AN ANALYSIS OF WORLD WIDE WEB (WWW) IS AN IMPORTANT SOURCE FOR RELEASING, BROADCASTING AND OBTAINING

ABSTRACT

Vertical search engines are attracting professionals and business users searching for niche topics and are providing them with the satisfactory user experience. A powerful vertical search engine can efficiently identify domain relevant resources, extracting critical information, and adapt the search results to specific user’s needs. To find patterns, trends, business characteristics and relevant interesting content from the ocean of web data is a very difficult task. To capture the freelancing and remote work opportunity from the web is an interesting area of research. It was observed that the professionals using general search engines were unable to find relevant business/work-related information. This happened as they were not trained in their use and the general search engines were not designed as business tools. This resulted in a low satisfaction for business user rating of forty percent for general search engines. As there is a huge demand for freelancing and remote work, so information about it is available on many forums, company sites, classifieds, blogs and marketplaces. Timely discovery of freelancing work is very crucial as it loses value with time. Existing platforms which provide information regarding freelancing work are dependent on their own databases and as they do not crawl the data across the web so these websites are not indexing and covering real time global freelancing work information. This system captures global freelancing real time data accurately and timely. The significant contributions of this thesis is the new approach of search framework which facilitates more accurate and timely discovery of freelancing and remote work in IT field. The machine learning approach is developed to find relevancy of web pages to the freelancing work in IT field. The performance evaluation of proposed algorithm is discussed using standard metrics precision and recall. The effectiveness of this approach is evaluated using different size of training and testing datasets

FULL TEXT