International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022
p-ISSN: 2395-0072
www.irjet.net
Tourism Based Hybrid Recommendation System Brian Davis Ukken1, Rhutuja Salunke2 1,2
Student, Department of Computer Engineering, Pillai HOC College of Engineering and Technology, Rasayani, Maharashtra, INDIA
---------------------------------------------------------------------***-------------------------------------------------------------------- New item problem: Not enough interactions with Abstract - Recommender system focuses on recommending the users.
appropriate packages to users based on their preferences and tastes by analyzing different reviews alongside the given ratings. However, users do not rate enough packages to make the collaborating filtering algorithm, leading to a cold start problem. The following solutions are included in this project to overcome the said problems: 1. the project combines ContentBased filtering along with the above algorithm to solve coldstart user problems and get higher accuracy and better precision. 2. Factors like hotels, destination, cost, and preferences are considered as a piece of additional information for a more personalised recommendation. 3. This model is also integrated with Aspect Based Sentiment Analysis to give better and more accurate results. Multiple tests were conducted on several datasets like TripAdvisor, and the results revealed that the proposed hybrid framework is competitive and superior to conventional approaches. The project also includes various elements such as a Semi-supervised Clustering algorithm which classifies the facets of the given vocabulary into nine pre-defined groups known as tour aspects. The ratings and reviews are stored in our database, which helps and enhances the desired solutions. Hence, the hybrid approach increases the efficiency of the results. collaborative filtering, recommender system, content-based filtering, tour package, predictions, ratings, hybrid model, sentiment analysis
1. INTRODUCTION
The primary methods included to develop this model are CF and CB. CB [1] will suggest articles to the users which was preferred by them earlier, and CF [2], will advise things that other individual, identical in tastes, liked before [3]. Individually, each method has its own drawbacks, which include:
Impact Factor value: 7.529
Cold start problem, etc.
several
contributions
and
A hybrid method to build a better and more efficient model that makes exchange between coverage and accuracy.
A proposition that resolves the cold start problem.
Better efficiency and precision are due to the integration of the two different algorithms.
The remaining paper contains the following sections: The next section defines the problem statement. Literature review work is represented in the third section. The fourth section represents our model along with the algorithms. The fifth section reveals the end results of our model. The paper concludes in Section 7 with the possible future of this work.
Limited content analysis: Not sufficient content for the algorithm to give the desired results
|
Our project includes improvements:
The recent boom in the Internet and the broad scope of Ecommerce has led to the flow of tremendous information. There is a massive demand for creating very sophisticated and superior systems that can process this massive surge of data. Furthermore, it should aid the users to make choices by proposing products, services, items, etc. that are similar to their respective preferences. Recommendation systems are a promising alternative to deal with these demands and issues.
© 2022, IRJET
New user problem: The newer users cannot be recommended items since the model isn’t aware of the user’s tastes.
A hybrid recommender system will be implemented where the CB and CF methods will be integrated to anticipate better predictions and get better of the impediments of each approach. This research proposes a hybrid system based on users' information, ratings, and written reviews. This mainly combines collaborative filtering (CF) and content-based (CB) into the recommendation system. Here, we alternate the hybrid direction, where the model will be enhanced by incorporating aspect-based sentiment analysis. This results in the cold start problem being eliminated, which in turn would give high-performance recommendation results. The details of each tour will be stored in an SQL database, which contains general tour information, its assessment and ratings, and an item-based CF technique to predict the unrated features of tours. The lexicon-based approach determines the sentiment orientation towards tour features, and semi-supervised clustering builds the vocabulary of tour aspects. For tour searching, we use context-based information to give a more accurate recommendation.
Key Words:
|
ISO 9001:2008 Certified Journal
|
Page 870