Beating Google: Development of the Novel Web Page Ranking Algorithm WordRank in a Search Engine

Overview

Objectives/Goals To develop a new algorithm that will more effectively rank web pages so that a search engine can return more meaningful results, especially for specific queries. Methods/Materials A search engine with a crawler and a lookup function was created in Python. A novel page ranking algorithm named WordRank, which ranks pages based on user inputted keywords and a page's content, was designed and implemented. For a comparative study, the two well-known ranking algorithms HITS and PageRank were also implemented in the search engine. The results from each of these algorithms and different combinations of algorithms were compared using the number of relevant links and quality scores. Results A comparative analysis of PageRank, HITS, and WordRank was performed. It was observed that the search results from HITS were of lower quality than those from PageRank. PageRank was thus used in conjunction with WordRank. When a combination strategy was applied to the PageRank and WordRank scores, it was determined that the results from a PageRank to WordRank ratio of 50:50 were generally more relevant and had higher quality scores than 25:75 and 0:100 ratios. This 50:50 ratio was then compared to pure PageRank. It was found that the combination returned better results overall, but returned especially relevant results for specific queries. Conclusions/Discussion To create a viable search engine with a better page ranking algorithm than PageRank or HITS, the novel WordRank algorithm was successfully designed and implemented in Python. A 50:50 ratio of PageRank to WordRank returned the best search results, especially for specific queries. This combination has large potential use in a research search engine for users looking for specialized results.

Competition history

• CSEF 2013 Mathematics & Software · Entry S1423

Resources

• CSEF project summary ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Computational Sciences projects
• CSEF Computational Sciences projects
• Every CSEF 2013 project