A Hybrid Two-Phase Recommendation for Group-Buying E-commerce Applications

The last two decades have witnessed an explosive growth of e-commerce applications. Existing online recommendation systems for e-commerce applications, particularly group-buying applications, suffer from scalability and data sparsity problems when confronted with exponentially increasing large-scale data. This leads to a poor recommendation effect of traditional collaborative filtering (CF) methods in group-buying applications. In order to address this challenge, this paper proposes a hybrid two-phase recommendation (HTPR) method which consists of offline preparation and online recommendation, combining clustering and collaborative filtering techniques. The user-item category tendency matrix is constructed after clustering items, and then users are clustered to facilitate personalized recommendation where items are generated by collaborative filtering technology. In addition, a parallelized strategy was developed to optimize the recommendation process. Extensive experiments on a real-world dataset were conducted by comparing HTPR with other three recommendation methods: traditional CF, user-clustering based CF, and item-clustering based CF. The experimental results show that the proposed HTPR method is effective and can improve the accuracy of online recommendation systems for group-buying applications.

Download Full-text

Large scale data processing in real world: From analytics to predictions

2014 14th International Conference on Advances in ICT for Emerging Regions (ICTer) ◽

10.1109/icter.2014.7083870 ◽

2014 ◽

Author(s):

Srinath Perera

Keyword(s):

Data Processing ◽

Real World ◽

Large Scale ◽

Large Scale Data ◽

Large Scale Data Processing ◽

Scale Data

Download Full-text

Internal and Contextual Attention Network for Cold-start Multi-channel Matching in Recommendation

Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2020/379 ◽

2020 ◽

Author(s):

Ruobing Xie ◽

Zhijie Qiu ◽

Jun Rao ◽

Yi Liu ◽

Bo Zhang ◽

...

Keyword(s):

Real World ◽

Large Scale ◽

Recommendation System ◽

Contextual Information ◽

Recommendation Systems ◽

Cold Start ◽

Personalized Recommendation ◽

Small Subset ◽

Multiple Channels ◽

Attention Network

Real-world integrated personalized recommendation systems usually deal with millions of heterogeneous items. It is extremely challenging to conduct full corpus retrieval with complicated models due to the tremendous computation costs. Hence, most large-scale recommendation systems consist of two modules: a multi-channel matching module to efficiently retrieve a small subset of candidates, and a ranking module for precise personalized recommendation. However, multi-channel matching usually suffers from cold-start problems when adding new channels or new data sources. To solve this issue, we propose a novel Internal and contextual attention network (ICAN), which highlights channel-specific contextual information and feature field interactions between multiple channels. In experiments, we conduct both offline and online evaluations with case studies on a real-world integrated recommendation system. The significant improvements confirm the effectiveness and robustness of ICAN, especially for cold-start channels. Currently, ICAN has been deployed on WeChat Top Stories used by millions of users. The source code can be obtained from https://github.com/zhijieqiu/ICAN.

Download Full-text

Large-Scale Data Storage Scheme in Blockchain Ledger Using IPFS and NoSQL

Large-Scale Data Streaming, Processing, and Blockchain Security - Advances in Information Security, Privacy, and Ethics ◽

10.4018/978-1-7998-3444-1.ch005 ◽

2021 ◽

pp. 91-116

Author(s):

Randhir Kumar ◽

Rakesh Tripathi

Keyword(s):

Data Storage ◽

Real World ◽

Large Scale ◽

Original Data ◽

Size Reduction ◽

Storage Model ◽

Large Scale Data ◽

Efficient Storage ◽

Storage Scheme ◽

Scale Data

The future applications of blockchain are expected to serve millions of users. To provide variety of services to the users, using underlying technology has to consider large-scale storage and assessment behind the scene. Most of the current applications of blockchain are working either on simulators or via small blockchain network. However, the storage issue in the real world is unpredictable. To address the issue of large-scale data storage, the authors have introduced the data storage scheme in blockchain (DSSB). The storage model executes behind the blockchain ledger to store large-scale data. In DSSB, they have used hybrid storage model using IPFS and MongoDB(NoSQL) in order to provide efficient storage for large-scale data in blockchain. In this storage model, they have maintained the content-addressed hash of the transactions on blockchain network to ensure provenance. In DSSB, they are storing the original data (large-scale data) into MongoDB and IPFS. The DSSB model not only provides efficient storage of large-scale data but also provides storage size reduction of blockchain ledger.

