Description

Online product configurations are prevailing tools in e-commerce industry to elicit customer needs. Yet current online configurations require customers to specify the choices of each product attribute, which poses a great challenge for customers with no background knowledge.

Hence, this project aims to develop a new online product configuration approach which enables customers to configure a product just by specifying their functional requirements, instead of the detailed design parameters.

For example,  when users input the keywords like “a large Screen Size laptop“, our approach could map the associated features and and give high quality  recommendations (Fig.1).

Fig. 1  Our approach to predicting the attributes based on the input “A large Screen Size laptop”

We are now developing auto tools to collect online user reviews as corpus, and training deep neural networks  to predict and display the items that a user would like to purchase with top accuracy.

Team members

Dr. WANG YUE,  Mr. Raymond ZHAO Wenlong (email: raymondzhao@hsu.edu.hk)

Presentation

• Dataset  (3 July 2018)
• Algorithm: word_embeddings_based_mapping  (15 August 2018 )
• Algorithm: SWEM Algs (28 August 2018)
• Algorithm: SVM based on word-embedding ( 11 September 2018)
• Convolutional Neural Networks for Sentence Classification (26 September 2018)
• The experiments based on CNN (9 October 2018)
• Algorithm: Recurrent Neural Networks (23 October 2018)
• Algorithm: LSTM (6 November 2018)
• Hier-attention Network (20 November 2018)
• Top-N Sort Alg (11 December 2018)