Article Preview
Top1. Introduction
Search engines are used to retrieve the relevant content from the World Wide Web (WWW), which is having more than a trillion of web pages. Conventional search engines such as Google, Yahoo search and Bing provide the search results based on the keywords mapped into the web content. But, it is derived from the fact that web content developers use these keywords deliberately to advertise their services and attract the users to visit their webpages. As a result, irrelevant content is also retrieved (because of the syntactic nature of the search process) in the search results. i.e. results are not meaningful and inappropriate to the user. Also, users are required to spend more time to identify the relevant results from the vast Uniform Resource Locator (URL). To overcome the abovementioned difficulties and optimize the traditional search engine performance, semantic search engines such as Swoogle and DuckDuckGo were introduced (Singh & Sharan, 2013). These engines achieve relevant and meaningful results based on the concepts and their relationships. Besides, it reduces the search time by retrieving the relevant documents in the first set of URLs returned (Roush, 2004).
The limitations of traditional search tools can be overcome by semantic engines with the use of semantic knowledge by using ontologies. Semantic mechanisms are effectively used in information retrieval and integration, service discovery, question answering, information management, annotation, recommendation and ranking. Semantic engines require the web content to be annotated earlier for improved search accuracy. In the past, it was the developer’s task to perform the annotation manually which usually prone to errors resulting from a number of factors such as user familiarity, personal motivation or the complexity of ontologies, which might negatively affect the annotation procedure (Rodríguez-García et al., 2014). Moreover, the downside of manual annotation applied in the search process is that it is time- consuming and expensive. Hence web content should require semantic annotation using Natural Language Processing (NLP) techniques.
In this paper, semantic search engine is implemented for crawling SLA templates from the World Wide Web and annotating these templates for meaningful search and retrieval. Currently, SLA templates are placed in the provider website which is very difficult for the consumer to search and compare the similar services offered by diversified service providers. Also, SLA templates are descriptive and written in the natural language (English), and consumers always do a tough task in searching the right provider (Van Surksum, 2013). Traditional search engines perform syntax based search in the vast web, resulting in wrong and inappropriate results which are not suitable for the consumer. In this paper, a semantically-enhanced search engine platform is developed which will assist in the process of discovering the SLAs in cloud that best match the user needs.
The proposed search platform working criteria is as follows. Consumer poses a query in the search tab with the brief description of their cloud service requirements and metrics in natural language. Then, the service requirements are mapped with the annotated SLAs residing in the repository, and retrieval is made with the matched documents. Based on the indexing, query results are presented in the form of SLA documents. If the results are not sufficient enough, dynamic search enables the user to fill up the specific requirement constraints and search results are produced using keyword based approach. In this way, this platform implements the semantic annotation based search along with the traditional keyword based search approaches.