Embedding Factorization Models for Jointly Recommending Items and User Generated Lists

Cao, Da; Nie, Liqiang; He, Xiangnan; Wei, Xiaochi; Zhu, Shengqi; Chua, Tat-Seng

doi:10.1145/3077136.3080779

Cited by 103 publications

(75 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…He et al [17] propose a hierarchical self-attentive model for recommending user-generated item lists (e.g., book lists and playlists) to right users. Besides, the List Recommending Model in [26] is proposed for recommending book lists and the Embedding Factorization Model in [3] is for recommending song playlists.…”

Section: Related Workmentioning

confidence: 99%

Consistency-Aware Recommendation for User-Generated Item List Continuation

Zhang

Liu

et al. 2020

Proceedings of the 13th International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

User-generated item lists are popular on many platforms. Examples include video-based playlists on YouTube, image-based lists (or "boards") on Pinterest, book-based lists on Goodreads, and answerbased lists on question-answer forums like Zhihu. As users create these lists, a common challenge is in identifying what items to curate next. Some lists are organized around particular genres or topics, while others are seemingly incoherent, reflecting individual preferences for what items belong together. Furthermore, this heterogeneity in item consistency may vary from platform to platform, and from sub-community to sub-community. Hence, this paper proposes a generalizable approach for user-generated item list continuation. Complementary to methods that exploit specific content patterns (e.g., as in song-based playlists that rely on audio features), the proposed approach models the consistency of item lists based on human curation patterns, and so can be deployed across a wide range of varying item types (e.g., videos, images, books). A key contribution is in intelligently combining two preference models via a novel consistency-aware gating network -a general user preference model that captures a user's overall interests, and a current preference priority model that captures a user's current (as of the most recent item) interests. In this way, the proposed consistency-aware recommender can dynamically adapt as user preferences evolve. Evaluation over four datasets (of songs, books, and answers) confirms these observations and demonstrates the effectiveness of the proposed model versus stateof-the-art alternatives. Further, all code and data are available at https://github.com/heyunh2015/ListContinuation_WSDM2020.

show abstract

Section: Related Workmentioning

confidence: 99%

Consistency-Aware Recommendation for User-Generated Item List Continuation

Zhang

Liu

et al. 2020

Proceedings of the 13th International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

show abstract

“…Following the preprocessing setup in [3,23], we filter out items appearing in fewer than 5 lists. To study the impact of data density on recommendation performance, we also filter out users who have interacted with lists fewer than 5 times on Spotify and Zhihu, so that they are denser than the Goodreads dataset.…”

Section: Datasets: Goodreads Spotify and Zhihumentioning

confidence: 99%

“…The first factor is the inner product between user latent factors and list latent factors; the second factor is the sum of inner products between the user factor and item latent factors within the list. [3] is also a Bayesianbased pair-wise model designed for list recommendation. Inspired by paragraph2vec model in [18], it calculates the shifted positive pointwise mutual information (SPPMI) value between lists and items and uses this information to boost the ranking performance.…”

Section: Baselinesmentioning

confidence: 99%

“…2 For LIRE and EFM, we use code from the authors of EFM. 3 We implement AttList with Keras, and Adam [16] is applied as the optimizer. For AttList and baseline methods, all hyper-parameters are tuned on the validation dataset, where early stopping strategy is applied such that we terminate training if validation performance does not improve over 10 iterations.…”

Section: Reproducibilitymentioning

confidence: 99%

See 1 more Smart Citation

A Hierarchical Self-Attentive Model for Recommending User-Generated Item Lists

Wang

Niu

et al. 2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

User-generated item lists are a popular feature of many different platforms. Examples include lists of books on Goodreads, playlists on Spotify and YouTube, collections of images on Pinterest, and lists of answers on question-answer sites like Zhihu. Recommending item lists is critical for increasing user engagement and connecting users to new items, but many approaches are designed for the item-based recommendation, without careful consideration of the complex relationships between items and lists. Hence, in this paper, we propose a novel user-generated list recommendation model called AttList. Two unique features of AttList are careful modeling of (i) hierarchical user preference, which aggregates items to characterize the list that they belong to, and then aggregates these lists to estimate the user preference, naturally fitting into the hierarchical structure of item lists; and (ii) item and list consistency, through a novel self-attentive aggregation layer designed for capturing the consistency of neighboring items and lists to better model user preference. Through experiments over three real-world datasets reflecting different kinds of user-generated item lists, we find that AttList results in significant improvements in NDCG, Precision@k, and Recall@k versus a suite of state-ofthe-art baselines. Furthermore, all code and data are available at https://github.com/heyunh2015/AttList.

show abstract

“…Another angle from latent feature analysis is location-aware topic modelling, i.e., finding implicit locations for entity names [4,8], again through extensions or additions for location to existing topic models. Preprocessing is needed to incorporate the location semantics.…”

Section: Introductionmentioning

confidence: 99%

Poisson Factorization Models for Spatiotemporal Retrieval

Silva

Ahlers

2017

Proceedings of the 11th Workshop on Geographic Information Retrieval

View full text Add to dashboard Cite

New retrieval models promise deeper integration of multiple features and sources of information. The inclusion of thematic and location features in a joint factorization model allows location to be modeled as a first-class feature and can improve a range of tasks in geographic information retrieval and recommendation. In this position paper, we describe these factorization models and how they can be useful for corpus and user need understanding and further GIR use cases. We argue that using joint factorization models can be a powerful tool in the integration of complex features and relationships present in many GIR data sources and applications.

show abstract

Embedding Factorization Models for Jointly Recommending Items and User Generated Lists

Cited by 103 publications

References 31 publications

Consistency-Aware Recommendation for User-Generated Item List Continuation

Consistency-Aware Recommendation for User-Generated Item List Continuation

A Hierarchical Self-Attentive Model for Recommending User-Generated Item Lists

Poisson Factorization Models for Spatiotemporal Retrieval

Contact Info

Product

Resources

About