In the context of the Internet, a search engine can be defined as a software program designed to help one access information, documents, and other content on the World Wide Web. The adoption and growth of the Internet in the last decade has been unprecedented. The World Wide Web has always been applauded for its simplicity and ease of use. This is evident looking at the extent of the knowledge one requires to build a Web page. The flexible nature of the Internet has enabled the rapid growth and adoption of it, making it hard to search for relevant information on the Web. The number of Web pages has been increasing at an astronomical pace, from around 2 million registered domains in 1995 to 233 million registered domains in 2004 (Consortium, 2004). The Internet, considered a distributed database of information, has the CRUD (create, retrieve, update, and delete) rule applied to it. While the Internet has been effective at creating, updating, and deleting content, it has considerably lacked in enabling the retrieval of relevant information. After all, there is no point in having a Web page that has little or no visibility on the Web. Since the 1990s when the first search program was released, we have come a long way in terms of searching for information. Although we are currently witnessing a tremendous growth in search engine technology, the growth of the Internet has overtaken it, leading to a state in which the existing search engine technology is falling short. When we apply the metrics of relevance, rigor, efficiency, and effectiveness to the search domain, it becomes very clear that we have progressed on the rigor and efficiency metrics by utilizing abundant computing power to produce faster searches with a lot of information. Rigor and efficiency are evident in the large number of indexed pages by the leading search engines (Barroso, Dean, & Holzle, 2003). However, more research needs to be done to address the relevance and effectiveness metrics. Users typically type in two to three keywords when searching, only to end up with a search result having thousands of Web pages! This has made it increasingly hard to effectively find any useful, relevant information. Search engines face a number of challenges today requiring them to perform rigorous searches with relevant results efficiently so that they are effective. These challenges include the following (“Search Engines,” 2004). 1. The Web is growing at a much faster rate than any present search engine technology can index. 2. Web pages are updated frequently, forcing search engines to revisit them periodically. 3. Dynamically generated Web sites may be slow or difficult to index, or may result in excessive results from a single Web site. 4. Many dynamically generated Web sites are not able to be indexed by search engines. 5. The commercial interests of a search engine can interfere with the order of relevant results the search engine shows. 6. Content that is behind a firewall or that is password protected is not accessible to search engines (such as those found in several digital libraries).1 7. Some Web sites have started using tricks such as spamdexing and cloaking to manipulate search engines to display them as the top results for a set of keywords. This can make the search results polluted, with more relevant links being pushed down in the result list. This is a result of the popularity of Web searches and the business potential search engines can generate today. 8. Search engines index all the content of the Web without any bounds on the sensitivity of information. This has raised a few security and privacy flags. With the above background and challenges in mind, we lay out the article as follows. In the next section, we begin with a discussion of search engine evolution. To facilitate the examination and discussion of the search engine development’s progress, we break down this discussion into the three generations of search engines. Figure 1 depicts this evolution pictorially and highlights the need for better search engine technologies. Next, we present a brief discussion on the contemporary state of search engine technology and various types of content searches available today. With this background, the next section documents various concerns about existing search engines setting the stage for better search engine technology. These concerns include information overload, relevance, representation, and categorization. Finally, we briefly address the research efforts under way to alleviate these concerns and then present our conclusion.
MACHINE LEARNING BASED SEARCH ENGINE WITH CRAWLING, INDEXING AND RANKING
