Practical Lessons for Job Recommendations in the Cold-Start Scenario

Lian, Jianxun; Zhang, Fuzheng; Hou, M.; Wang, Hongwei; Xie, Xing; Sun, Guoqiang

doi:10.1145/3124791.3124794

Cited by 23 publications

(19 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the large number of possible category variables, the converted one-hot features are usually of high dimensionality but sparse [11], and simply using raw features rarely provides optimal results. On this occasion, the interactions among different features act as the winning formula for a wide range of data mining tasks [7], [12], [13]. The interactions among multiple raw features are usually termed as cross features [7] (a.k.a.…”

Section: Introductionmentioning

confidence: 99%

“…In short, there are two major trends of improvements over the plain FM. One is to make the model "deep" with multi-layer network structures in order to exhaustively extract useful information from feature interactions, e.g., the residual network in DeepCross [7], the pairwise product layer in PNN [21], and the compressed interaction network in xDeepFM [12]. The other is to make the model "wide" by considering multiple feature interactions in varied domains (usually coupled with "deep" structures), e.g., separately modelling user logs and texts with CoFM [15], or fusing shallow low-order output with dense high-order output via Wide&Deep [18], DeepFM [20] and xDeepFM [19].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Sequence-Aware Factorization Machines for Temporal Predictive Analytics

Chen

Yin

Nguyen

et al. 2020

2020 IEEE 36th International Conference on Data Engineering (ICDE)

View full text Add to dashboard Cite

In various web applications like targeted advertising and recommender systems, the available categorical features (e.g., product type) are often of great importance but sparse. As a widely adopted solution, models based on Factorization Machines (FMs) are capable of modelling high-order interactions among features for effective sparse predictive analytics. As the volume of web-scale data grows exponentially over time, sparse predictive analytics inevitably involves dynamic and sequential features. However, existing FM-based models assume no temporal orders in the data, and are unable to capture the sequential dependencies or patterns within the dynamic features, impeding the performance and adaptivity of these methods. Hence, in this paper, we propose a novel Sequence-Aware Factorization Machine (SeqFM) for temporal predictive analytics, which models feature interactions by fully investigating the effect of sequential dependencies. As static features (e.g., user gender) and dynamic features (e.g., user interacted items) express different semantics, we innovatively devise a multi-view self-attention scheme that separately models the effect of static features, dynamic features and the mutual interactions between static and dynamic features in three different views. In SeqFM, we further map the learned representations of feature interactions to the desired output with a shared residual network. To showcase the versatility and generalizability of SeqFM, we test SeqFM in three popular application scenarios for FM-based models, namely ranking, classification and regression tasks. Extensive experimental results on six large-scale datasets demonstrate the superior effectiveness and efficiency of SeqFM.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Sequence-Aware Factorization Machines for Temporal Predictive Analytics

Chen

Yin

Nguyen

et al. 2020

2020 IEEE 36th International Conference on Data Engineering (ICDE)

View full text Add to dashboard Cite

show abstract

“…Due to the large number of possible category variables, the converted one-hot features are usually of high dimensionality but sparse [69], and simply using raw features rarely provides optimal results. On this occasion, the interactions among different features act as the winning formula for a wide range of data mining tasks [95][96][97] information for user profiling with cross features, such as ( junior, engineer) and (senior, lecturer). To avoid the high cost of task-specific manual feature engineering, factorization machines (FMs) [20] are proposed to embed raw features into a latent space, and model the interactions among features via the inner product of their embedding vectors.…”

Section: Predictive Analytics Under Sparsitymentioning

confidence: 99%

“…However, these In short, there are two major trends of improvements over the plain FM. One is to make the model "deep" with multi-layer network structures in order to exhaustively extract useful information from feature interactions, e.g., the residual network in DeepCross [95], the pairwise product layer in PNN [99], and the compressed interaction network in xDeepFM [96]. The other is to make the model "wide" by considering multiple feature interactions in varied domains (usually coupled with "deep" structures), e.g., separately modelling user logs and texts with CoFM [84], or fusing shallow low-order output with dense high-order output via Wide&Deep [101], DeepFM [100] and xDeepFM [70].…”

Section: Evolution Of Factorization Machinesmentioning

confidence: 99%

See 1 more Smart Citation

Sequence modelling for e-commerce

Chen¹

View full text Add to dashboard Cite

With the proliferation of electronic commerce (e-commerce), the data generated by both customers and service providers can accumulate at a fast rate. As such, analyzing the rich but subtle patterns within the e-commerce data offers a prominent opportunity of refining user experience and increasing business revenue. Due to the high velocity of e-commerce data, sequence modelling plays a pivotal role in delivering timely predictive analytics and recommendations. Based on the granularity of data, sequence modelling for e-commerce is mainly conducted at two levels, namely macro-level modelling and micro-level modelling. When researching on e-commerce data, macro-level sequence modelling aims During my 3.5-year PhD candidature at The University of Queensland (UQ), I have come across a lot of wonderful people that offered me tremendous help, support, and advice. I would give my supervisors, Dr. Hongzhi Yin and Prof. Xue Li a big thanks for their continuous academic guidance, scholarship support, and patient advisory along my way of research. Dr. Hongzhi Yin has set me a great example of a young scholar with superb visions, capabilities and humbleness, and I have been so lucky to join his research team and contribute to the research goal that we believe will greatly pay back to the society. He is my constant reminder that I still have a lot to learn about what we are working on. Prof. Xue Li is the guide that opened the door of conducting research at UQ for me. Despite his seniority, I can always feel his endless energy and passion for research whenever we have a chat, while his kindness and insistence are the benchmarks I need to bear in mind throughout my lifetime. Apart from my advisory team, I would like to thank many other academic and professional staff at UQ and other institutes, especially A

show abstract