Open AccessTravel Attractions Recommendation based on Max-negative the Gated Recurrent Unit trajectory mining Representation
Author(s) -
Shunyao Zhang,
Phatpicha Yochum,
Chenzhong Bin,
Liang Chang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1437/1/012047
Subject(s) - computer science , pairwise comparison , recurrent neural network , recommender system , trajectory , representation (politics) , recall , artificial intelligence , tourism , sequence (biology) , machine learning , data mining , artificial neural network , geography , psychology , physics , archaeology , astronomy , politics , biology , political science , law , cognitive psychology , genetics
Although the traditional recommendation algorithm has achieved good results in the field of travel recommendation, due to the lack of data, the cold start and data sparseness problems and the neglect of the semantic problems hidden in the travel track, the low recommendation accuracy remains unresolved. Recently, the RNN model performed very well in recommending system sequence learning. We use RNN to model the travel sequence in the travel recommendation. In the pairwise, we can achieve a more accurate recommendation by effectively processing the negative samples and then training to generate a smaller loss function. Our method (Max-GRU) is optimized by adding additional negative sample and finding Max-negative on the Gated Recurrent Unit trajectory mining Representation Model. On the Shanghai tourism data and Guilin tourism data, both MRR@10 and RECALL@10 have been significantly improved compared to the use of the RNN model and baselines.