Evaluating Item-Item Similarity Algorithms for Movies

Colucci, Lucas; Doshi, Prachi; Lee, Kun-Lin; Liang, Jiajie; Lin, Yin; Vashishtha, Ishan; Zhang, Jia; Jude, Alvin

doi:10.1145/2851581.2892362

Cited by 17 publications

(16 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our goal was to compare supervised similarity methods against unsupervised methods with a user test. Here we first describe the previous work [CDL16] for context as it supplied the labels used, inspired the user interface of our study, and provided a benchmark against which we would compare our methods. Then we how we built and tested methods offline to decide which machine learning methods and features should be used in the user study, which is presented in Section 4 and Section 5.…”

Section: Building Similarity Methodsmentioning

confidence: 99%

“…Colucci et. al [CDL16] evaluated existing movie similarity methods from a user perspective, and showed these methods matched user perspective about half the time. They implemented four similarity methods, two based on CB similarity and two based on CF similarity.…”

Section: Existing Dataset and Methodsmentioning

confidence: 99%

“…The full plot was longer than plot and therefore could have more relevant keywords. Previous work [CDL16] proposed that an older candidate movie may be seen as less similar than a newer one, so we engineered the feature age difference. Let M 1 and M 2 be the two movies for which we are evaluating similarity, then:…”

Section: Our Methodsmentioning

confidence: 99%

“…We know that movie pairs with high CF similarity can be perceived to be similar [CDL16]. We further believed that CF and CB can be hybridised to produce a single method that considers both, where two movies are considered similar if they shared metadata and had common raters.…”

Section: Our Methodsmentioning

confidence: 99%

“…In this paper, we use a small dataset from Colucci et al [CDL16] to learn four different ML models. Experimentally, we show that the four different models can be used to predict movie similarity in a way that is consistent with users' perceived similarity.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Finding Similar Movies: Dataset, Tools, and Methods

Leng¹,

Paulino²,

Haider³

et al. 2018

CSRN

Self Cite

View full text Add to dashboard Cite

Recommender systems are becoming ubiquitous in online commerce as well as in video-on-demand (VOD) and music streaming services. A popular form of giving recommendations is to base them on a currently selected product (or items), and provide "More Like This," "Items Similar to This," or "People Who Bought This also Bought" functionality. These recommendations are based on similarity computations, also known as item-item similarity computations. Such computations are typically implemented by heuristic algorithms, which may not match the perceived item-item similarity of users. In contrast, we study in this paper a data-driven approach to similarity for movies using labels crowdsourced from a previous work. Specifically, we develop four similarity methods and investigate how user-contributed labels can be used to improve similarity computations to better match user perceptions in movie recommendations. These four methods were tested against the best known method with a user experiment (n = 114) using the MovieLens 20M dataset. Our experiment showed that all our supervised methods beat the unsupervised benchmark and the differences were both statistically and practically significant. This paper's main contributions include user evaluation of similarity methods for movies, user-contributed labels indicating movie similarities, and code for the annotation tool which can be found at http://MovieSim.org.

show abstract

Section: Building Similarity Methodsmentioning

confidence: 99%

Section: Existing Dataset and Methodsmentioning

confidence: 99%

Section: Our Methodsmentioning

confidence: 99%

Section: Our Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations