China 
Africa 
Explore chapters and articles related to this topic
A new preference-based model to solve the cold start problem in a recommender system
Published in Xiaoling Jia, Feng Wu, Electromechanical Control Technology and Transportation, 2017
Kun Liu, Wei Liu, Xiaoyun Chen
In this paper, we presented a new recommender model to alleviate the user cold start problems of RS. We first analyzed the shortages of traditional CF methods. In our proposed model, we first classified all users into groups on the basis of their preferences. Then, we obtained the nearest neighbors of cold users using a similarity measure. Finally, popular items of cold users’ neighbors were recommended. For demonstrating the effectiveness of our metric, several experiments were conducted compared to different methods on two widely used data sets. In this paper, we also illustrated why we chose these methods for the comparison. From the experimental results in the end, we can see that our model shows the best performance and it improves more than 100% compared to CF methods. In our method, we alleviate the cold start problem of users with few ratings. When users have no rating, we call this phenomenon the pure cold start problem. In future, we will focus on the pure cold start problem.
The evolution of recommender systems: From the beginning to the Big Data era
Published in Matthias Dehmer, Frank Emmert-Streib, Frontiers in Data Science, 2017
Beatrice Paoli, Monika Laner, Beat Tödtli, Jouri Semenov
Although the field of recommender systems has undergone a significant development, certain problems are still challenging. Great concern must be given to aspects such as the quality of the recommendations, the sparsity of the data, scalability, and how to cope with the so-called cold start problem [19,20] that deals with users and items with limited or no previous information. The recommendations need to attract the user’s interest and be useful. The items that a user has already purchased should not be recommended again, as well as the items that are not matching the user’s taste. By providing high-quality recommendations, the user’s trust to the recommender system is augmented, and he or she is likely to continue using it.
BERT- and FastText-Based Research Paper Recommender System
Published in Pallavi Vijay Chavan, Parikshit N Mahalle, Ramchandra Mangrulkar, Idongesit Williams, Data Science, 2022
Nemil Shah, Yash Goda, Naitik Rathod, Vatsal Khandor, Pankaj Kulkarni, Ramchandra Mangrulkar
However, like all other recommender systems, this recommender system also suffers from the problem of ‘cold start’. The cold start problem as in Volkovs et al., 2017 [32] occurs when new users or new items are added to the recommendation systems. During the addition of new items to the recommender systems, the new items must undergo the same training process to alleviate this problem. If a different training process is used, the results might differ from the original results leading to discrepancies. Also, newer data must be regularly added to the recommender systems so that the latest recommendations are readily available to the users.
Memory-augmented meta-learning on meta-path for fast adaptation cold-start recommendation
Published in Connection Science, 2022
Tianyuan Li, Xin Su, Wei Liu, Wei Liang, Meng-Yen Hsieh, Zhuhui Chen, XuChong Liu, Hong Zhang
There may be a different proportion of new users and new items in the recommendation system, and interaction between these users and items is sparse. As a result, personalised recommendation for new users is challenging due to the cold-start problem. Deep learning (Liang, Xie, et al., 2020) has achieved great results in a variety of artificial intelligence domains. However, in order to obtain significant generalisation, a large number of examples must be trained. Deep learning becomes ineffective when used in a cold-start recommendation scenario with sparse user–item interactions. Data augmentation at the data level or the provision of auxiliary data (Zhu et al., 2019) are the most typical solutions to cold-start recommendations. There are also some methods involving the high-level representation of the data, such as capturing the rich heterogeneous data (Chang et al., 2021) of the items and the users, using the data representation of the heterogeneous information network, in addition to considering the basic characteristics of the data. Alternatively, a semantic network can be built using a knowledge graph, in which nodes represent entities and edges to reflect various semantic relationships between items. There are also cross-domain recommendations based on mapping of neighbour user attributes, or recommendations based on mining friend lists on social networks. These techniques rely largely on data and have a number of drawbacks.
iTourSPOT: a context-aware framework for next POI recommendation in location-based social networks
Published in International Journal of Digital Earth, 2022
Lin Wan, Han Wang, Yuming Hong, Ran Li, Wei Chen, Zhou Huang
The personalized top-N recommendation can seem like synthesizing rating prediction and ranking tasks. Regarding the rating prediction stage, the inherent cold-start issue always poses great challenges to the recommendation accuracy. At the same time, this is even more serious in POI recommender systems as there are many tourists without any footprints. To relieve this problem, we introduced collective wisdom for the preference prediction of such cold-boot travelers utilizing existing tourists' characteristics. Specifically, based on the assumption that cold-boot tourists will plan their first trips more likely than people who just experienced the interest in travel, we randomly selected α of tourists with less than two footprints. We used the mean value of their preference embedding vectors to represent the preferences of a specific cold-boot tourist. Due to the sequential recommendation scenario, we primarily calculate historical ratings for each POI and leverage the results to evaluate the current rating scores of each user. As shown in Equation (14), the ratings of historical traveler u in POI poi up to date t, depends on the temporal information and check-in times. where is an indication function, it equals 1 if , equals 0 otherwise.
An intelligent location recommender system utilising multi-agent induced cognitive behavioural model
Published in Enterprise Information Systems, 2021
Logesh Ravi, Malathi Devarajan, Vijayakumar V, Arun Kumar Sangaiah, Lipo Wang, Sasikumar A, V Subramaniyaswamy
Recommender system suggests items on the basis of active user’s preferences either explicitly or implicitly. When new user or item enters the recommender system, the inability to predict and suggest relevant information to the new user without prior experience about the active user leads to a cold-start issues. The collaborative filtering recommender system cannot facilitate cold-start problem, since it has no earlier knowledge or experience about the new user or item such as item ratings, user preferences and interests. Though Content-based filtering recommender system produce recommendations based on item description, they are likely to accomplish lower precision. Moreover, cold-start problem is related to density and diversity of information and it is categorised into recommendations for new users, new items and new items for new users.