and conference proceedings for conferences. SBR characterises conferences in terms of the topics covered in their proceedings and applying cosine pairwise similarity computation to produce a set of recommendations.
We evaluated SBR on Springer conference series in the field of Computer Science with the help of seven publishing editors and seven researchers from The Open University. The evaluation consisted of two parts:
1) A quantitative analysis, which focused on analysing how evaluators rated each SBR recommendations for two conference series of their choosing.
2) A qualitative analysis, which focused on analysing the answers of the surveys that the evaluators completed after a hands-on session with SBR.
This compressed file contains the following two CSV files: i) Quantitative - Recommendation Ratings, and ii) Qualitative - Survey Responses. Below we describe the datasets.