Full Citation

Title: “Where Far Can Be Close”: Finding Distant Neighbors In Recommender Systems

Citation Type: Miscellaneous

Publication Year: 2015

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: Location and its corollary, distance, are critical concepts in social computing. Recommender systems that incorporate location have generally assumed that the utility of location- awareness monotonically decreases as entities get farther apart. However, it is well known in geography that places that are distant “as the crow flies” can be more similar and connected than nearby places (e.g., by demographics, expe- riences, or socioeconomic). We adopt theory and statistical methods from geography to demonstrate that a more nu- anced consideration of distance in which “far can be close” – that is, grouping users with their “distant neighbors” – mod- erately improves both traditional and location-aware rec- ommender systems. We show that the distant neighbors approach leads to small improvements in predictive accu- racy and recommender utility of an item-item recommender compared to a “nearby neighbors” approach as well as other baselines. We also highlight an increase in recommender utility for new users with the use of distant neighbors com- pared to other traditional approaches.

Url: https://pdfs.semanticscholar.org/2784/703839ebee08fe14acc611475cdb84758f7e.pdf

User Submitted?: No

Authors: Kumar, Vikas; Jarratt, Daniel; Anand, Rahul; Konstan, Joseph, A; Hecht, Brent

Publisher: University of Minnesota, Twin Cities

Data Collections: IPUMS USA

Topics: Population Data Science

Countries: United States