Synthesis Essay

            It is a fact that people use search engines for so many things and their use of search engines is increasing with the passage of time. The beauty of search engines is that they have really made a great influence on people that how they look for information on the internet. It has been observed that entry points for web pages have been provided by these search engines. That’s why when people search for anything on these search engines, few web pages are more visible and comes early in search results than many other web pages. It happens, because these web pages have been ranked well on these search engines through keywords. That’s why it is the top priority of companies to get top ranking on web so that they can generate more traffic to their web pages.  For this purpose, one method has become the most important one in this regard, and that method is called Search Engine Optimization (SEO). The ranking algorithms are used by SEO method to rank website on these search engines such as Google. The Google ranking algorithm can be a good one to understand the fact that which web pages will be shown to visitors with regards to their enterer search key words. The web masters can optimize web pages through such ranking methods to get better results ¹

        It is important to understand the fact that how PageRank algorithm originally work. Its work method is that to improve search query ranking, it uses single vector to compute results. In this process, the web’s link structure is also used so that web pages with relevant importance can be captured. However, a research study proposed that if more accurate results have to be yielded then set of PageRank vectors should be used for computing results. When search queries for ordinary keywords are made, then topic sensitive PageRank scores are computed for the pages so that query can be satisfied. The study also proposed that when context is used to make any searches, for instance, when words in web pages are highlighted through search query, then scores for topic sensitive PageRank can be computed by generating content specific scores at the time of query. Such method depicted that as compared to using single vector for generic PageRank, the use of this method come up with better results in terms of finding more accurate rankings ²

        The other important thing associated with page ranking is bias in search algorithms. So, it is important to investigate that what kind of bias is there in approach of page ranking. A study was conducted to investigate that how Google and various other search engines use random walk in algorithm of PageRank. First of all, it was explained that Markov Chain is the one, which is used for the random walk work. On the basis of this fact, study further explains that how graph cycles are formed by matrix on specific locations. The great thing about these particular cycles is that queues of PageRank can be controlled with the help of them. Moreover, these cycles have capability of ranking value from one online web page to the other web pages, which is next on the list. After analyzing things through this method, the results indicated that for PageRank, the basis is provided by adjacency matrix, which may have contributed in biased spaces, which should be considered while ranking web pages. The study also throw light on the aspect that still PageRank algorithms face various issues and obstacles to work efficiently ³

            Keeping this bias approach in search algorithms in mind, it is important to further investigate more research literature so that more details can be found on this. One of the research studies looked at Meta bias in search algorithm as a necessity. It was described that in search space probability, the role of bias is crucial for learning. This Meta bias in algorithm of search is introduced by the search algorithm designer, who works to design these algorithms. One of the examples of such design is search operator. The bias does not show any changes, so whenever the execution of stochastic search algorithm will be made, the answer will be different. It means that bias shows static value in this process. In many problems, the purpose of search algorithms is to be reused so that these problems can be known better. In terms of probability distribution, a problem class and search algorithm could be seen associated with it. The research study analyzed various aspects of Meta bias and search algorithm, and it concluded that if many instance of different problems have to use search algorithm, then there is great need to use Meta bias so that process can be complete successfully ⁴

        It has been observed so for that bias in the search algorithm has been considered important and it can be useful. But it is not the case always, as opposing view was discussed in another research article. The topic of this research article was revolving around the fact that search engine should stop being so evil, and search engines like Google should look for unbiased search. It was explained in early part of the research that how Google has made its contribution since its start from 1990s. The search algorithms of Google have allowed people to hear ideas of others, obtain any type of information, which they want to see. Google made this information and data finding method very easy and convenient for internet users. But this credit of Google is now fading away as they are getting far away from openness of internet, which they used to champion. Now, their own content is being promoted by their search engine, which is squeezing the space for its competitors. The Google has got so much power in this field and need of the hour is to leverage this absolute power as much as possible so that competition can stop experiencing potential damage. So, it is responsibility of search engines like Google to stop doing so, and eliminate bias from their search algorithm approach. An unbiased Google is important for future course of this field as well as competitors ⁵

