PNKASR — A Positive Negative Keyword-Aware Service Recommendation methodology for big data applications

A recommendation system has been appearing as an important method for giving suitable suggestions to users. In the latest years, the most of clients, administrations and network connected resources had developed quickly, yielding the huge data research problem for the administration recommendation system. The Recommendation system applies learning, revelation systems to the issue of making customized suggestions for data, items or administrations amid the live communication. The colossal development in the measure of accessible data and the increase in guests to the sites in the late years represents some key difficulties for recommender system. Today there is a major assortment of various methodologies and calculations of information separating and suggestion. A recommendation process is applied on items, items can be books, trains, music, video, hotels and things on which persons have many numbers of choices are available. These available many numbers of choices are anticipated using two essential methodologies first content-based method which include attributes of an item and second collaborative filtering method which concentrate on past behavior of clients. Additionally, the greater part of present service recommender system show the equal rankings and ratings of services to various clients by not considering different clients requirements and therefore not able to fulfill clients actualize needs. For this method, we introduce a Positive Negative Keyword-Aware Service Recommendation strategy, abbreviated PNKASR, to minimize the raised difficulties and for accuracy point of view it also concentrate on negative review of the previous user. It aims to exhibit a customized service suggestion list and prescribing the most proper services to the clients adequately. To upgrade its efficiency and scalability in large size data environment, PNKASR is implemented on open-source framework Hadoop, a mostly used platform by taking MapReduce parallel processing tool.