Download Full-text

Recommendation Framework Combining User Interests with Fashion Trends in Apparel Online Shopping

Applied Sciences ◽

10.3390/app9132634 ◽

2019 ◽

Vol 9 (13) ◽

pp. 2634 ◽

Cited By ~ 1

Author(s):

Ok ◽

Lee ◽

Kim

Keyword(s):

Random Walk ◽

Collaborative Filtering ◽

Real World ◽

Large Scale ◽

Online Shopping ◽

Experimental Studies ◽

Empirical Evaluation ◽

Personalized Recommendation ◽

User Interests ◽

Sparsity Problem

Although fashion-related products account for most of the online shopping categories, it becomes more difficult for users to search and find products matching their taste and needs as the number of items available online increases explosively. Personalized recommendation of items is the best method for both reducing user effort on searching for items and expanding sales opportunity for sellers. Unfortunately, experimental studies and research on fashion item recommendation for online shopping users are lacking. In this paper, we propose a novel recommendation framework suitable for online apparel items. To overcome the rating sparsity problem of online apparel datasets, we derive implicit ratings from user log data and generate predicted ratings for item clusters by user-based collaborative filtering. The ratings are combined with a network constructed by an item click trend, which serves as a personalized recommendation through a random walk. An empirical evaluation on a large-scale real-world dataset obtained from an apparel retailer demonstrates the effectiveness of our method.

Download Full-text

Collaborative filtering recommendation algorithm based on variational inference

International Journal of Crowd Science ◽

10.1108/ijcs-10-2019-0030 ◽

2020 ◽

Vol 4 (1) ◽

pp. 31-44

Author(s):

Kai Zheng ◽

Xianjun Yang ◽

Yilei Wang ◽

Yingjie Wu ◽

Xianghan Zheng

Keyword(s):

Probability Distribution ◽

Collaborative Filtering ◽

Large Scale ◽

Variational Inference ◽

Data Sets ◽

Rating Data ◽

Content Type ◽

Basic Model ◽

Large Scale Data ◽

Scale Data

Purpose The purpose of this paper is to alleviate the problem of poor robustness and over-fitting caused by large-scale data in collaborative filtering recommendation algorithms. Design/methodology/approach Interpreting user behavior from the probabilistic perspective of hidden variables is helpful to improve robustness and over-fitting problems. Constructing a recommendation network by variational inference can effectively solve the complex distribution calculation in the probabilistic recommendation model. Based on the aforementioned analysis, this paper uses variational auto-encoder to construct a generating network, which can restore user-rating data to solve the problem of poor robustness and over-fitting caused by large-scale data. Meanwhile, for the existing KL-vanishing problem in the variational inference deep learning model, this paper optimizes the model by the KL annealing and Free Bits methods. Findings The effect of the basic model is considerably improved after using the KL annealing or Free Bits method to solve KL vanishing. The proposed models evidently perform worse than competitors on small data sets, such as MovieLens 1 M. By contrast, they have better effects on large data sets such as MovieLens 10 M and MovieLens 20 M. Originality/value This paper presents the usage of the variational inference model for collaborative filtering recommendation and introduces the KL annealing and Free Bits methods to improve the basic model effect. Because the variational inference training denotes the probability distribution of the hidden vector, the problem of poor robustness and overfitting is alleviated. When the amount of data is relatively large in the actual application scenario, the probability distribution of the fitted actual data can better represent the user and the item. Therefore, using variational inference for collaborative filtering recommendation is of practical value.

Download Full-text