        Keeping this bias approach from Google, it is necessary to look at research paper, which has reviewed its bias approach in a real world context. It was reviewed in a research that Google comes with Bias for Partisan audience. There is a consensus on the fact that influence of online platforms is systematic on democratic process, and this influence is gradually increasing with the passage of time. But this research has been attributed only to social media platforms, and other than social media, the research in this regard is very limited. The research paper analyzed the audit of partisan audience bias through mixed method algorithm to see that how Google has been bias in its approach. After the inauguration of Donald Trump, the researchers took 187 contributors, who were given a survey to complete, and they were asked to install extension in the browser, which helps to obtain data regarding Search Engine Results Pages. In this process, the Twitter panel was used in domain level score to get real numbers from registered voters. The results from the study showed that Google’s ranking showed a clear shift towards the SERPs lean, which was on the unweighted average of right side. It was also found out that search results of Google as well as different components of Google are evolving with the passage of time, and more audits are needed to look at their bias approach in an online information ecosystem. The future research can be conducted to get more deep analysis of the facts, revealed in this current research paper ^{6, 2}

        On the other hand, it is important to keep in mind that web search engines are important tools for people to gather information as well as access information, which they require for certain purposes. But it is vital to see that how these web search engines put their emphasis on content. The fact of the matter is that web search engines are used by millions of users around the world. That’s why web search engines have applied ranking algorithms to make an influence on the pages to be visited by users. It is true that in certain period of time, specific points or sites are favored by these search engines. How content is emphasized on these web search engines, the study used a method called PAWS, which is helpful in knowing the differences, which are there between these search engines. The PAWS method in this research was applied to two search engines, Bing and Google, as they are one of the major in terms of user rankings. The results of the study showed that these search engines did not emphasize on results which can be attributed to products of the company. However, it was found out that certain news sites were emphasized by these search engines, and those pages were also favored amongst others, which have used ads of their companies. It means that sites containing ads of competitors were avoided by these search engines. It means that there was some bias as well as content emphasis shown by these web search engines. So, when web pages are containing ads of these companies, then their search engines do rank those pages higher than the other pages ^{7, 3}

        It has been mentioned earlier that people do use internet to access and obtain information, and many search engines are there. But most important thing for these users is to look for those search engines, which are relatively easy to use and users have to put little efforts for their information seeking. Moreover, they like search engines, which come up with desired, accurate as well as relevant results to their actual search, and it also should not take too much time. So, a study proposed that a search algorithm should be used on the basis of correlation, where judgments of explicit users are accepted in its relevancy to the documents of web pages. To makes sure that biased judgments are removed in this process, the feedback of implicit users were included in the search algorithm in the form of click hits, Dwell time as well as duration of session. In this research, the focus was given to three major search engines and number of TREC queries used was 25 to get actual comparison as well as accurate evaluation. The study came with an important finding that feedback of implicit users used along with feedback of explicit users is very important to evaluate anything about tools of web searches. The correlation approach used in this research was very beneficial to get an idea about different search tools through their evaluation and it also provided a roadmap for future research studies as well 

References of Work Cited of Synthesis Essay

Alam, Mohammed A. and Doug Downey. 2014. 22 April 2019. <http://users.cs.northwestern.edu/~ddowney/publications/alam_content_emphasis.pdf>.

Alkhalifa, Eshaa Mohamed Hamed. "Investigating Bias in The Page Ranking Approach." 2015 International Conference on Information and Communication Technology Research. IEEE, 2015. 294-297.

Goutam, Rajesh Kumar. "Correlation Based Evaluation for Search Tools." 2018 5th International Conference on Signal Processing and Integrated Networks (SPIN). IEEE, 2018.

Haveliwala, Taher H. "Topic-Sensitive PageRank: A Context-Sensitive Ranking Algorithm for Web Search." IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 15.4 (2003): 784-796.

Hazan, Joshua G. "Stop Being Evil: A Proposal for Unbiased Google Search." Michigan Law Review 111.5 (2013): 789-820.

Robertson, Ronald E., et al. "Auditing Partisan Audience Bias within Google Search." Proceedings of the ACM on Human-Computer Interaction - CSCW 2.CSCW (2018).

Su, Ao-Jan, et al. "How to Improve Your Google Ranking: Myths and Reality." 2010 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology. IEEE, n.d. 50-57.

Woodward, John R. "The Necessity of Meta Bias in Search Algorithms." 2010 International Conference on Computational Intelligence and Software Engineering. IEEE, 2010.