Tugas5 Komputer&masyarakat

PageRanking WordNet Synsets: An Application to Opinion Minin

Penulis :andrea Esuli and Fabrizio Sebastiani

URL: https://www.researchgate.net/publication/220874736_PageRanking_WordNet_Synsets_An_Application_to_Opinion_Mining

Several works have recently tackled the automateddetermination of term polarity.

Hatzivassiloglou andMcKeown (1997) determine the polarity of adjec-tives by mining pairs of conjoined adjectives fromtext, and observing that conjunctions such as andtend to conjoin adjectives of the same polarity whileconjunctions such as but tend to conjoin adjectivesof opposite polarity.

Turney and Littman (2003) de-termine the polarity of generic terms by computingthe pointwise mutual information (PMI) between thetarget term and each of a set of “seed” terms ofknown positivity or negativity, where the marginaland joint probabilities needed for PMI computationare equated to the fractions of documents from agiven corpus that contain the terms, individually orjointly. Kamps et al.

(2004) determine the polarityof adjectives by checking whether the target adjec-tive is closer to the term good or to the term badin the graph induced on WordNet by the synonymyrelation.

 Kim and Hovy (2004) determine the po-larity of generic terms by means of two alternativelearning-free methods that use two sets of seed termsof known positivity and negativity, and are basedon the frequency with which synonyms of the targetterm also appear in the respective seed sets. Amongthese works, (Turney and Littman, 2003) has provenby far the most effective, but it is also by far the mostcomputationally intensive

 

The Mathematics Behind Google’s PageRank

Penulis: Ilse Ipsen

URL : https://ipsen.math.ncsu.edu/ps/slides_man.pdf

Google Matrix

Convex combination G = α S + (1 − α ) 1 1 v T | {z } rank 1 • Stochastic matrix S • Damping factor 0 ≤ α < 1 e.g. α = .85 • Column vector of all ones 1 1 • Personalization vector v ≥ 0 k v k 1 = 1 Models teleportation

Properties of Matrix S

• Row i of S: Links from page i to other pages • Column i of S: Links into page i • S is a stochastic matrix: All elements in [0 , 1] Elements in each row sum to 1 • Dominant left eigenvector: ω T S = ω T ω ≥ 0 k ω k 1 = 1 • ω i is probability of visiting page i • But: ω not unique

Google PageRank

PENULIS:???

URL : https://www.mathworks.com/content/dam/mathworks/mathworks-dot-com/moler/exm/chapters/pagerank.pdf

Further Reading Further reading on matrix computation includes books by Demmel [?], Golub and Van Loan [?], Stewart [?, ?], and Trefethen and Bau [?]. The definitive references on Fortran matrix computation software are the LAPACK Users’ Guide and Web site [?]. The Matlab sparse matrix data structure and operations are described in [?]. Information available on Web sites about PageRank includes a brief explanation at Google [?], a technical report by Page, Brin, and colleagues [?], and a comprehensive survey by Langville and Meyer [?]. Recap %% Page Rank Chapter Recap % This is an executable program that illustrates the statements % introduced in the Page Rank Chapter of "Experiments in MATLAB". % You can access it with % % pagerank_recap % edit pagerank_recap % publish pagerank_recap % % Related EXM programs % % surfer % pagerank %% Sparse matrices 10 Chapter 7. Google PageRank n = 6 i = [2 6 3 4 4 5 6 1 1] j = [1 1 2 2 3 3 3 4 6] G = sparse(i,j,1,n,n) spy(G) %% Page Rank p = 0.85; delta = (1-p)/n; c = sum(G,1); k = find(c~=0); D = sparse(k,k,1./c(k),n,n); e = ones(n,1);j I = speye(n,n); x = (I - p*G*D)\e; x = x/sum(x) %% Conventional power method z = ((1-p)*(c~=0) + (c==0))/n; A = p*G*D + e*z; x = e/n; oldx = zeros(n,1); while norm(x - oldx) > .01 oldx = x; x = A*x; end x = x/sum(x) %% Sparse power method G = p*G*D; x = e/n; oldx = zeros(n,1); while norm(x - oldx) > .01 oldx = x; x = G*x + e*(z*x); end x = x/sum(x) %% Inverse iteration x = (I - A)\e; x = x/sum(x) %% Bar graph bar(x) title(’Page Rank’)

 

A Survey and Comparative Study of Different PageRank Algorithms

Penulis : Tahseen A. Jilani University of Karachi Karachi, Pakistan

 Ubaida Fatima NED University Karachi, Pakistan

Mirza Mahmood Baig NED University Karachi, Pakistan

Saba Mahmood IoBM Karachi, Pakistan

url : https://www.researchgate.net/publication/281170066_A_Survey_and_Comparative_Study_of_Different_PageRank_Algorithms

 

In  contemporary  computer  generation,  internet  has become

indispensable in our lives and cognizance is only a click away.

We just open our desired  search engines, like Google, Yahoo,

Bing, and the search engine will show the webpages appropriate

for our search. 

Google’s  humongous  triumph  as  a  search  engine  can  be

ascribing  to  numerous  elements,  including  its  naiveness,

acceleration and ease of employ. Conversely, the most eminent

grounds  for  their  triumph  is  due  to  their  search  algorithm;

contrast  to  other  search  engines;  Google  provides  the  most

significant ensue first.

Google as a search engine inevitably to be able to execute two

errands. First it requires to acquire and retain all of the webpages

it is able to; this attain by crawling the web ad indexing the data

that it encounters. Second, it requires being able to figure out the

order  of  pages  resumed  by  any  search  survey.  This  is

accomplished  through  Google’s  PageRank  algorithm  as

proposed  by  Wills[1],  which  assesses  each  webpage  and  its

status relative to  other webpages. “Significance”,  as defined by

PageRank algorithm, is  subject on the number of other pages,

linking to a webpage. This is a prejudiced factor which includes

the importance of  other pages, meaning that a high importance

webpage will impart more importance to a linked page  than an

irrelevant page

 

The anatomy of a large-scale hypertextual Web search engine

Penulis : omputer Networks and ISDN Systems 30 ( 1998) 107- 117

URL : http://www.cse.fau.edu/~xqzhu/courses/cap6777/google.search.engine.pdf

The Web creates new challenges for information retrieval. The amount of information on the Web is growing rapidly, as well as the number of new users inexperienced in the art of Web research. People are ’ Corresponding author. ’ There are two versions of this paper - a longer full version and a shorter printed version. The full version is available on the Web and the conference CD-ROM. ’ E-mail: (sergey, page] @cs.stanford.edu likely to surf the Web using its link graph, often starting with high quality human maintained indices such as Yahoo! 3 or with search engines. Human maintained lists cover popular topics effectively but are subjective, expensive to build and maintain, slow to improve, and cannot cover all esoteric topics. Automated search engines that rely on keyword matching usually return too many low quality matches. To make matters worse, some advertisers attempt to gain people’s attention by taking measures meant to mislead ’ http://www.yahoo.com 0169-7552/9X/$19.00 0 1998 Published by Elsevier Science B.V. All rights reservedautomated search engines. We have built a large-scale search engine which addresses many of the problems of existing systems. It makes especially heavy use of the additional structure present in hypertext to provide much higher quality search results. We chose our system name, Google, because it is a common spelling of googol, or 10100 and fits well with our goal of building very large-scale search engines.