You are on page 1of 2

Project Abstract

The last part of the registration is your Project Abstract. We recommend that yo
u type your Project Abstract off-line and "Copy and Paste" the sections into the
application form.
The abstract must be one typewritten page or less using Times 12 point font. Be
aware that your abstract is saved in our database and displayed to judges withou
t formatting. Please use a text editor, such as notepad, to edit your abstract b
efore you "Copy and Paste" it into the form below. Follow the format provided be
low. Please keep responses short. All parts must be completed in order to submit
form.
Project Abstract Format
Problem Statement: What is the question to be answered through your experime
nt?
Hypothesis: What is your educated/testable answer to your question.?
Materials: List major equipment, all chemicals and materials used.
Procedure:
A. What major steps did you follow to investigate the problem?
B. What factor (variable) did you alter in order to test your hypothesis?
C. Describe your sample size and number of trials.
D. What measurements did you take?
Results: Describe briefly the measurements you obtained from your experiment
s as well as the results of any mathematical calculations.
Conclusion: Briefly describe the answer to your initial problem based on the
results you obtained in your experiments.
Problem Statement (*)
Web search engines such as Google often return irrelevant results in response to
a users query.
Hypothesis (*)
This problem may be due to the fact that the currently used ranking algorithms m
easure the popularity of web pages. A search engine that has a better page-ranki
ng algorithm than Google can be developed. This novel page-ranking algorithm wil
l be able to differentiate web pages based on their content and will be implemen
ted in Python.
Materials (*)
Computer (Intel i5 CPU@ 2.67 GHz, 4.00 GB RAM), Python 2.7.3, Microsoft Excel
Procedure (*)
(1) Create a web crawler that will extract text and links and index them.
(2) Implement the HITS and PageRank (used by Google) algorithms in Python, which
will be used to rank the pages in the index.
(3) Design and implement WordRank in Python, a novel ranking algorithm that rank
s pages based upon user inputted words and the content of the page.
(4) Make a look-up function that will take in a searched word and return a list
of relevant pages in order according to the ranks.
(5) Compare the ranking algorithms (e.g. PageRank vs HITS, PageRank vs WordRank)
using the number of relevant links and the quality scores. Also, test various c
ombinations of ranking algorithms and compare the results.
Results (*)

After the PageRank and HITS algorithms were implemented in the search engine, it
was discovered that PageRank returned more relevant results than HITS. PageRank
was therefore used with WordRank when computing the ranks. When the two were us
ed together, the scores were taken from each and placed in PageRank to WordRank
ratios of 50:50, 25:75, and 0:100. It was observed that the number of relevant l
inks and the quality score of the returned pages from the 50:50 ratio were the h
ighest. 50:50 PageRank to WordRank (50:50 PR:WR) was then compared to pure PageR
ank. It was determined that 50:50 PR:WR returned more relevant results than pure
PageRank overall. In addition, when used on specific queries that returned fewe
r links, 50:50 PR:WR was able to find far more relevant results and had a much h
igher quality score than PageRank.
Conclusions (*)
To create a viable search engine with a better page ranking algorithm than PageR
ank and HITS, the novel ranking algorithm WordRank was successfully designed and
implemented in Python. WordRank ranks pages based on their content and user inp
utted keywords rather than on their popularity. It was observed that a 50:50 rat
io of PageRank to WordRank yielded the most relevant results. When compared to p
ure PageRank, the 50:50 ratio returned superior results. This search engine can
potentially be used as for research purposes, as WordRank returns highly relevan
t results to specific queries.

You might also like