Meta Data based Conceptualization and Temporal Semantics in Hybrid Recommender

Meta Data based Conceptualization and Temporal Semantics in Hybrid Recommender

M. Venu Gopalachari (Chaitanya Bharathi Institute of Technology, Hyderabad, India) and Porika Sammulal (JNTUH College of Engineering Jagtial, Karimnagar, India)
Copyright: © 2017 |Pages: 18
DOI: 10.4018/IJRSDA.2017100104
OnDemand PDF Download:
No Current Special Offers


Modern recommender systems target the satisfaction of the end user through the personalization techniques that collects the history of the user's navigation. But the sole dependency on the user profile by means of navigation history alone cannot promise the quality of recommendations because of the lack of semantics. Though the literature provides many techniques to conceptualize the process they lead to high computational complexity due to considering the content data as input information. In this paper a hybrid recommender framework is developed that considers Meta data based conceptual semantics and the temporal patterns on top of the usage history. This framework also includes an online process that identifies the conceptual drift of the usage dynamically. The experimental results shown the effectiveness of the proposed framework when compared to the existing modern recommenders also indicate that the proposed model can resolve a cold start problem yet accurate suggestions reducing computational complexity.
Article Preview


World Wide Web bombards the information with continuous rapid advancements in the services that affects the human in all aspects of life. But the enormous information repository also paves the information overload problem to the user limiting to the access to the most relevant information of the needy topic. Web usage mining (Liu, 2011, Nguyen, 2012) is the devoted research field that generates user profiles by means of the web log over the past decade. The web log records the details of the navigation of a user that consists of IP address, Session id, URL, Title of the page and Protocol used etc. The user profiles generated are used to assist the user, as well as to improve the market of the vendor by means of recommender, target marketing (Cheng, 2012). But the traditional approaches in web mining did not able to consider the semantics in their processing at its full extent leads to limit the quality of recommendations to the user (Rios, 2008). However, the semantic web usage mining, which includes the domain knowledge, is annotating ontological entity to each web page makes better understanding the usage profiles. The domain ontology can be generated from free text documents in many ways, but integrating these methods with web mining would make the web application complex (Rios, 2008, Salin, 2009).

Recommender Systems expanded their scope for many topics such as tourism (Cheng, 2012), movies (Winto, 2010), songs (Lee, 2010), web search (Ha, 2002), books (Crespo, 2011) etc. Content based and collaborative filtering are the two major approaches in recommending web pages having their own merits and demerits. However, hybrid recommenders could able to grab the optimal in minimizing the disadvantages of these approaches by combining the features from both the kinds. In general, traditional approaches of recommender systems suffer from the issues such as cold start and data sparsity (Martin, 2014). Data sparsity in ratings of the products considerably affects the quality of recommendations, for example, in an e-commerce website the user will not get recommendations of a product either if that user does not have history about that product or if the product did not have support of enough number of accesses. Cold start problem occurs when trying to suggest the new user who does not have much access log so far. These issues arise with lack of semantics and due of conceptualization in recommendation process. In order to incorporate semantics into the recommender process, it needs the construction of knowledge from the domain of the corresponding web application. In many recommenders the web content data used to be treated as domain that leads to the high computational complexity of the web service. An alternative is much needed to ensure the quality of recommendations and also leads to the light weight application with respect to the computational complexity.

Complete Article List

Search this Journal:
Volume 8: 1 Issue (2022): Forthcoming, Available for Pre-Order
Volume 7: 4 Issues (2021): 1 Released, 3 Forthcoming
Volume 6: 3 Issues (2019)
Volume 5: 4 Issues (2018)
Volume 4: 4 Issues (2017)
Volume 3: 4 Issues (2016)
Volume 2: 2 Issues (2015)
Volume 1: 2 Issues (2014)
View Complete Journal Contents Listing