Semantic-Based Service Recommendation Method on MapReduce Using User-Generated Feedback
Service recommender systems provide same recommendations to different users based on ratings and rankings only, without considering the preference of an individual user. These ratings are based on the single criteria of a service ignoring its multiple asp
- PDF / 397,432 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 101 Downloads / 145 Views
Abstract Service recommender systems provide same recommendations to different users based on ratings and rankings only, without considering the preference of an individual user. These ratings are based on the single criteria of a service ignoring its multiple aspects. Big data also affects these recommender systems with issues like scalability and inefficiency. Proposed system enhances existing recommendations systems and generates recommendations based on the categorical preferences of the present user by matching them with the feedback/comments of the past users. System semantically analyzes the users feedback and distinguishes it into positive and negative preferences to eliminate the unnecessary reviews of the users which boosts the system accuracy. Approximate and exact similarity between the preferences of present and past users is computed and thus the recommendations are generated using SBSR algorithm. To improve the performance, i.e., scalability and efficiency in big data environment, SBSR is ported on distributed computing platform, Hadoop.
⋅
⋅
Keywords Service recommender systems Big data Semantic analysis Jaccard co-efficient Cosine similarity Hadoop MapReduce
⋅
⋅
⋅
⋅
1 Introduction The overabundance of data on the web drives away the focus of users, landing them to surf for the data that they were not searching for initially. Information filtering systems are used to overcome these problems and to eliminate the unnecessary information before presenting it to the user. The subclass of these systems, called as Ruchita Tatiya (✉) ⋅ Archana Vaidya Gokhale Education Society’s R. H. Sapat College of Engineering, Management Studies and Research, P T A Kulkarni Vidyanagar, Nashik, Maharashtra, India e-mail: [email protected] Archana Vaidya e-mail: [email protected] © Springer Science+Business Media Singapore 2017 S.C. Satapathy et al. (eds.), Proceedings of the International Conference on Data Engineering and Communication Technology, Advances in Intelligent Systems and Computing 468, DOI 10.1007/978-981-10-1675-2_15
131
132
Ruchita Tatiya and Archana Vaidya
recommendation systems assist by predicting the services or items that the user would like. Service recommender systems [1] provide appropriate recommendations and have become popular in variety of practical applications like recommending the users about hotels, books, movies, music, travel, etc. [2, 3]. The enlarged number of Internet users is contributing to immense amount of data everyday [4]. Such immense data, known as Big data, is not only difficult to capture and store but also managing, processing, and analysing such data with the available current technology within the tolerable speed and time is a difficult task.
1.1
Motivation
The service recommender systems present the same ratings and rankings of the services to the different users and also provide the same recommendations to them without considering the user’s personal likings and taste [1]. Also many recommendation systems provides single-criteria ratings i.e. just
Data Loading...
